Kapoor 2016
-
Upload
anonymous-syxsnn -
Category
Documents
-
view
225 -
download
0
Transcript of Kapoor 2016
-
8/18/2019 Kapoor 2016
1/9
High strain rate compression response of woven Kevlar reinforcedpolypropylene composites
Rajat Kapoor a , Laxman Pangeni a, Aswani Kumar Bandaru a , *, Suhail Ahmad a,Naresh Bhatnagar b
a Department of Applied Mechanics, Indian Institute of Technology Delhi, New Delhi, Indiab Department of Mechanical Engineering, Indian Institute of Technology Delhi, New Delhi, India
a r t i c l e i n f o
Article history:
Received 24 August 2015
Received in revised form
26 September 2015
Accepted 20 November 2015
Available online 12 January 2016
Keywords:
A. Polymer-matrix composites
B. Delamination
A. Thermoplastic resin
D. Electron microscopy
Kevlar
a b s t r a c t
In this study, experimental investigations on Kevlar ber reinforced polypropylene (PP) woven com-
posites under high strain rate compression loading are discussed. Kevlar/PP composite laminates with 8
and 24 layers are fabricated using vacuum assisted compression molding technique. Maleic anhydride
grafted-PP (MAg-PP) is added to PP to improve the interfacial property between Kevlar ber and PP resin.
The through-thickness properties at high strain rates from 1370 to 6066 s1 are obtained using split
Hopkinson pressure bar (SHPB) setup. The behavior of PP resin is found to be different than the
commonly used thermoset resins, such as epoxy. Dynamic stressestrain relations are drawn to reveal the
mechanical properties at high strain rates and these relations appear to be rate sensitive. As a result, the
peak stress increased by three times, toughness increased by almost ten times and strain at peak stress
increased by as much as two times with an increase in the strain rate. The nal failure of the specimens is
examined by scanning electron microscopy (SEM) to explore the possible failure mechanisms such as,
delamination, ber failure and shear fracture.© 2016 Elsevier Ltd. All rights reserved.
1. Introduction
Composite materials possess high specic strength and specic
stiffness with less fatigue. Due to this advantage of composite
materials, it's been widely used in military, aerospace and other
structural applications where weight of the structure is a signicant
parameter. Composite structures undergo different loading condi-
tions, such as, static and dynamic loads during their service life.
When the composite laminate is used as a body armor material,
the armor undergoes dynamic loading when projectile impacts the
target [1]. Among all the composite materials, Kevlar nds its majorapplication in body/vehicle armors as it exhibits an improved
impact resistance with lightweight. As the necessity increases, it is
very important to understand the effect of high strain rate on the
impact performance of Kevlar composite laminates. The response
of the material under different strain rates should be clearly known
for the effective useof materials [2]. Through-the-thickness loading
is one of the crucial condition in the ballistic impact applications.
The compressive properties of composite armor materials under
high strain rate conditions are highly desirable to assess the bal-
listic impact response.
Composite materials have been extensively characterized under
quasi-static tensile, compressive and shear loading conditions [3,4].
However, understanding the mechanical behavior of these mate-
rials under dynamic loading conditions is limited due to the asso-
ciated technical hitches at high strain rates. Most widely used
technique to characterize the materials at high strain rates is a split
Hopkinson pressure bar (SHPB) [2,5e7]. Recent works on the high
strain rate behavior of polymer composites were based on the
thermoset-based laminates made from glass and carbon bers[8e15]. Few works [7,16e19] reported on the dynamic compressive
response of Kevlar composites majorly based on the thermoset-
based matrix.
Woo et al. [7] used an acoustic emission technique to charac-
terize the failure progress in Kevlar/epoxy composites under high
strain rate impact. The peak stress and toughness of the Kevlar-
woven fabric specimen were increased almost two times, with an
increase in strain rate in the range of 1182e1460 s1, whereas the
strain at peak stress decreased by approximately 16%. An experi-
mental investigation was carried out by Daniel and Liber [16] to
assess the strain rate dependence on the tensile behavior of Kevlar/* Corresponding author.
E-mail address: [email protected] (A.K. Bandaru).
Contents lists available at ScienceDirect
Composites Part B
j o u r n a l h o m e p a g e : w w w . e l s e v i e r . co m / l o c a t e / c o m p o si t e s b
http://dx.doi.org/10.1016/j.compositesb.2015.11.044
1359-8368/©
2016 Elsevier Ltd. All rights reserved.
Composites Part B 89 (2016) 374e382
mailto:[email protected]://www.sciencedirect.com/science/journal/13598368http://www.elsevier.com/locate/compositesbhttp://dx.doi.org/10.1016/j.compositesb.2015.11.044http://dx.doi.org/10.1016/j.compositesb.2015.11.044http://dx.doi.org/10.1016/j.compositesb.2015.11.044http://dx.doi.org/10.1016/j.compositesb.2015.11.044http://dx.doi.org/10.1016/j.compositesb.2015.11.044http://dx.doi.org/10.1016/j.compositesb.2015.11.044http://www.elsevier.com/locate/compositesbhttp://www.sciencedirect.com/science/journal/13598368http://crossmark.crossref.org/dialog/?doi=10.1016/j.compositesb.2015.11.044&domain=pdfmailto:[email protected]
-
8/18/2019 Kapoor 2016
2/9
epoxy at a strain rate up to 27 s1 on the modied universal testing
machine (UTM). It was observed that, with an increase in strain
rate, modulus and strength of Kevlar/epoxy increased. Dynamic
response of Kevlar/Polyester composites was investigated by Har-
ding and Welsh [17] for cylindrical projectile up to a strain rate of
400 s1. They found that the tensile modulus increased within the
strain rate range from 104 to 103 s1 and reported the non-linear
response in dynamic tension. Zhu et al. [18] carried out static and
dynamic tests on Kevlar/polyester laminates. The damage pattern
observed for dynamic loading was different than that of the cor-
responding quasi-static case. Jacob et al. [20] carried out a detailed
review of the strain rate dependence on the mechanical properties
of polymer composites and a lot of contradiction in the data,
regarding the strain rate effect was reported.
From the above literature, it is evident that the mechanical
properties of the ber reinforced composite laminates are sensitive
to strain rate. Numerous studies have been conducted to charac-
terize the mechanical properties of Kevlar bers [19e22] and
Kevlar fabrics [23]. Wang and Xia [19,21] investigated the inuence
of strain rates (104 to 103 s1) and temperature(60 to 90 C) on
Kevlar 49 ber bundles. Results show that the mechanical prop-
erties of Kevlar 49 bers are sensitive to strain rate and tempera-
ture. It was also reported that at a constant temperature, initialelastic modulus, strength, and failure strain increase with an in-
crease in the strain rate, and for a xed strain rate, the initial elastic
modulus decreases and failure strain increases with increase in test
temperature. The experimental investigation of Lim et al. [22]
included three high-performance bers (Kevlar, Kevlar 129, and
Twaron) fabricated at different times over a period of ten years.
