You are on page 1of 16

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/260941272

Experimental and numerical analysis of pinned-joints composite laminates:


Effects of stacking sequences

Article  in  Journal of Composite Materials · December 2013


DOI: 10.1177/0021998312464891

CITATIONS READS

29 1,423

5 authors, including:

U.A. Khashaba T.A. Sebaey


Zagazig University Prince Sultan University
58 PUBLICATIONS   1,927 CITATIONS    58 PUBLICATIONS   672 CITATIONS   

SEE PROFILE SEE PROFILE

A. I. Selmy
Zagazig University
28 PUBLICATIONS   622 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

glass-polyamide hybrid composites View project

Severe Plastic Deformation View project

All content following this page was uploaded by T.A. Sebaey on 09 September 2015.

The user has requested enhancement of the downloaded file.


Journal of Composite Materials http://jcm.sagepub.com/

Experimental and numerical analysis of pinned-joints composite laminates: Effects of stacking


sequences
UA Khashaba, TA Sebaey, FF Mahmoud, AI Selmy and RM Hamouda
Journal of Composite Materials published online 9 November 2012
DOI: 10.1177/0021998312464891

The online version of this article can be found at:


http://jcm.sagepub.com/content/early/2012/11/07/0021998312464891

Published by:

http://www.sagepublications.com

On behalf of:

American Society for Composites

Additional services and information for Journal of Composite Materials can be found at:

Email Alerts: http://jcm.sagepub.com/cgi/alerts

Subscriptions: http://jcm.sagepub.com/subscriptions

Reprints: http://www.sagepub.com/journalsReprints.nav

Permissions: http://www.sagepub.com/journalsPermissions.nav

>> OnlineFirst Version of Record - Nov 9, 2012


What is This?

Downloaded from jcm.sagepub.com at Zagazig University on November 9, 2012


JOURNAL OF
COMPOSITE
Article M AT E R I A L S
Journal of Composite Materials
0(0) 1–14
! The Author(s) 2012
Experimental and numerical analysis Reprints and permissions:
sagepub.co.uk/journalsPermissions.nav
of pinned-joints composite laminates: DOI: 10.1177/0021998312464891
jcm.sagepub.com
Effects of stacking sequences

UA Khashaba1, TA Sebaey2, FF Mahmoud2,3, AI Selmy2 and


RM Hamouda4

Abstract
The present study investigates the effect of stacking sequence on the failure loads (strength) and modes of pinned-joints
glass-fiber reinforced epoxy composite laminates. Specimens with [0/90]2S, [15/75]2S, [30/60]2S and [45/45]2S
stacking sequences were investigated both experimentally and numerically. A series of ASTM tests were performed
on unidirectional [8]0 glass-fiber reinforced epoxy composite laminate to determine the properties of the single lamina
that was needed for the finite element analysis. A 3D progressive damage model was built with the aid of ABAQUS
software, failure criteria and property degradation rules to simulate the problem. The results showed that the [0/90]2S
laminate has the highest ultimate strength. The minimum bearing and ultimate strength was observed for [30/60]2S
laminate. Loading the specimens up to the ultimate value lead to shear-out failure mode for [0/90]2S, [15/75]2S and
[30/60]2S stacking sequences, while specimens with [45/45]2S stacking sequence are characterized by bearing failure
mode. The experimental and numerical results agree well with a maximum Euclidean error norm of 8.57%.

Keywords
Pinned-joints, finite element, stacking sequence, bearing strength, ultimate strength, bearing stiffness

The effect of stacking sequence on the behavior of


Introduction
composite joints was addressed by several authors.4-8
Fiber-reinforced composite materials and structures are Ondurucu et al.4 compared the induced damage of
being used in space, air, land and marine structures. [0/902/0]s and [0/90]2s glass/epoxy. Turvey6 studied the
Bolts provide the primary means of connecting com- effect of the angle between pultrusion and tension direc-
posite parts in the construction of aircrafts, aerospace tion of bolted joint test for Glass Reinforced Plastic
and automotive vehicles. The joint design has a special (GRP) laminates. For off-axis angle 30 , the bearing
significance in fiber-reinforced composite structures for failure did not arise at any geometrical configuration.
two reasons:1 (1) joints are often the weakest points in a Icten et al.7 presented an experimental and numerical
composite structure and (2) the composite materials do
1
not possess the forgiven characteristics of ductile Mechanical Engineering Department, Faculty of Engineering, King
metals, namely, their capability to redistribute local Abdulaziz University, Jeddah, Saudi Arabia
2
Mechanical Design and Production Engineering Department, Faculty of
high stresses by yielding
Engineering, Zagazig University, Zagazig, Egypt
Experimentally, the mechanically fasted joints fail 3
Visiting Professor, University of Texas, Dallas, TX, USA
under three basic mechanisms: net tension, shear out 4
Department of Design and Production Engineering, Faculty of
and bearing failure. Typical damage caused by each Engineering, Ain-Shams University, Massaken Nasr-City, Cairo, Egypt
mechanism is shown in Figure 1.2 Moreover, cleavage
failure takes place in laminates having 0 fibers along Corresponding author:
UA Khashaba, Mechanical Engineering Department, Faculty of
the loading direction.3 The net tension, shear out and Engineering, King Abdulaziz University, P.O. Box 80204, Jeddah 21589,
cleavage failure modes are not desirable due to the cata- Saudi Arabia.
strophic nature of final failure. Email: khashabu@hotmail.com; khashabu@zu.edu.eg