Their experimental results show that longitudinal tensile strength
of the bers weakly dependent on the ber gage length and re-
ported the insignicant strain rate effect on tensile strengths (only
by a few percent). Tensile tests of Zhu et al. [23] on Kevlar 49 plain
weave fabric at strain rates ranging from 25 to 170 s1 reported that
the dynamic material properties in terms of Young's modulus,
tensile strength, maximum strain and toughness increase with an
increase in the strain rate.There is a lack of experimental studies on the composite laminas
made from Kevlar bers reinforced with a thermoplastic resin. The
majority of the works reported in the literature were concentrated
on the high strain rate behavior of thermoset-based composite
laminates. Limited research has been reported on the dynamic
response of thermoplastic-based Kevlar composites [24,25].
Rodriguez et al. [24] examined the effects of high strain rate on
polyethylene and aramid woven fabrics. They deduced from their
results that dynamic stressestrain curve is more linear as
compared to the static one. Dynamic compression tests and quasi-
static tests of Viswanathan et al. [25] on Kevlar 29/polyethylene
showed signicant increase in the tensile strength and decrease in
failure strain at high strain rates as compared to that of the quasi-
static tests. Carillo et al. [26] reported that the addition of poly-propylene (PP) to aramid (Kevlar 129) fabrics shows improved
impact resistance. However, low adhesion was reported between
aramid fabrics and PP matrix. There is no consistent data available
for the characterization of dynamic response of thermoplastic-
based composite laminates made from Kevlar bers and PP ma-
trix. To the best of the author's knowledge, the compressive prop-
erties of Kevlar ber reinforced thermoplastic composites in the
through-thickness direction at high strain rates have not been re-
ported in the open literature.
In the present study, the through-thickness compressive prop-
erties of thermoplastic-based composite laminates made from
Kevlar ber and maleic anhydride grafted-PP (MAg-PP) matrix are
reported at high strain rates ranging from 1370 to 6066 s1. Grafted
maleic anhydride-PP is added to PP resin to improve the adhesion
between Kevlar ber and PP resin. Kevlar/MAg-PP (K-MPP) com-
posite laminates consisting of 8 and 24 layers are fabricated using
vacuum assisted compression molding machine. High strain rate
compressive properties of the K-MPP composite laminates with
respect to through-the-thickness direction are characterized using
SHPB. The compressive stressestrain behavior, toughness,
compressivepeak stressand strain at peak stress are studied for the
fabricated composite laminates. Following the experiments, failure
mechanisms are characterized through scanning electron micro-
scopy (SEM).
2. Experiments
2.1. Fabrication of composite laminates
Kevlar 29 yarns with 1000 denier were woven into plain woven
fabric with areal density of 364 g/m2. Yarns made of thousands of
bers, were woven into this fabric. The yarn tenacity was 14.91 g/
den with 40 ends/inch in both weft and warp directions. Thermo-
plastic polymer PP was selected as matrix due to its lightweight and
density of 0.855 g/cm3. The interfacial property between Kevlar
fabric and PP was improved by adding a coupling agent called,
maleic anhydride grafted PP (MAg-PP). The main function of thiscoupling agent is linking of bers to the polymer matrix and
reducing the pull out while increasing the impact and tensile
strength. Kevlar/MAg-PP (K-MPP) composite laminates were
fabricated using vacuum assisted compression molding machine.
PP sheets were coated with grafted maleic anhydride (MAg) to
improve the interfacial property between Kevlar and PP resin. The
alternate layers of Kevlar fabric and MAg-PP sheets were stacked
and placed in the vacuum chamber containing at plate molds. The
fabric weave pattern and stacking sequence of the preform are
shown in Fig. 1. The specimens were heated at a temperature of
200 C under 10 bar pressure and cooled to room temperature. The
matrix (MAg-PP sheet) thickness was 0.05e0.1 mm and the Kevlar
fabric thickness was 0.15e0.2 mm. To avoid formation of voids
during the fabrication process, a vacuum was maintained at550 mm of Hg in a vacuum chamber. K-MPP laminates with 8 layers
(1.6e1.7 mm thick) and 24 layers (4.3e4.6 mm thick) were cut out
from a square plate of 160 160 mm, through laser cutting to
prepare the test specimen for dynamic characterization (Fig.1). The
ber volume fraction was measured through burn off test according
to ASTM-D-2584-02 and it was observed to be 50e57%.
The critical slenderness ratio (l/d) is imperative to ensure that
the results obtained from the experiment reect the desired ma-
terial properties. Inertial effects produce stress waves along the
radial and the axial directions of the specimen. If the ends of the
specimen are well lubricated, it minimizes the inertial effects. The
correction for friction depends on the l/d ratio of the specimen. The
l/d ratio of 0.3e0.5 was shown to be good criteria of SHPB specimen
design [27]. However, it was also suggested that the tests on thematerials with high ow stress and low density are less prone to
such inertial errors [28]. As the material used in the present study is
of low density, low l/d ratio was considered to avoid inertial effects.
The diameter of the specimen was considered in the range of
15.5e16 mm. Length of the specimen was governed by the thick-
ness of the laminate which comes out to be approximately
1.6e1.7 mm for 8-layer specimens and 4.3e4.6 mm for 24-layer
specimens. Therefore, the l/d ratio of the present specimens was in
the range of 0.1e0.3.
2.2. High strain rate test
High strain rate experiments were performed using SHPB
apparatus available at Impact Mechanics Lab, Department of
R. Kapoor et al. / Composites Part B 89 (2016) 374e 382 375
-
8/18/2019 Kapoor 2016
3/9
Applied Mechanics, IIT Delhi. The schematic representation of the
SHPB setup is shown in Fig. 2.
The specications of the SHPB setup were as follows:
Bar diameter e Striker bar, Incident bar and Transmission bar e
20 mm
Bar length e Striker bar e 300 mm; Incident bar e 1500 mm;
Transmission bar e 1500 mm
Bar material properties e Maraging steel: r e 8000 kg/m3; E e
183 GPa; Elastic wave speed (C e) of bars e 5091 m/s
Strain gage properties e Gage factor(G) e 5; Initial voltage (U ) e
5 V; and Amplication factor ( A) e 11
Detailed theory and technique involved in SHPB are well
described in the literature [29,30]. Therefore, a detailed discussion
is avoided in the present study. A dynamic stressestrain relation
can be obtained by measuring the incident, transmitted and re-
ected stress waves. The strain gage mounted on the incident bar
measures the both incident pulse (V i) and the reected pulse (V r)
while the strain gage on the output bar measures the transmitted
Striker Specimen Transmission bar Incident bar
Strain gauge 2
Signal conditioner & amplifier Signal conditioner & amplifier
Data acquisition &
storage
Strain gauge 3
Fig. 2. Schematic setup of compression SHPB.
!!!!!!
(a) (b) (c)
MAg-PP matrix(25 layers)
Kevlar fabric
24 la ers
8 layers
24 layers
Fig. 1. Schematic diagrams: (a) 2D plain weave pattern, (b) stacking sequence for 24-layer composite and (c) dynamic test specimen.
(a) (b)
0.0154 0.0156 0.0158 0.0160
-0.05
0.00
0.05
0.10
V o l t a g e ( V )
Time (sec)
8 layer 5335/s
Incident & reflective wave
Transmitted wave
0.0112 0.0114 0.0116 0.0118
-0.05
0.00
0.05
0.10
V o l t a g e ( V
)
Time (sec)
8 layer 6066/s
Incident & reflective wave
Transmitted wave
Fig. 3. Input and output wave signals of 8-layer K-MPP laminates in through-the-thickness compression: (a) 5335 s
1
and (b) 6066 s
1
.
R. Kapoor et al. / Composites Part B 89 (2016) 374e 382376
-
8/18/2019 Kapoor 2016
4/9
pulse (V t). These values are in volts and were converted into strain
signals, including transmitted strain signal 3t and reected strain
signal 3r by using the following equations:
3r ¼ 2V rGUA
(1)
3t ¼ 2V tGUA
(2)
where G is the gage factor, U is the input voltage and A is the
amplication factor.
By using Eqs. (1) and (2), engineering stress (s), strain ( 3) andstrain rate ð _ 3Þ dened on the specimen length are measured using
the following equations [31]:
s ¼ E Ab As
3t (3)
3¼ 2C e
Ls
Z t
0
3rdt (4)
_ 3¼ 2C e
Ls 3r (5)
where Ab, As, Ls are the cross-sectional areas of pressure bar, cross-
sectional area and length of the specimen, respectively; E and C e are
Young's modulus and the wave velocity in the pressure bars,
respectively. Then the engineering stress is converted into true
stress by using the following equations [32,33]:
strue ¼ s½1 3 (6)
(a) (b)
0.0112 0.0114 0.0116 0.0118
-0.05
0.00
0.05
0.10
V o l t a g e ( V )
Time (sec)
24 layer 2538/s
Incident & reflective wave
Transmitted wave
0.0098 0.0100 0.0102 0.0104
-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
V o l t a g e ( V )
Time (sec)
24 layer 3440/s
Incident & reflective wave
Transmitted wave
Fig. 4. Input and output wave signals of 24-layer K-MPP laminates in through-the-thickness compression: (a) 2538 s 1 and (b) 3440 s1.
0.00000 0.00005 0.00010 0.00015 0.00020 0.00025
0
100
200
300
400
500
600
S t r e s s ( M P a )
Time (sec)
24 layer K-MPP
2005/s
3239/s
0.00000 0.00005 0.00010 0.00015 0.00020
0
4
8
12
16
S t r a i n ( % )
Time (sec)
24 layer K-MPP
2005/s
3239/s
Fig. 5. Variation of stress and strain with time at different strain rates.
R. Kapoor et al. / Composites Part B 89 (2016) 374e 382 377
-
8/18/2019 Kapoor 2016
5/9
3true ¼ ln½1 3 (7)
3. Results and discussion
3.1. Dynamic stressestrain behavior
High strain rate tests are performed on SHPB for 8-layer and 24-
layer K-MPP laminates at strain rates from 1370 to 6066 s1. The
strain waves sensed by strain gages mounted on the incident and
transmission bars at strain rates of 5335 and 6066 s1
for 8-layerlaminate and 2538 and 3440 s1 for 24-layer laminate are shown in
Figs. 3 and 4.
The transient data for each specimen tested under high strain
rate is recorded and saved. The incident wave is a function of
impact velocity and the data is initiated as the wave reaches the
location of strain gage on the incident bar. The reected wave is not
uniform for a long time. The reected wave reaches a maximum
value and oscillates at a constant value and approaches zero. Strain
rate versus time and stress versus time response are saved, to plot
dynamic stressestrain behavior. The time at which transmitting
pulse starts deviating from zero, is selected as the starting time and
when it drops tozerois selectedas the end time. The portions of the
reected pulse are taken to the equivalent time range and incor-
porated to obtain strain versus stress data using Eqs. (3)e(5) at each
strain rate. The stress versus time and strain rate versus time plots
for high strain rate compressive behavior of 24-layer K-MPP lami-
nate is shown in Fig. 5.
The dynamic true stress-true strain behavior of 24-layer and 8-
layer K-MPP laminates through-the-thickness direction at high
strain rates is shown in Fig. 6.
0 3 6 9 12 15 18
0
100
200
300
400
500
600
T r u e S t r e s
s ( M P a )
True Strain (%)
8 layer Kevlar/MAg-PP
@ 5335/s strain rate
@ 6066/s strain rate
0 2 4 6 8 10 12
0.0
2.0x108
4.0x108
6.0x108
24 layer Kevlar/MAg-PP
@ 2538/s strarin rate
@ 3239/s strain rate
T r u e S t r e s s
( M P a )
True Strain (%)
(a) (b)
Fig. 6. Dynamic true stresse
true strain response of K-MPP (a) 8 layers and (b) 24 layers.
Fig. 7. Dynamic true stresse
true strain behavior of 24-layer K-MPP laminate.
R. Kapoor et al. / Composites Part B 89 (2016) 374e 382378
-
8/18/2019 Kapoor 2016
6/9
It is observed that for 8-layer K-MPP laminates lots of un-
dulations and inconsistencies were recorded in the experimental
results (Fig. 6a). These inconsistencies could be attributed to the
low l/d ratio of the test specimen. Further, a series of initial ex-
periments were conducted on the specimens with different diam-
eter values and consistent results were obtained for 24-layer
specimens with an l/d ratio of 0.3 (Fig. 6b). Hence, further discus-
sions are emphasized for 24-layer composites. From the stresse-
strain behavior of laminates under various strain rates, the effect of
strain rate on the compressive dynamic properties of K-MPP lam-
inates are studied in terms of compressive peak stress, strain at
peak stress, compression modulus and toughness.
Fig. 7 shows the dynamic true stressetrue strain curves of plain
woven 24-layer K-MPP laminate in the through-thickness direction
subjected to different strain rates. As the matrix is dominant in
through-the-thickness behavior of composites, the compressive
stress vs strain behavior obtained from the present tests are qual-
itatively comparable with the dynamic stress vs strain behavior of
PP [34] (Fig. 7b). Due to the presence of Kevlar bers, exact and
quantitative comparison of the results cannot be made. The high
strain rate compressive properties measured by SHPB tests for 8-
layer and 24-layer K-MPP laminates, are listed in Table 1.
3.2. In uence of strain rate on the dynamic compressive properties
The inuence of strain rate on the dynamic compressive
response of 24-layer K-MPP specimens is dened in terms of peak
stress, strain at peak stress, toughness and compressive stiffness.
Fig. 8 shows the effect of strain rate on the compressive peak
stress of K-MPP (24 layers) laminate at various strain rates. In this
case the peak stress, increased from 220.32 MPa to 652.91 MPa at
strain rates of 1370 s1 and 4264 s1, respectively, with an increase
of 196.35% over the given range of strain rates.