Downloaded from jcm.sagepub.com at Zagazig University on November 9, 2012


2 Journal of Composite Materials 0(0)

Figure 1. Failure modes of fibrous composite mechanical joints: (a) tension failure; (b) shear failure and (c) bearing failure.2

Table 1. The constituent materials of glass-fiber reinforced epoxy (GFRE) laminate

Mechanical properties

Tensile Tensile Shear


Density, Modulus, strength, Poisson’s modulus,
Constituent materials (g/cm3) (GPa) (GPa) ratio (GPa)

Epoxy resin: Araldite PY 1092 (100 part by weight); 1.1 3.2 0.069 0.36 1.18
Hardener HY 1092 (50 part by weight)
E-roving glass fiber linear density 2.54 72.40 3.45 0.22 30.13
¼ 2.10 g/m, d ¼ 17  2 mm

study to compare the response of [(0/90)3]S and [(45)3]S effect of specimen stacking sequences on the mean first-
stacking sequences of glass-fiber reinforced epoxy peak bearing strength, mean ultimate failure stress,
(GFRE). The [(45)3]S specimens failed as bearing fail- failure displacement and bearing stiffness. For this
ure mode while the [(0/90)3]S failed as shear out failure objective, a cross-ply, [0/90]2s, GFRE laminates were
mode with slight evidence of bearing damage around the manufactured using the hand lay-up technique. The
hole. In terms of bearing strength, the bearing strength laminates were cut at different off-axis angles (0, 15,
of cross-ply laminates was greater than it for angle ply 30, 45 ) to give, respectively [0/90]2s, [15/75]2s,
ones. Similar results were presented by Park8 for [903/ [30/60]2s and [45/45]2s stacking sequences.
þ453/453/03]S, [903/03/þ453/453]S, [906/06]S, [03/þ453/ The second objective was to determine the mechan-
453/903]S and [06/906]S stacking sequences. ical properties of the single layer, to be used in the
The present study aims to investigate the effect of progressive failure analysis of GFRE pinned-joints
off-axis angle (stacking sequence) both experimentally laminated with different stacking sequences. For this
and numerically on the bearing strength and failure purpose, [0]8 unidirectional GFRE composite laminates
mode for [0/90]2S GFRE laminates. The off-axis were manufactured using the hand lay-up technique.
angles are 0, 15, 30, 45 to produce specimens with The constituent materials of composite laminates
[0/90]2S, [15/75]2S, [30/60]2S and [45/45]2S stacking and their mechanical properties were illustrated in
sequence. In order to perform a numerical analysis, the Table 1. Details about the manufacturing technique
mechanical characteristics of the single lamina are are presented elsewhere, Khashaba et al.11 The lamin-
obtained experimentally. A 3D analysis is performed ate was precured under uniform pressure for 24 h at
using the ABAQUS software. A progressive damage room temperature, ISO 1268, and post cured at room
model (PDM), considering the Hashin failure criteria9 temperature for further 21 days. Fiber volume fraction
and the property degradation rules presented by Riccio (Vf) was determined experimentally using the ignition
and Scaramuzzino,10 is used. technique according to ASTM D3171. The average
value of Vf was 40  0.5%. The thickness of the cross-
ply laminate was 3.2  0.1 mm.
Experimental work
Fabrication of composite laminates Mechanical tests
Two main objectives were achieved from the experi- All the mechanical tests were performed on com-
mental work. The first objective was to investigate the puter controlled Universal Testing Machine

Downloaded from jcm.sagepub.com at Zagazig University on November 9, 2012


Khashaba et al. 3

(a) 3.2 (b)


36 8
To machine grip

φ10 Bolt

135 6.5

φ 6 Steel pin
6
3.2 Composite Specimen
18

Applied load from


testing machine

Figure 2. (a) Dimensions of pinned-joint specimens and (b) pinned-joint fixture.

(Testometric 200 kN). The head-speed of the testing manufactured from stainless steel according to the
machine was 2 mm/min for all the mechanical tests. geometry illustrated in Figure 2(b).
Forty-five specimens were tested in this work. Twenty In the aircraft industry, 60% of all rejected parts
of them were tested in bearing tests (five specimens for during final assembly of an aircraft structure are due
each stacking sequence). Ten unidirectional specimens to delamination-associated drilling.13 Drilling at high
were tested in tension: five of them were tested in the feed results in high thrust force, which is the direct
longitudinal direction and five specimens were tested in cause for the delamination onset in drilling GFRE com-
the transverse direction. Similarly, ten unidirectional posites and subsequently the lower bearing
specimens were tested in compression. The last five uni- strength.12-15 Therefore, special attention is taken to
directional specimens were used for the Iosipescu shear machining delamination-free holes with 6 mm diameter
test to determine the in-plane shear properties normal in the pin bearing specimen.
to the fiber direction. The strength values were deter- The pins were neat-fit the hole (zero clearance). Five
mined based on the average values. The apparent load- specimens were tested for each condition and the mean
displacement diagrams were printed through the PC of values were used to develop the different relationships.
the testing machine. The actual strains were measured The crosshead speed of the loading member was
for various testing using strain gages that connected to 2 mm/min. The bearing stress was calculated from
Digital Strain Meter (Tc-21 K model 232). equation (1).