The relation between the compressive peak stresses (smax) with
strain rate can be tted through a linear curve with the following
equation:
smax ¼ 0:165_ 3 21:362 (8)
Fig. 9 depicts the effect of strain rate on the failure strain at
different strain rates for 24-layer K-MPP laminates. The strain at
peak stress increases linearly with strain rate and it is evident fromFig. 7a. This can be attributed to the ductile failure of the specimen,
as a result of which the damage propagation in the material be-
comes slow with an increase in the strain rate, which indicates
higher impact resistance of the material. The strain at peak stress,
increased approximately by 133.8% over the given range of strain
rates.
The linear relationship between the strains at peak stress ( 3f )
and strain rate can be tted as:
3f ¼ 0:003_ 3þ 2:686 (9)
In the case of Kevlar/epoxy laminates [7], strain at peak stress
decreased as the strain rate increased. But in the present study, for
K-MPP laminates, strain at peak stress, increases with the strainrate due to the ductile behavior of Kevlar ber and PP matrix. This
phenomenon makes the present material desirable for armor ap-
plications due to increased energy absorption.
Fig. 10 illustrates the toughness of 24-layer K-MPP laminates at
various strain rates. It can be observed that the toughness increases
with the strain rate and the trend of increase is polynomial in na-
ture. The increase in toughness is 808.8%.
Table 1
Dynamic properties of K-MPP laminates.
No. of layers Strain rate (s1) Young's modulus (GPa) Toughness (MPa) Peak stress (MPa) Failure strain
8 5335 102 60.62 407.98 0.170
6066 143 66.18 544.00 0.146
24 1370 38 11.66 220.32 0.065
2005 69 21.37 264.40 0.086
2538 81 33.80 405.00 0.100
3239 105 54.22 515.81 0.1143440 111 62.55 591.34 0.117
4264 131 105.97 652.91 0.152
1000 1500 2000 2500 3000 3500 4000 4500
100
200
300
400
500
600
700
800
P e a k s t r e s s ( M P a )
Strain rate (s-1)
Experimemt
Linear Fit of Peak stress
Fig. 8. In
uence of strain rate on compressive peak stress (24 layers). Fig. 9. In
uence of strain rate on failure strain at peak stress (24 layers).
R. Kapoor et al. / Composites Part B 89 (2016) 374e 382 379
-
8/18/2019 Kapoor 2016
7/9
The polynomial relationship between toughness and strain rate
can be tted as following equation:
E t ¼ 0:000009_ 32 0:017_ 3þ 20:013 (10)
Fig. 11 shows the variation of compression stiffness at various
strain rates. As the strain rate increases the compressive stiffness
increases. The increase of compression stiffness is almost linear and
can be tted with the following equation:
E C ¼ 0:032_ 3 0:062 (11)
The peak stress values obtained for K-MPP laminates are
compared with Kevlar/epoxy reported in Ref. [7]. But the length of
the specimens in Woo and Kim [7] was 8 mm while in the present
study it is 4 mm. According to Eqs. (3) and (6), the dynamic stress is
independent of the length and depends primarily on the cross
section area of the bar and the specimen. However, for the purposeof comparison, analytical calculations were performed in MATLAB
by doubling the length of the present specimen, to obtain an esti-
mate of the peak stress value. The response obtained from the
present study for 8 mm length was not exact, but it is assumed that
the results remain in the similar range. The strain rate values of K-
MPP laminate, which are in close range with the Kevlar/epoxy are
considered for the comparison. Peak stress values of Ref. [7] are
376 MPa and 161 MPa for strain rates of 1460 s1 and 1182 s1,
respectively. Assuming that the stress value remains unaffected by
the length of the specimen, the obtained peak stress from the
present study is 437 MPa and 190 MPa for strain rates of 1452 s 1
and 1173 s1, respectively. Hence, it indicates that the compressive
properties of K-MPP laminateshas better impact response than that
of the Kevlar/epoxy because the failure behavior of Kevlar/epoxy is
brittle and for K-MPP it is ductile in nature.
3.3. Fractography
The morphologies of the composite specimens are investigated
by SEM. The SEM micrographs of 24-layer K-MPP laminates at
different strain rates are studied. Since the loading is in through-
the-thickness direction, the failure initiates from the surface edge
of the specimen (Fig.12), which is in line with the loading direction.
As the strain rate increases the edge failure increases and inter
laminar shear stresses generate which are expected to be the
probable cause for the occurrence delamination. The progress of
the delamination at different strain rates can be seen in Fig. 12.
The failure modes of the samples in through-the-thickness di-rection at strain rates of 2005 s1 and 3440 s1 are illustrated in
Fig. 13a and b respectively, through SEM micrographs. The SEM
micrographs choose from selected locations on the samples to
highlight the dominant failure modes. These micrographs illustrate
clearly the failure modes through delamination, ber deformation,
shear fracture, ber failure, ber buckling and matrix cracking.
Fig. 13(a) shows the shear fracture, delamination, deformation
of bers and ber crush on the impact face. The crush is induced by
the compressive loading. It is observed that, after the impact, the
specimen is marginally damaged and is characterized by ber crush
and the shedding of bers (Fig.13(b)). It indicates that the specimen
undergoes a signicant compressive stress during impact, severe
shedding of bers and damage due to surcial friction. While
Fig. 10. Inuence of strain rate on toughness (24 layers).
Fig. 11. Inuence of strain rate on compressive stiffness (24 layers).
Fig. 12. SEM micrographs of the edge failure at strain rates of 2005, 3239 and 4349 s
1
.
R. Kapoor et al. / Composites Part B 89 (2016) 374e 382380
-
8/18/2019 Kapoor 2016
8/9
comparing Fig. 13(a) to (b) at higher strain rates, the material
showed more delamination and several ber breakages which may
possibly due to the changes in local boundary bonding.
4. Conclusions
In the present study, experimental investigations were carried
out on 8- and 24-layer Kevlar/MAg-PP (K-MPP) composite lami-
nates in the through-thickness direction under high strain rate
loading. Interfacial behavior between Kevlar and PP was improved
by adding grafted maleic anhydride PP to the matrix. The high
strain rate tests were conducted using split Hopkinson pressure bar
(SHPB) apparatus. The samples were tested in the strain rate range
of 1370 s1 to 6066 s1. The dynamic true stressetrue strain
behavior of composite specimens wasobtained and the inuence of
strain rate on the through-thickness compression properties was
studied. At high strain loading, the though-thickness compressive
properties were rate sensitive. Especially, the strain at peak stress
for K-MPP increased with the strain rate, whereas for Kevlar/epoxylaminates it decreased. This behavior can be attributed to the
ductile behavior of Kevlar fabric and PP matrix. Fractography
through SEM revealed various failure modes such as, matrix
cracking, shear fracture, delamination, ber failure, ber defor-
mation, friction between bers and the shedding of bers. The
ductile response of K-MPP laminates observed in the present study
can be adopted in the design of composite laminates under high
strain rate loading conditions, such as ballistic impact and impul-
sive loading.
References
[1] Sun B, Gu B, Ding X. Compressive behavior of 3-D angle-interlock woven
fabric composites at various strain rates. Polym Test 2005;24(4):447e
54.
[2] Naik NK, Yernamma P, Thoram NM, Gadipatri R, Kavala VR. High strain ratetensile behavior of woven fabric E-glass/epoxy composite. Polym Test2010;29(1):14e22.
[3] Sevkat E, Liaw B, Delale F, Basavaraju B. A combined experimental and nu-merical approach to study ballistic impact response of S2-glass ber/tough-
ened epoxy composite beams. Compos Sci Technol 2009;69(7e
8):965e
82.[4] Valença SL, Griza S, de Oliveira VG, Sussuchi EM, de Cunha FGC. Evaluation of
the mechanical behavior of epoxy composite reinforced with Kevlar plainfabric and glass/Kevlar hybrid fabric. Compos Part B Eng 2015;70(1):1e8.
[5] Chen WW, Song B. Split Hopkinson bar design, testing and applications.Mechanical engineering series. London: Springer; 2011. p. 11e80.
[6] Foroutan R, Nemes J, Ghiasi H, Hubert P. Experimental investigation of highstrain-rate behavior of fabric composites. Compos Struct 2013;106:264e9.
[7] Woo SC, Kim WT. High-strain-rate impact in Kevlar woven composites andfracture analysis using acoustic emission. Compos Part B Eng 2014;60:125e36.
[8] Hosur MV, Abraham A, Jeelani S, Vaidya UK. Studies on the inuence of through-the-thickness reinforcement on low velocity and high strain rateresponse of S2-glass/vinylester composites. J Compos Mater 2001;35(12):1111e33.
[9] Karkkainen RL. Dynamic micromechanical modeling of textile compositestrength under impact and multi-axial loading. Compos Part B Eng 2015;83:27e35.
[10] Walter TR, Subhash G, Sankar BV, Yen CF. Damage modes in 3D glass ber
epoxy woven composites under high rate of impact loading. Compos Part BEng 2009;40(6):584e9.
[11] Song Z, Wang Z, Ma HW, Xuan HJ. Mechanical behavior and failure mode of woven carbon/epoxy laminate composites under dynamic compressiveloading. Compos Part B Eng 2014;60:531e6.
[12] Li DS, Lu ZX, Jiang N, Fang DN. High strain rate behavior and failure mecha-nism of three-dimensional ve-directional carbon/phenolic braided compos-ites under transverse compression. Compos Part B Eng 2011;42(2):309e17.
[13] Arbaoui J, Tarfaoui M, Alaoui AEM. Mechanical behavior and damage kineticsof woven E-glass/vinylester laminate composites under high strain rate dy-namic compressive loading: experimental and numerical investigation. Int JImpact Eng 2016;87.
[14] Hosur MV, Alexander J, Vaidya UK, Jeelani S. High strain rate compressionresponse of carbon/epoxy laminate composites. Compos Struct 2001;52(3e4):405e17.
[15] Jadhav A, Woldesenbet E, Pang SS. High strain rate properties of balancedangle-ply graphite/epoxy composites. Compos Part B Eng 2003;34(4):339e46.
[16] Daniel I, Liber T. Strain rate effects on mechanical properties of ber com-posites. Report NASA CR-135087, Part-III, 4. 1976. p. 18.
Fig. 13. SEM micrographs of K-MPP specimen at strain rates of (a) 2005 s1 and (b) 3239 s1.
R. Kapoor et al. / Composites Part B 89 (2016) 374e 382 381
http://refhub.elsevier.com/S1359-8368(15)00758-1/sref1http://refhub.elsevier.com/S1359-8368(15)00758-1/sref1http://refhub.elsevier.com/S1359-8368(15)00758-1/sref1http://refhub.elsevier.com/S1359-8368(15)00758-1/sref2http://refhub.elsevier.com/S1359-8368(15)00758-1/sref2http://refhub.elsevier.com/S1359-8368(15)00758-1/sref2http://refhub.elsevier.com/S1359-8368(15)00758-1/sref2http://refhub.elsevier.com/S1359-8368(15)00758-1/sref3http://refhub.elsevier.com/S1359-8368(15)00758-1/sref3http://refhub.elsevier.com/S1359-8368(15)00758-1/sref3http://refhub.elsevier.com/S1359-8368(15)00758-1/sref3http://refhub.elsevier.com/S1359-8368(15)00758-1/sref3http://refhub.elsevier.com/S1359-8368(15)00758-1/sref3http://refhub.elsevier.com/S1359-8368(15)00758-1/sref3http://refhub.elsevier.com/S1359-8368(15)00758-1/sref4http://refhub.elsevier.com/S1359-8368(15)00758-1/sref4http://refhub.elsevier.com/S1359-8368(15)00758-1/sref4http://refhub.elsevier.com/S1359-8368(15)00758-1/sref4http://refhub.elsevier.com/S1359-8368(15)00758-1/sref5http://refhub.elsevier.com/S1359-8368(15)00758-1/sref5http://refhub.elsevier.com/S1359-8368(15)00758-1/sref5http://refhub.elsevier.com/S1359-8368(15)00758-1/sref6http://refhub.elsevier.com/S1359-8368(15)00758-1/sref6http://refhub.elsevier.com/S1359-8368(15)00758-1/sref6http://refhub.elsevier.com/S1359-8368(15)00758-1/sref7http://refhub.elsevier.com/S1359-8368(15)00758-1/sref7http://refhub.elsevier.com/S1359-8368(15)00758-1/sref7http://refhub.elsevier.com/S1359-8368(15)00758-1/sref7http://refhub.elsevier.com/S1359-8368(15)00758-1/sref8http://refhub.elsevier.com/S1359-8368(15)00758-1/sref8http://refhub.elsevier.com/S1359-8368(15)00758-1/sref8http://refhub.elsevier.com/S1359-8368(15)00758-1/sref8http://refhub.elsevier.com/S1359-8368(15)00758-1/sref8http://refhub.elsevier.com/S1359-8368(15)00758-1/sref8http://refhub.elsevier.com/S1359-8368(15)00758-1/sref8http://refhub.elsevier.com/S1359-8368(15)00758-1/sref8http://refhub.elsevier.com/S1359-8368(15)00758-1/sref9http://refhub.elsevier.com/S1359-8368(15)00758-1/sref9http://refhub.elsevier.com/S1359-8368(15)00758-1/sref9http://refhub.elsevier.com/S1359-8368(15)00758-1/sref9http://refhub.elsevier.com/S1359-8368(15)00758-1/sref10http://refhub.elsevier.com/S1359-8368(15)00758-1/sref10http://refhub.elsevier.com/S1359-8368(15)00758-1/sref10http://refhub.elsevier.com/S1359-8368(15)00758-1/sref10http://refhub.elsevier.com/S1359-8368(15)00758-1/sref10http://refhub.elsevier.com/S1359-8368(15)00758-1/sref10http://refhub.elsevier.com/S1359-8368