Bearing tests. A series of pin-bearing ASTM D 5961 P


b ¼ ð1Þ
tests were conducted on GFRE specimens with differ- Dh  t
ent stacking sequences using universal testing machine
(Testometric 200 kN) at room temperature. The where Dh ¼ hole diameter, t ¼ specimen thickness and
machine was fully computer controlled and allows for P ¼ failure load. In accordance with the standard, the
acquisition of load and displacement. actual hole diameters (Dh) and the thickness (t) in the
Standard test specimens were used to obtain bearing vicinity of the hole for each individual joint were used
failure mode rather than net tension or shear-out in the calculations, rather than nominal values.16 The
modes,11,12 which had lower loads associated with cata- failure load (P) was defined by two different levels. The
strophic fracture. The dimensions of the test specimen first load level is the load at the first-peak of the load-
are illustrated in Figure 2(a). The test fixture was displacement curve, which define the first-peak bearing

Downloaded from jcm.sagepub.com at Zagazig University on November 9, 2012


4 Journal of Composite Materials 0(0)

strength.17-19 The second one is the ultimate load on (strength and Young’s modulus) and the Poisson’s
load-displacement curve, which define the ultimate fail- ratio are presented in Table 2.
ure strength.8,12-14,20,21
Compression tests. The compressive properties of unidir-
ectional composite laminates are determined experi-
Tension tests. The tensile properties of unidirectional mentally according to the ASTM D695. The test
composite laminates are determined experimentally fixture and the dimensions of the test specimens are
according to the ASTM D3039. The dimensions of illustrated in Figure 6. Four rectangular aluminum
each test specimen is illustrated in Figure 3. Two end tabs are bonded at the gripping portions. The func-
strain gauges are mounted back-to-back, both at the tions of these tabs are to prevent the buckling of the
center of the unidirectional specimen and normal to specimens under compressive load and ensure that the
each other to determine the actual values of the tensile failure will be at/near the specimen midpoint. The load-
modulus and Poisson’s ratio. Four rectangular alumi- displacement diagrams are illustrated in Figure 7 and
num end tabs are bonded at the gripping portions, in the average values of the longitudinal and transverse
order to reduce the stress concentration owing to the compressive strength are indicated in Table 2.
lateral compressive stresses of grips serration and pre-
vent the slipping of the test specimen from the grip. In-plane shear tests. In-plane shear tests are implemented
End-tabs also smoothly transfer the lateral compressive on unidirectional GFRE composite laminate using
load owing to the grips of the testing machine to the Iosipescu shear test according to ASTM D5379. The
specimen and prevent the crushing of the test specimens fixture, Figure 8(a), was manufactured and modified
between the grips.22 The load-displacement diagrams by machining a guide way on the spacer block of the
and the stress–strain relationships are illustrated in left half to prevent the twisting of the movable (right)
Figures 4 and 5, respectively. The average values of half.23,24 The left half of the fixture is fixed rigidly to the
the longitudinal and transverse tensile properties base plate along with the spacer block. The front face of
the specimen is visible during the testing. Accordingly,
the progress of failure can be monitored visually. The
210 principle of the test is to apply a set of prescribed dis-
90 placements on the V-notch specimen, so that the central
3.5 Tab 3 region of the sample is under a state of predominant
3.2 shear. These displacements are achieved through rela-
tive movement of the movable grip with respect to the
fixed grip.
27
The test specimens were cut to strips with
Strain gauges 75  19 mm2 and a 90 double V-notches were
machined through the thickness of the specimen,
Figure 3. Dimension of tension specimen. using form-milling cutter, to the dimensions final

(a) 40 (b) 1.6

35

30 1.2

25
Load (kN)
Load (kN)

20 0.8

15

10 0.4

0 0

0 3 6 9 12 0 0.2 0.4 0.6 0.8 1 1.2


Displacement (mm) Displacement (mm)

Figure 4. Load-displacement curves of UD-glass-fiber reinforced epoxy (GFRE) composites in tensile: (a) loaded in the fiber
direction and (b) loaded in the transverse direction.

Downloaded from jcm.sagepub.com at Zagazig University on November 9, 2012


Khashaba et al. 5

dimensions, Figure 8(b). To measure the shear strain modulus is determined from the initial slop of the
(g12) and shear modulus (G12), two strain gages were shear stress–strain curve, Figure 10(b) as:
bonded at þ45 and 45 at the center of the test spe-
cimen, Figure 8(b). 12
G12 ¼ ð4Þ
The load-displacement diagram is illustrated in 12
Figure 9. The value of shear strength (S) was calculated
from the following equation: where  12 is the in-plane shear stress.

Pmax
S¼ ð2Þ Experimental results and discussions
A
Load-displacement behavior of pinned-joints
where A is the cross-sectional area between the roots of Figure 11 illustrates the load-displacement curves of
two V-notches and Pmax is the maximum load. The pin-loaded specimens with [0/90]2s, [15/75]2s,
average value of the measured shear strength is pre- [30/60]2s and [45/45]2s stacking sequences. Figure
sented in Table 2. 11(a)–(c) indicates that the load-displacement curves
The relationship between shear-stress and strains have a linear relationships up to a point at which the
of þ45 and 45 strain gauges is illustrated in deviation from linearity was observed. The nonlinear
Figure 10(a). The value of shear strain ( 12) was calcu- relationship is due to the progressive debonding of
lated as:24 fiber/matrix interface owing to the bearing load.
Progressive loading leads to ‘brooming’ failure at the
12 ¼ "45  "þ45 ð3Þ edge of the hole at which the first beak on the load-
displacement curve is observed. The load subsequently
where "45 is tensile while "þ45 is compression, so the increases in a nonlinear fashion accompanied with sev-
shear strain is positive. Thus, the in-plane shear eral peaks up to the ultimate load. The number of peaks

Stress (MPa)
(a) 700

600 (b) 8
[90]8
500
6
400
Stress (MPa)

300 4

200
[0]8 2
100 Transverse
Longitudinal
0 0
0 0.05 0.1 0.15
–1 –0.5 0 0.5 1 1.5 2
Strain (%)
Strain (%)

Figure 5. Stress–strain curves of (a) [0]8 laminate and (b) [90]8 laminate.

Table 2. Mechanical properties of glass-fiber reinforced epoxy (GFRE) unidirectional laminates

E1 E2 G12 N12 A Xt Xc Yt Yc S
GPa GPa GPa (MPa)3 MPa MPa MPa MPa MPa

32.1 5.74 1.24 0.33 1.9  106 722 230 14 34 54.6

Downloaded from jcm.sagepub.com at Zagazig University on November 9, 2012


6 Journal of Composite Materials 0(0)

3
(a) (b)

80 10

3.2
17

Figure 6. (a) Dimensions of compression test specimen and (b) experimental setup.

14 load, without the presence of peaks, Figure 11(d)


[0]8 curve (ii). The first-peak load in this case is determined
12 [90]8 at the point where the slope of the nonlinear curve is
sharply decreased. Referring to curves (i) and (ii) in
10 Figure 11(d), the two points are approximately the
same. The final failure of the test specimens with differ-
Load (kN)

8 ent stacking sequences is the gradual drop from the


ultimate loads without catastrophic fracture due to
6 the energy absorption through different failure mech-
anisms of the composite layers, Figure 11(a)–(d).
4

Failure displacements
2
Figure 12 shows the effect of stacking sequences on the
0 first-peak and ultimate failure displacements of pinned-
0 0.5 1 1.5 2 2.5 joints GFRE composite laminates. Specimens with
Displacement (mm) [0/90]2s stacking sequence have the highest displacement
to the ultimate failure load, compared to the other stack-
Figure 7. Load-displacement curves of glass-fiber reinforced
ing sequences. This result is due to the higher energy
epoxy (GFRE) unidirectional composites in compression.
dissipated in breaking and pulling out the transverse
fibers in 90 layers compared to the other laminates
on the load-displacement curves denotes the complexity with different stacking sequences. Therefore, this stack-
of failure mechanisms for specimen with different stack- ing sequence increases the energy absorption and delay-
ing sequence. ing the catastrophic failure compared to the other
The presence of 45 layers in specimens with stacking sequences. Increasing the absorbed energy in
[45/45]2s stacking sequence resists the in-plane shear FRP composite joints is very important issue for the
stresses consequently, delays the observation of the designer with FRP composite materials. On the other
peaks, Figure 11(d). The load-displacement curves of hand, specimens with [45/45]2s stacking sequences
[45/45]2s stacking sequence show a relatively linear have the maximum distance to failure based on the
behavior up to about 60% of the first-peak failure first-peak criterion. This result because the [45/45]2s
load, Figure 11(d) curve (i), followed by a slight non- stacking sequence has the maximum in-plane shear
linearity up to the first beak. This nonlinearity is due to modulus compared to the other stacking sequences.24
the failure of fiber/matrix interface that followed by
‘‘brooming’’ failure at the edge of the hole that
Bearing stiffness
observed as the first-peak on the load-displacement
curve. In many cases, the load-displacement curves The effect of stacking sequences on the stiffness of
show a smooth nonlinear behavior up to the ultimate pinned-joints GFRE composite laminates is illustrated

Downloaded from jcm.sagepub.com at Zagazig University on November 9, 2012


Khashaba et al. 7

(a) P from the loading member


of the testing machine

Pivot
Fixed grip
Moving grip

Specimen Strain gages

Thumb
Adjustable Screw
Specimen alignment pin Jaw
Guide way to
prevent twisting
the moving grip
Spacer
block

Four clamping
screws Base plate

90°
(b)

w =11.9 Strain gauges 19

75

Figure 8. (a) Iosipescu shear tests fixture and (b) dimension of shear test specimen.

2 in Figure 13. The figure indicates that specimens with


[0/90]2s lay-up have the highest bearing stiffness com-
pared to the other stacking sequences. This result is due
1.5 to the presence of 0 layers, which resists the displace-
ment in the load direction resulting in maximum appar-
P ent bearing stiffness. Figure 13 also shows that the
Load (kN)

values of the apparent bearing stiffness are decreased


1
in a linear fashion with the off-axis angle of the cross-
ply laminates. This behavior qualitatively agrees with
P the predicted values of Young’s modulus of cross-ply
0.5 laminates with different off-axis angle using classical
laminate theory, Figure 13.