(15)00758-1/sref11http://refhub.elsevier.com/S1359-8368(15)00758-1/sref11http://refhub.elsevier.com/S1359-8368(15)00758-1/sref11http://refhub.elsevier.com/S1359-8368(15)00758-1/sref11http://refhub.elsevier.com/S1359-8368(15)00758-1/sref11http://refhub.elsevier.com/S1359-8368(15)00758-1/sref12http://refhub.elsevier.com/S1359-8368(15)00758-1/sref12http://refhub.elsevier.com/S1359-8368(15)00758-1/sref12http://refhub.elsevier.com/S1359-8368(15)00758-1/sref12http://refhub.elsevier.com/S1359-8368(15)00758-1/sref12http://refhub.elsevier.com/S1359-8368(15)00758-1/sref12http://refhub.elsevier.com/S1359-8368(15)00758-1/sref13http://refhub.elsevier.com/S1359-8368(15)00758-1/sref13http://refhub.elsevier.com/S1359-8368(15)00758-1/sref13http://refhub.elsevier.com/S1359-8368(15)00758-1/sref13http://refhub.elsevier.com/S1359-8368(15)00758-1/sref14http://refhub.elsevier.com/S1359-8368(15)00758-1/sref14http://refhub.elsevier.com/S1359-8368(15)00758-1/sref14http://refhub.elsevier.com/S1359-8368(15)00758-1/sref14http://refhub.elsevier.com/S1359-8368(15)00758-1/sref14http://refhub.elsevier.com/S1359-8368(15)00758-1/sref15http://refhub.elsevier.com/S1359-8368(15)00758-1/sref15http://refhub.elsevier.com/S1359-8368(15)00758-1/sref15http://refhub.elsevier.com/S1359-8368(15)00758-1/sref15http://refhub.elsevier.com/S1359-8368(15)00758-1/sref16http://refhub.elsevier.com/S1359-8368(15)00758-1/sref16http://refhub.elsevier.com/S1359-8368(15)00758-1/sref16http://refhub.elsevier.com/S1359-8368(15)00758-1/sref16http://refhub.elsevier.com/S1359-8368(15)00758-1/sref16http://refhub.elsevier.com/S1359-8368(15)00758-1/sref16http://refhub.elsevier.com/S1359-8368(15)00758-1/sref16http://refhub.elsevier.com/S1359-8368(15)00758-1/sref15http://refhub.elsevier.com/S1359-8368(15)00758-1/sref15http://refhub.elsevier.com/S1359-8368(15)00758-1/sref15http://refhub.elsevier.com/S1359-8368(15)00758-1/sref14http://refhub.elsevier.com/S1359-8368(15)00758-1/sref14http://refhub.elsevier.com/S1359-8368(15)00758-1/sref14http://refhub.elsevier.com/S1359-8368(15)00758-1/sref14http://refhub.elsevier.com/S1359-8368(15)00758-1/sref14http://refhub.elsevier.com/S1359-8368(15)00758-1/sref13http://refhub.elsevier.com/S1359-8368(15)00758-1/sref13http://refhub.elsevier.com/S1359-8368(15)00758-1/sref13http://refhub.elsevier.com/S1359-8368(15)00758-1/sref13http://refhub.elsevier.com/S1359-8368(15)00758-1/sref12http://refhub.elsevier.com/S1359-8368(15)00758-1/sref12http://refhub.elsevier.com/S1359-8368(15)00758-1/sref12http://refhub.elsevier.com/S1359-8368(15)00758-1/sref12http://refhub.elsevier.com/S1359-8368(15)00758-1/sref11http://refhub.elsevier.com/S1359-8368(15)00758-1/sref11http://refhub.elsevier.com/S1359-8368(15)00758-1/sref11http://refhub.elsevier.com/S1359-8368(15)00758-1/sref11http://refhub.elsevier.com/S1359-8368(15)00758-1/sref10http://refhub.elsevier.com/S1359-8368(15)00758-1/sref10http://refhub.elsevier.com/S1359-8368(15)00758-1/sref10http://refhub.elsevier.com/S1359-8368(15)00758-1/sref10http://refhub.elsevier.com/S1359-8368(15)00758-1/sref9http://refhub.elsevier.com/S1359-8368(15)00758-1/sref9http://refhub.elsevier.com/S1359-8368(15)00758-1/sref9http://refhub.elsevier.com/S1359-8368(15)00758-1/sref9http://refhub.elsevier.com/S1359-8368(15)00758-1/sref8http://refhub.elsevier.com/S1359-8368(15)00758-1/sref8http://refhub.elsevier.com/S1359-8368(15)00758-1/sref8http://refhub.elsevier.com/S1359-8368(15)00758-1/sref8http://refhub.elsevier.com/S1359-8368(15)00758-1/sref8http://refhub.elsevier.com/S1359-8368(15)00758-1/sref7http://refhub.elsevier.com/S1359-8368(15)00758-1/sref7http://refhub.elsevier.com/S1359-8368(15)00758-1/sref7http://refhub.elsevier.com/S1359-8368(15)00758-1/sref7http://refhub.elsevier.com/S1359-8368(15)00758-1/sref6http://refhub.elsevier.com/S1359-8368(15)00758-1/sref6http://refhub.elsevier.com/S1359-8368(15)00758-1/sref6http://refhub.elsevier.com/S1359-8368(15)00758-1/sref5http://refhub.elsevier.com/S1359-8368(15)00758-1/sref5http://refhub.elsevier.com/S1359-8368(15)00758-1/sref5http://refhub.elsevier.com/S1359-8368(15)00758-1/sref4http://refhub.elsevier.com/S1359-8368(15)00758-1/sref4http://refhub.elsevier.com/S1359-8368(15)00758-1/sref4http://refhub.elsevier.com/S1359-8368(15)00758-1/sref4http://refhub.elsevier.com/S1359-8368(15)00758-1/sref3http://refhub.elsevier.com/S1359-8368(15)00758-1/sref3http://refhub.elsevier.com/S1359-8368(15)00758-1/sref3http://refhub.elsevier.com/S1359-8368(15)00758-1/sref3http://refhub.elsevier.com/S1359-8368(15)00758-1/sref3http://refhub.elsevier.com/S1359-8368(15)00758-1/sref2http://refhub.elsevier.com/S1359-8368(15)00758-1/sref2http://refhub.elsevier.com/S1359-8368(15)00758-1/sref2http://refhub.elsevier.com/S1359-8368(15)00758-1/sref2http://refhub.elsevier.com/S1359-8368(15)00758-1/sref1http://refhub.elsevier.com/S1359-8368(15)00758-1/sref1http://refhub.elsevier.com/S1359-8368(15)00758-1/sref1
-
8/18/2019 Kapoor 2016
9/9
[17] Harding J, Welsh LM. A tensile testing technique for bre-reinforced com-posites at impact rates of strain. J Mater Sci 1983;18:1810e26.
[18] Zhu G, Goldsmith W, Dharan CKH. Penetration of laminated Kevlar by pro- jectiles I e experimental investigations. Int J Solids Struct 1992;29(4):399e420.
[19] Wang Y, Xia Y. The effects of strain rate on the mechanical behaviour of Kevlarbre bundles: an experimental and theoretical study. Compos Part A1998;29(11):1411e5.