0 Bearing strength and ultimate strength


0 0.5 1 1.5 2 2.5
Figure 14 illustrates the effect of the stacking sequences
Displacement (mm)
(off-axis angles) on the mean bearing and ultimate
Figure 9. Load–displacement diagram of Iosipescu shear test strengths in pin-loaded GFRE composites. Compared
specimen. to the other stacking sequences, [45/45]2s specimens

Downloaded from jcm.sagepub.com at Zagazig University on November 9, 2012


8 Journal of Composite Materials 0(0)

(a) 10 (b) 40
τxy
ε of + 45° strain gauge Gxy =
γxy

Shear stress, τ12 (MPa)


5 ε of –45° strain gauge 30

Strain, ε (%)
0 20

–5 10

–10 0
0 10 20 30 40 0 5 10 15 20
Shear stress, τ12 (MPa) Shear strain, γ12 (%)

Figure 10. Shear stress–strains (a) shear-stress vs. strains of þ45 and 45 strain gauges and (b) shear-stress vs. shear-strain of the
Iosipescu specimen.

(a) 7 Transverse (b) 7


cracks due to [0/90]2s Slope of tangent [15/-75] 2s
buckling of 0o line = apparent
6 layers 6 stiffness (kN/mm)
Brooming
5 failure 5
mode
1st peak
Interlaminar
Load (kN)

Load (kN)

4 shear 4

In-plane
3 shear 3

2 Bearing 2 Start of fiber/matrix


failure interface failure
1 1

0 0
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7
Displacement (mm) Displacement (mm)
(c) 7 (d) 7
[30/-60] 2s [45/-45] 2s
6 6
1st peak (ii)
1st peak (i)
5 5
1st peak
Load (kN)

Load (kN)

4 4

3 3

2 2

1 1

0 0
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7
Displacement (mm) Displacement (mm)

Figure 11. Load-displacement diagrams of pinned-joints glass-fiber reinforced epoxy (GFRE) composites with different stacking
sequences.

Downloaded from jcm.sagepub.com at Zagazig University on November 9, 2012


Khashaba et al. 9

6
Ultimate Ultimate
400
Bearing Bearing
Failure displacement (mm)

300
4

Strength (MPa)
200

2
100

0
0 [0/90]2s [15/-75]2s [30/-60]2s [45/-45]2s
[0/90]2s [15/-75]2s [30/-60]2s [45/-45]2s Stacking sequence
Stacking sequence
Figure 14. Ultimate and bearing strengths of pinned-joints
Figure 12. Ultimate and bearing failure displacements of glass-fiber reinforced epoxy (GFRE) specimens with different
pinned-joints glass-fiber reinforced epoxy (GFRE) specimens with stacking sequence.
different stacking sequences.
The minimum first-peak bearing were recorded for
specimens with [15/75]2s stacking sequence, Figure 14,
while the minimum ultimate strength were observed for
specimens with [30/60]2s stacking sequence. The latter
6 results are in agreement with the observation of
Pakdil.25 He reported that the minimum and maximum
30 strengths were observed, respectively, for [02/60]s and
[02/452]s compared to [02/302]s and [02/902]s stacking
sequences.
Young's modulus (GPa)

4
Stiffness (kN/mm)

20
Finite element simulations
A numerical model to simulate this problem should
2
take into account the evolution of interlaminar
10 damage (delamination) as well as that of the interlami-
nar one to capture the coupling between them. That
Bearing stiffness
was the aim of de Morais et al.26 and Shokrieh and
Predicted Young's modulus
Attar27 using the Visual Crack Closure technique or
0 0 that of Wisnom28 using cohesive zone models for
0 15 30 45 both delamination and intraply damage. These
Off-axisangle(Deg.) approaches are computationally expensive and, there-
fore, usually restricted to the analysis of short series or
Figure 13. Bearing stiffness of pinned-joints glass-fiber rein-
forced epoxy (GFRE) specimens with different off-axis angles to 2D problems.29 More simplified and computation-
(stacking sequence). ally nonexpensive approach is adopted in this study by
ignoring the delamination effect and assuming that the
coefficient of friction is to be constant in all directions.
The boundary
is assumed to consist of three dis-
have the maximum bearing capacity based on the first- joint measurable parts DC, DD and DF, which are the
peak criterion. This result because the [45/45]2s stack- portions of the boundary on which displacement and
ing sequence has the maximum in-plane shear strength traction are prescribed, respectively, Figure 15. The
compared to the other stacking sequences.24 Based on parts are meshed using the 8-node linear isoperimetric
the ultimate failure load, specimens with [0/90]2s stack- element. DC is the candidate contact zone containing
ing sequence have the maximum value compared to the the adjacent contact interfaces. Fixed constraints are
other stacking sequences. made on the bolt heads DD1. An important fixed

Downloaded from jcm.sagepub.com at Zagazig University on November 9, 2012


10 Journal of Composite Materials 0(0)

constraint is applied to the end of the plate DD2 to contact behavior. The contacting nodes are allowed to
simulate the contact between the specimen and the separate after contact. The detailed contact algorithm
machine jaws. The least fixed constraint allows the can be seen elsewhere.30
nodes to deform in the loading direction only. A uni- The PDM is an integration of stress analysis, failure
form distributed load is applied on the other end of the criteria and property degradation rules. It is imple-
specimen DF. A frictional contact boundary condition is mented by developing the ABAQUS finite element soft-
applied between the bolt and the hole and between the ware. After stress analysis, with a certain interaction
bolt heads and the outer layers, DC, Figure 15. between the ABAQUS and the subroutine USDFLD,
Before proceeding with the numerical simulations, the model performs a failure analysis using the non-
the mesh of each individual part was checked, in linear Hashin failure criteria. Once the failure index
terms of the shape and size, using the ABAQUS stand- exceeds the unity, at any integration point, the proper-
ard verification tool. No errors were detected, however, ties of this element was degraded by means of the prop-
some warnings were commented regarding to the elem- erty degradation rules. The model is explained by
ent aspect ratio. The solution is not expected to be means of the flowchart in Figure 16.
dependent on these elements because their in-plane pos- The failure analysis of composite materials is based
ition was bit far away from the bolt/hole interface. on three main assumptions of equal strengths in all
Due to the contact constraint and the nonlinearity of fibers, linear elastic behavior up to failure and equal
the material itself, the solution was expected to be longitudinal strains in fibers and matrix. While the
mesh-dependent, i.e. accurate solution requires highly assumption regarding linear elastic behavior up to fail-
refined mesh. On the other hand, highly refined mesh ure is not valid for many ductile matrix materials, the
means unreasonable computational time. To clarify this errors generated by this assumption are believed to be
dependency, a parametric study was conducted using small.31 The nonlinear Hashin failure criteria9 are
several mesh refinements. Figure 15 shows the optimum extension of the linear Hashin criteria assuming a non-
mesh at which the computational time is reasonable linear behavior. The criteria were selected because it
and more refinements showed insignificant effect on can take into consideration the nonlinear shear
the contact stress distribution and convergence criteria.
In order to simulate the bolt preload (tighten
torque), a shrink fit is considered between the bolt
head and the outer layers. In other words, the bolt is Start
made shorter than the laminate thickness by a small
value of interference. This initial interpenetration
yields a tensile stress on the bolt and a compression Materials and Geometry
stress on the surface between the washer and the speci-
men, which has the same effect of the torque.
Initial load Pi
Throughout this analysis, the frictional contact is
adopted using the coefficient of friction of 0.1. The
Augmented Lagrange method is used to simulate Linear analysis

L
F Check for No
Final failure
convergence

L Yes Stop
D2
Stress computation

Check for Yes Properties


failure degradation

No
L L Load incrementation
C D1
Pi = Pi –1+ P

Figure 15. Bolted joint specimen and bolt. Figure 16. Progressive damage model.

Downloaded from jcm.sagepub.com at Zagazig University on November 9, 2012


Khashaba et al. 11

response of the composite material. The modified cri- variables must be continuous within the elements. The
teria with respect to failure modes are as follows: second stated that the element size must not reduce to
Fiber tensile failure ( 11 > 0): zero.32 From these two requirements, we can return the
failure of the joint to the large deformation of the joint.
 2
11 On the other hand, the joint can be considered to fail
þT  1 ð5Þ when hole elongated by 4% of its original diameter,
Xt
safe strength,33 the divergence or ultimate strength.
Fiber compressive failure ( 11 < 0):

j11 j Numerical results and discussion


1 ð6Þ
Xc Despite the simplifications introduced to the numerical
simulations, the numerical results show a reasonable
Matrix tensile failure ( 22 > 0): agreement with the experimental ones (the maximum
Euclidean error norm is 8.57%), Figure 17. The results
 2
22 show that the ultimate strength decreases as the off-axis
þT  1 ð7Þ angle increases until 30 and then increases. This influ-
Yt
ence is a result of two different effects; the membrane
Matrix compressive Failure ( 22 < 0): stiffness and the failure mode. The specimens with
[0/90]2S has the maximum ultimate strength, due to
 2
22 the higher value of the in-plane stiffness. As the off-
þT  1 ð8Þ axis angle increases, ultimate strength decreases due
Yc

where

 2 1 þ 3 G  2 400
12 12 12
T¼ 2
ð9Þ
S 1 þ 32 G12 S2 Experimental

and  ij are the stress components, Xt is the tensile 360 Numerical


strength in the fiber direction, Yt is the tensile strength
Ultimate strength (MPa)

in the transverse direction, Xc is the compression


strength in the fiber direction, Yc is the compression 320
strength in the transverse direction, S is the shear
strength,  12 is the in-plane shear stress, a is a nonlinear
coefficient measured experimentally and G12 is the shear 280
modulus of elasticity.
The degradation rules set most of the material prop-
erties of the failed portions to 0.1 of the nonfailed prop-
240
erties. They use the value of 0.1 instead of zero to avoid
the convergence problems before the final failure of the
problem, Table 3.
Through the present study, the divergence of the 200
0 15 30 45
solution is assumed to occur when the joint is com-
Off-axisangle (Deg.)
pletely failed. The sequence converges to the exact solu-
tion if the inter polynomial satisfies two convergence Figure 17. Experimental vs. numerical ultimate strength at
requirements. The first one stated that the field different off-axis angles.