[20] Jacob GC, Starbuck JM, Fellers JF, Simunovic S, Boeman RG. Strain rate effectson the mechanical properties of polymer composite materials. J Appl Polym
Sci 2004;94:296e
301.[21] Wang Y, Xia Y. Experimental and theoretical study on the strain rate and
temperature dependence of mechanical behavior of Kevlar ber. Compos PartA 1999;30:1251e7.
[22] Lim J, Zheng JQ, Masters K, Chen WW. Effects of gage length, loading rates, anddamage on the strength of PPTA bers. Int J Impact Eng 2011;38(4):219e27.
[23] Zhu D, Mobasher B, Rajan SD. Dynamic tensile testing of Kevlar 49 fabrics. J Mater Civil Eng 2011;23(3):230e9.
[24] Rodriguez J, Chocron IS, Martinez MA, Sanchez-Galvez V. High strain rateproperties of aramid and polyethylene woven fabric composites. Compos PartB Eng 1996;27(2):147e54.
[25] Viswanathan B, Vinson JR, Scott BR. High strain rate compression testing of Kevlar 29/polyethylene composite with very high ber volume fraction. In:AIAA/ASME/ASCE/AHS/ASC structures, structural dynamics, and materials
conference and exhibit, 39th and AIAA/ASME/AHS adaptive structures forum;1998. Long Beach, CA.
[26] Carrillo JG, Gamboa RA, Flores-Johnson EA, Gonzalez-Chi PI. Ballistic perfor-mance of thermoplastic composite laminates made from aramid woven fabricand polypropylene matrix. Polym Test 2012;31(4):512e9.
[27] Gorham DC. Specimen inertia in high strain-rate compression. J Phys 1989;22:1888e93.
[28] Ramesh KT. High strain rate and impact experiment. Springer handbook of experimental solid mechanics. New York: Springer; 2008. p. 910e50.
[29] Kolsky H. An investigation of the mechanical studies in plastic wave propa-
gation. J Mech Phys Solids 1949;10:195e
223.[30] Ruiz C, Mines RAW. The Hopkinson pressure bar: an alternative to the
instrumented pendulum for Charpy tests. Int J Fract 1985;29:101e9.[31] Meng H, Li QM. Correlation between the accuracy of a SHPB test and the stress
uniformity basedon numerical experiments.Int J Impact Eng 2003;28:537e55.[32] Frew DJ, Forrestal MJ, Chen W. Pulse shaping techniques for testing elastic
plastic materials with a split Hopkinson pressure bar. Exp Mech 2005;45:186e95.
[33] Lu YB, Li QM. Dynamic behavior of polymers at high strain-rates based on splitHopkinson pressure bar tests. Int J Impact Eng 2011;38:41e50.
[34] Bahlouli N, R emond Y, Matadi R, Wang K, Ahzi S. Dynamic behavior of virginand recycled PP/EPR and PP/EPDM/TALC materials for car bumpers. In: 18thInternational conference on composite materials (ICCM), Korea. 18 2011.p. 3e6.
R. Kapoor et al. / Composites Part B 89 (2016) 374e 382382
http://refhub.elsevier.com/S1359-8368(15)00758-1/sref17http://refhub.elsevier.com/S1359-8368(15)00758-1/sref17http://refhub.elsevier.com/S1359-8368(15)00758-1/sref17http://refhub.elsevier.com/S1359-8368(15)00758-1/sref17http://refhub.elsevier.com/S1359-8368(15)00758-1/sref17http://refhub.elsevier.com/S1359-8368(15)00758-1/sref17http://refhub.elsevier.com/S1359-8368(15)00758-1/sref18http://refhub.elsevier.com/S1359-8368(15)00758-1/sref18http://refhub.elsevier.com/S1359-8368(15)00758-1/sref18http://refhub.elsevier.com/S1359-8368(15)00758-1/sref18http://refhub.elsevier.com/S1359-8368(15)00758-1/sref18http://refhub.elsevier.com/S1359-8368(15)00758-1/sref19http://refhub.elsevier.com/S1359-8368(15)00758-1/sref19http://refhub.elsevier.com/S1359-8368(15)00758-1/sref19http://refhub.elsevier.com/S1359-8368(15)00758-1/sref19http://refhub.elsevier.com/S1359-8368(15)00758-1/sref19http://refhub.elsevier.com/S1359-8368(15)00758-1/sref20http://refhub.elsevier.com/S1359-8368(15)00758-1/sref20http://refhub.elsevier.com/S1359-8368(15)00758-1/sref20http://refhub.elsevier.com/S1359-8368(15)00758-1/sref20http://refhub.elsevier.com/S1359-8368(15)00758-1/sref21http://refhub.elsevier.com/S1359-8368(15)00758-1/sref21http://refhub.elsevier.com/S1359-8368(15)00758-1/sref21http://refhub.elsevier.com/S1359-8368(15)00758-1/sref21http://refhub.elsevier.com/S1359-8368(15)00758-1/sref21http://refhub.elsevier.com/S1359-8368(15)00758-1/sref21http://refhub.elsevier.com/S1359-8368(15)00758-1/sref22http://refhub.elsevier.com/S1359-8368(15)00758-1/sref22http://refhub.elsevier.com/S1359-8368(15)00758-1/sref22http://refhub.elsevier.com/S1359-8368(15)00758-1/sref22http://refhub.elsevier.com/S1359-8368(15)00758-1/sref22http://refhub.elsevier.com/S1359-8368(15)00758-1/sref23http://refhub.elsevier.com/S1359-8368(15)00758-1/sref23http://refhub.elsevier.com/S1359-8368(15)00758-1/sref23http://refhub.elsevier.com/S1359-8368(15)00758-1/sref24http://refhub.elsevier.com/S1359-8368(15)00758-1/sref24http://refhub.elsevier.com/S1359-8368(15)00758-1/sref24http://refhub.elsevier.com/S1359-8368(15)00758-1/sref24http://refhub.elsevier.com/S1359-8368(15)00758-1/sref24http://refhub.elsevier.com/S1359-8368(15)00758-1/sref24http://refhub.elsevier.com/S1359-8368(15)00758-1/sref25http://refhub.elsevier.com/S1359-8368(15)00758-1/sref25http://refhub.elsevier.com/S1359-8368(15)00758-1/sref25http://refhub.elsevier.com/S1359-8368(15)00758-1/sref25http://refhub.elsevier.com/S1359-8368(15)00758-1/sref25http://refhub.elsevier.com/S1359-8368(15)00758-1/sref25http://refhub.elsevier.com/S1359-8368(15)00758-1/sref25http://refhub.elsevier.com/S1359-8368(15)00758-1/sref26http://refhub.elsevier.com/S1359-8368(15)00758-1/sref26http://refhub.elsevier.com/S1359-8368(15)00758-1/sref26http://refhub.elsevier.com/S1359-8368(15)00758-1/sref26http://refhub.elsevier.com/S1359-8