Table 3. The properties degradation rules.10

Failure mode Property degradation rules

Matrix tensile failure ( 2  0) Ed2 ¼ 0.1E2, Ed3 ¼ 0.1E3, Gd23 ¼ 0.1G23


Matrix compression failure ( 2  0) Ed2 ¼ 0.1E2, Ed3 ¼ 0.1E3, Gd23 ¼ 0.1G23
Fiber tensile and compression failure Ed1 ¼ 0.1E1

Downloaded from jcm.sagepub.com at Zagazig University on November 9, 2012


12 Journal of Composite Materials 0(0)

Matrix failure Fiber failure Matrix failure Fiber failure in


in 0o layer in 0o layer in 90o layer 90o layer

(a) [0/90] 2S

Matrix failure Fiber failure in Matrix failure Fiber failure in


in 15o layer 15o layer in -75o layer -75o layer

(b) [15/-75]2S

Matrix failure Fiber failure in Matrix failure Fiber failure in


in 30o layer 30o layer in -60o layer -60o layer

(c) [30/-60]2S

Matrix failure Fiber failure in Matrix failure Fiber failure in


in 45o layer 45o layer in -45o layer -45o layer

(d) [45/-45]2S

Figure 18. Predicted failure mode for various stacking sequences.

to the stiffness reduction. At 45 off-axis angle, the Figure 18 shows the in-situ failure mode for each
strength increases due to the difference in the joint stacking sequence. For [0/90]2S stacking sequence, the
main failure mechanism. The specimen with [30/ shear-out failure mode appears at layer with 0 orien-
60]2S has the minimum strength. These results are in tation angle due to matrix failure with a tendency to net
agreement with the experimental date. tension appears at layers of 90 orientation angle due to
The mechanical properties of composite mater- fiber and matrix failures. The bearing appears in both
ials have a remarkable scatter even when manufactured 0 and 90 due to fiber and matrix failure, respectively.
and tested under the same conditions. Figure 17 The [15/75]2S and [30/60]2S laminates had the same
shows reasonable agreement between the experimental failure mode but at different angles, following to the
and predicted strength. The Euclidean error reinforcement angle.
norm between the predicted strength and the average It is worth remarking that, for specimen with
experimental results is 5.39%, 4.50%, 7.21% and [45/45]2S stacking sequence, the two layers of 45
8.57% for specimens with 0 , 15 , 30 and 45 , and 45 orientation angles fails in the same way but
respectively. perpendicular to each other’s corresponding to the

Downloaded from jcm.sagepub.com at Zagazig University on November 9, 2012


Khashaba et al. 13

loading axis. This can be believed as they integrate each 2. Lim TS, Kim BC and Lee DG. Fatigue characteristics of
other i.e. the weakest axis of the layer 45 is the same the bolted joints for unidirectional composite laminates.
and the strongest axis of the layer 45 , with respect to Compos Struct 2006; 72: 58–68.
the loading axis, which leads to vanishing the shear 3. Mazumdar SK. Composites manufacturing; materials,
product and process engineering. New York: CRC Press,
failure and the only remain failure mode is the bearing
2002.
failure mode. 4. Ondurucu A, Esendemir U and Tunay RF. Progressive
failure analysis of glass–epoxy laminated composite
Conclusions pinned-joints. Mater Design 2012; 36: 617–625.
5. Seike S, Takao Y, Wang WX, et al. Bearing damage evo-
In this paper, a numerical and experimental study on lution of a pinned joint in CFRP laminates under
the effect of the stacking sequence of GFRP was con- repeated tensile loading. Int J Fatigue 2010; 32: 72–81.
ducted. A PDM was developed using the Hashin failure 6. Turvey GJ. Single-bolt tension joint tests on pultruded
criteria and a property degradation rule. The results GRP plate-effects of tension direction relative to pultru-
showed that such a simplified model could predict sion direction. Compos Struct 1998; 42: 341–351.
well the response of bolted joints in composite materials 7. Icten BM, Okutan B and Karakuzu R. Failure strength
of woven glass fiber- epoxy composites pin jointed.
with a reasonable computational time. The in-situ pro-
J Compos Mater 2003; 37: 1337–1350.
gressive damage helped to understand the specimen 8. Park HJ. Effect of stacking sequence and clamping force
final failure mode. Despite the simplifications intro- on the bearing strengths of mechanically fastened joints
duced to the numerical simulations, the numerical in composite laminates. Compos Struct 2001; 53: 213–221.
results show a reasonable agreement with the experi- 9. Hashin Z. Failure criteria for unidirectional Eber com-
mental ones (the maximum Euclidean error norm posites. J Appl Mech 1980; 47: 329–34.
is 8.57%). 10. Riccio A and Scaramuzzino F. Influence of damage onset
The highest ultimate strength was observed for and propagation on the tensile structural behavior of
pinned-joints in [0/90]2S stacking sequence. The min- protruding composite joints. In: 4th GRACM Congress
imum bearing and ultimate strength was observed for on Computational Mechanics, Patras, 27–29 June, 2002.
pinned-joints in [30/60]2S laminate. Loading the 11. Khashaba UA. Notched and pin bearing strengths of
GFRP composite laminates. J Compos Mater 1996; 30:
pinned-joints up to the ultimate value lead to shear-
2042–2055.
out failure mode for [0/90]2S, [15/75]2S and 12. Khashaba UA, Sallam HEM, Al-Shorbagy AE, et al.
[30/60]2S stacking sequences, while specimens with Effect of washer size and tightening torque on the per-
[45/45]2S stacking sequence are characterized by bear- formance of bolted joints in composite structures.
ing failure mode. Compos Struct 2006; 73: 310–317.
From the experimental discussion and the numerical 13. Khashaba UA, El-Sonbaty IA, Selmy AI, et al. Drilling
predicted failure mode, it can be recommended that for analysis of woven GFR/epoxy composites. Accepted for
mechanical FRP composite parts that assembled and Publication. J Compos Mater 2012. DOI: 10.1177/
subjected to bearing loads introducing 45 layer is 0021998312438620.
important to maximize the bearing load and displace- 14. Khashaba UA, El-Sonbaty IA, Selmy AI, et al.
ment. Layers with 90 are important for resisting the Machinability analysis in drilling woven GFR/epoxy
composites: Part I- Effect of Machining Parameters.
displacement of the pin and increasing the energy
Compos Part A 2010; 41: 391–400.
absorbed, which results in delaying the sudden failure 15. Khashaba UA. Drilling of polymer matrix composites:
of the component and, accordingly, increasing the A Review. J Composite Materials 2012. DOI: 10.1177/
structural reliability/durability. The presence of the 0 0021998312451609.
layer is important for maximizing the ultimate load to 16. McCarthy MA, Lawlor VP, Stanley WF, et al. Bolt-hole
failure and the apparent bearing stiffness of the joint. clearance effects and strength criteria in single-bolt,
single-lap, composite bolted joints. Compos Sci Technol
Funding 2002; 62: 1415–1431.
17. Lin HJ and Tsai CC. Failure analysis of bolted connec-
This research received no specific grant from any funding tions of composites with drilled and molded-in hole.
agency in the public, commercial, or not-for-profit sectors. Compos Struct 1995; 30: 159–168.
18. Maikuma H and Kubomura K. Bearing strength damage
Conflict of interest progress for PAM-based and pitch-based carbon fiber
None declared. composites. J Compos Mater 1993; 27: 1739–1761.
19. Irisarri FX, Laurin F, Carrere N, et al. Progressive
damage and failure of mechanically fastened joints in
References CFRP laminates – Part I: Refined Finite Element model-
1. Mallick PK. Fiber-reinforced composites, 2nd ed. New ling of single-fastener joints. Compos Struct 2012; 94:
York: Marcel Dekker Press, 1993. 2269–2277.

Downloaded from jcm.sagepub.com at Zagazig University on November 9, 2012


14 Journal of Composite Materials 0(0)

20. Yan Y, Wen W-D, Chang F-K, et al. Experimental study 27. Shokrieh MM and Attar PHM. A new method for mod-
on clamping effects on the tensile strength of composite eling of initiation and propagation of delamination
plates with a bolt-filled hole. Compos: Part A 1999; 30: between [0/y] layers of laminated composites. Appl
1215–1229. Compos Mater 2010; 17: 441–452.
21. Hollmann K. Failure analysis of bolted composite joints 28. Wisnom MR. Modeling discrete failures in composites
exhibiting in-plane failure modes. J Compos Mater 1996; with interface elements. Compos Part A 2010; 41:
30: 359–383. 795–805.
22. Khashaba UA, Aldousari SM and Najjar IMR. Behavior 29. Sebaey TA, Blanco N, Lopes CS, et al. Numerical inves-
of [0]8 woven composites under combined bending and tigation to prevent crack jumping in Double Cantilever
tension loading: Part - I experimental and analytical. Beam tests of multidirectional composite laminates.
J Compos Mater 2012; 46(11): 1345–1355. Compos Sci Technol 2011; 71: 1587–1592.
23. Adams DF and Walrath DE. Current status of the 30. Mahmoud FF, Ali-eldin SS, Hassan MM, et al. An incre-
Iosipescu shear test method. J Compos Mater 1987; 21: mental mathematical programming model for solving
494–507. multi-phase contact problems. Comput Struct 1998; 68:
24. Khashaba UA. In-plane shear properties of cross-ply 567–581.
laminate with different off-axis angles. Compos Struct 31. Gibson RF. Principles of composite material mechanics.
2004; 65: 167–177. New York: McGraw-Hill Publication, 1994.
25. Pakdil M. Failure analysis of composite single bolted- 32. Rao SS. The finite element method in engineering, 2nd ed.
joint subjected to pretension. Indian J Eng Mater Sci Elmsford, New York: Pergamon Press, 1989.
2009; 16: 79–85. 33. Zienkiewicz OC and Taylor RL. The finite element
26. De Morais AB, De Moura MF, Marques AT, et al. method. The basics. Vol. 1, 5th ed. Bristol, UK:
Mode-I interlaminar fracture of carbon/epoxy cross-ply Butterworth-Heinemann Publication, 2000.
composites. Compos Sci Technol 2002; 62: 679–686.

Downloaded from jcm.sagepub.com at Zagazig University on November 9, 2012

View publication stats

You might also like