368(15)00758-1/sref27http://refhub.elsevier.com/S1359-8368(15)00758-1/sref27http://refhub.elsevier.com/S1359-8368(15)00758-1/sref27http://refhub.elsevier.com/S1359-8368(15)00758-1/sref28http://refhub.elsevier.com/S1359-8368(15)00758-1/sref28http://refhub.elsevier.com/S1359-8368(15)00758-1/sref28http://refhub.elsevier.com/S1359-8368(15)00758-1/sref29http://refhub.elsevier.com/S1359-8368(15)00758-1/sref29http://refhub.elsevier.com/S1359-8368(15)00758-1/sref29http://refhub.elsevier.com/S1359-8368(15)00758-1/sref30http://refhub.elsevier.com/S1359-8368(15)00758-1/sref30http://refhub.elsevier.com/S1359-8368(15)00758-1/sref30http://refhub.elsevier.com/S1359-8368(15)00758-1/sref31http://refhub.elsevier.com/S1359-8368(15)00758-1/sref31http://refhub.elsevier.com/S1359-8368(15)00758-1/sref31http://refhub.elsevier.com/S1359-8368(15)00758-1/sref32http://refhub.elsevier.com/S1359-8368(15)00758-1/sref32http://refhub.elsevier.com/S1359-8368(15)00758-1/sref32http://refhub.elsevier.com/S1359-8368(15)00758-1/sref32http://refhub.elsevier.com/S1359-8368(15)00758-1/sref32http://refhub.elsevier.com/S1359-8368(15)00758-1/sref33http://refhub.elsevier.com/S1359-8368(15)00758-1/sref33http://refhub.elsevier.com/S1359-8368(15)00758-1/sref33http://refhub.elsevier.com/S1359-8368(15)00758-1/sref33http://refhub.elsevier.com/S1359-8368(15)00758-1/sref34http://refhub.elsevier.com/S1359-8368(15)00758-1/sref34http://refhub.elsevier.com/S1359-8368(15)00758-1/sref34http://refhub.elsevier.com/S1359-8368(15)00758-1/sref34http://refhub.elsevier.com/S1359-8368(15)00758-1/sref34http://refhub.elsevier.com/S1359-8368(15)00758-1/sref34http://refhub.elsevier.com/S1359-8368(15)00758-1/sref34http://refhub.elsevier.com/S1359-8368(15)00758-1/sref34http://refhub.elsevier.com/S1359-8368(15)00758-1/sref34http://refhub.elsevier.com/S1359-8368(15)00758-1/sref34http://refhub.elsevier.com/S1359-8368(15)00758-1/sref34http://refhub.elsevier.com/S1359-8368(15)00758-1/sref34http://refhub.elsevier.com/S1359-8368(15)00758-1/sref33http://refhub.elsevier.com/S1359-8368(15)00758-1/sref33http://refhub.elsevier.com/S1359-8368(15)00758-1/sref33http://refhub.elsevier.com/S1359-8368(15)00758-1/sref32http://refhub.elsevier.com/S1359-8368(15)00758-1/sref32http://refhub.elsevier.com/S1359-8368(15)00758-1/sref32http://refhub.elsevier.com/S1359-8368(15)00758-1/sref32http://refhub.elsevier.com/S1359-8368(15)00758-1/sref31http://refhub.elsevier.com/S1359-8368(15)00758-1/sref31http://refhub.elsevier.com/S1359-8368(15)00758-1/sref31http://refhub.elsevier.com/S1359-8368(15)00758-1/sref30http://refhub.elsevier.com/S1359-8368(15)00758-1/sref30http://refhub.elsevier.com/S1359-8368(15)00758-1/sref30http://refhub.elsevier.com/S1359-8368(15)00758-1/sref29http://refhub.elsevier.com/S1359-8368(15)00758-1/sref29http://refhub.elsevier.com/S1359-8368(15)00758-1/sref29http://refhub.elsevier.com/S1359-8368(15)00758-1/sref28http://refhub.elsevier.com/S1359-8368(15)00758-1/sref28http://refhub.elsevier.com/S1359-8368(15)00758-1/sref28http://refhub.elsevier.com/S1359-8368(15)00758-1/sref27http://refhub.elsevier.com/S1359-8368(15)00758-1/sref27http://refhub.elsevier.com/S1359-8368(15)00758-1/sref27http://refhub.elsevier.com/S1359-8368(15)00758-1/sref26http://refhub.elsevier.com/S1359-8368(15)00758-1/sref26http://refhub.elsevier.com/S1359-8368(15)00758-1/sref26http://refhub.elsevier.com/S1359-8368(15)00758-1/sref26http://refhub.elsevier.com/S1359-8368(15)00758-1/sref25http://refhub.elsevier.com/S1359-8368(15)00758-1/sref25http://refhub.elsevier.com/S1359-8368(15)00758-1/sref25http://refhub.elsevier.com/S1359-8368(15)00758-1/sref25http://refhub.elsevier.com/S1359-8368(15)00758-1/sref25http://refhub.elsevier.com/S1359-8368(15)00758-1/sref24http://refhub.elsevier.com/S1359-8368(15)00758-1/sref24http://refhub.elsevier.com/S1359-8368(15)00758-1/sref24http://refhub.elsevier.com/S1359-8368(15)00758-1/sref24http://refhub.elsevier.com/S1359-8368(15)00758-1/sref24http://refhub.elsevier.com/S1359-8368(15)00758-1/sref24http://refhub.elsevier.com/S1359-8368(15)00758-1/sref23http://refhub.elsevier.com/S1359-8368(15)00758-1/sref23http://refhub.elsevier.com/S1359-8368(15)00758-1/sref23http://refhub.elsevier.com/S1359-8368(15)00758-1/sref22http://refhub.elsevier.com/S1359-8368(15)00758-1/sref22http://refhub.elsevier.com/S1359-8368(15)00758-1/sref22http://refhub.elsevier.com/S1359-8368(15)00758-1/sref21http://refhub.elsevier.com/S1359-8368(15)00758-1/sref21http://refhub.elsevier.com/S1359-8368(15)00758-1/sref21http://refhub.elsevier.com/S1359-8368(15)00758-1/sref21http://refhub.elsevier.com/S1359-8368(15)00758-1/sref20http://refhub.elsevier.com/S1359-8368(15)00758-1/sref20http://refhub.elsevier.com/S1359-8368(15)00758-1/sref20http://refhub.elsevier.com/S1359-8368(15)00758-1/sref20http://refhub.elsevier.com/S1359-8368(15)00758-1/sref19http://refhub.elsevier.com/S1359-8368(15)00758-1/sref19http://refhub.elsevier.com/S1359-8368(15)00758-1/sref19http://refhub.elsevier.com/S1359-8368(15)00758-1/sref19http://refhub.elsevier.com/S1359-8368(15)00758-1/sref18http://refhub.elsevier.com/S1359-8368(15)00758-1/sref18http://refhub.elsevier.com/S1359-8368(15)00758-1/sref18http://refhub.elsevier.com/S1359-8368(15)00758-1/sref18http://refhub.elsevier.com/S1359-8368(15)00758-1/sref18http://refhub.elsevier.com/S1359-8368(15)00758-1/sref17http://refhub.elsevier.com/S1359-8368(15)00758-1/sref17http://refhub.elsevier.com/S1359-8368(15)00758-1/sref17