Vous êtes sur la page 1sur 10

Original article

Journal of Reinforced Plastics


and Composites

Elastic anisotropy of kenaf fibre and 0(0) 1–10


! The Author(s) 2016
Reprints and permissions:
micromechanical modeling of nonwoven sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/0731684416652740
kenaf fibre/epoxy composites jrp.sagepub.com

NG Andre1, Dody Ariawan1,2 and ZA Mohd Ishak1,3

Abstract
In this work, nonwoven kenaf fibre/epoxy composites were produced by using resin transfer moulding. The effect of
kenaf fibre volume fraction on the composites’ tensile properties and Poisson’s ratio was investigated. Experimental
results show that highest tensile properties and Poisson’s ratio were attained at volume fraction ¼ 0.42. A simple method
has been developed to predict the fibre transverse modulus and has allowed the characterisation of kenaf fibre’s elastic
anisotropy. The performance of the Tsai–Pagano model in predicting the composites’ tensile modulus and Poisson’s ratio
was compared with the Manera and Cox-Krenchel model. Results showed that the consideration of fibre’s elastic
anisotropy in the Tsai–Pagano model yielded a good prediction of both composites’ modulus and Poisson’s ratio.
Meanwhile, the Bowyer–Bader model produced a better tensile strength prediction owing to the inclusion of fibre
length and orientation factors in the model.

Keywords
Polymer composites, micromechanical modeling, elastic anisotropy, Kenaf, natural fibre

fibre/epoxy composites. It is important to note that


Introduction the intrinsic properties of fibres are key factors in pro-
The current upsurge in the use of natural fibres in ducing an accurate prediction of the composites’ tensile
various applications has led to the utilisation of green properties.
nonwovens for interior automotive components.1 This Owing to the complex structure of most natural
is because it is lightweight and strong with sound effi- fibres, elastic anisotropy should be expected in which
ciency, flexibility, and versatility, as well as having an the fibres’ longitudinal properties are greater than their
attractive cost to performance ratio.2 Nonwovens also transverse properties.5,6 Unfortunately, many studies
provide excellent z-directional properties that reduce fail to predict the stiffness of randomly oriented natural
delamination issues.1,2 Resin transfer moulding fibre composites accurately due to the assumption that
(RTM) is a suitable processing method used to produce the fibres were isotropic.7–9 Therefore, to produce an
nonwoven composites as this method retains the shapes accurate model, the anisotropy of fibre properties
and dimensions of the nonwoven mat. Furthermore,
the laminates produced by RTM have superior mech- 1
School of Materials and Mineral Resources Engineering, Universiti Sains
anical properties compared to those produced via com-
Malaysia, Nibong Tebal, Malaysia
pression moulding, due to hollow cylinders of needle 2
Faculty of Engineering, Universitas Sebelas Maret, Surakarta, Indonesia
path that ease the penetration of matrix into the non- 3
Cluster for Polymer Composites, Science and Engineering Research
woven which consequently led to lower void content.3,4 Centre, Universiti Sains Malaysia, Nibong Tebal, Malaysia
To avoid costly and time-consuming experiments,
prediction of mechanical properties such as tensile Corresponding author:
ZA Mohd Ishak, Cluster for Polymer Composites, Science and
properties can be made through established mathemat- Engineering Research Centre, Universiti Sains Malaysia, Engineering
ical models. So far, there has been limited work on the Campus, 14300 Nibong Tebal, Pulau Pinang, Malaysia.
prediction of tensile properties of nonwoven natural Email: zarifin@usm.my

Downloaded from jrp.sagepub.com at TEXAS SOUTHERN UNIVERSITY on June 6, 2016


2 Journal of Reinforced Plastics and Composites 0(0)

should be taken into consideration. The Tsai–Pagano Experimental


model takes into account the anisotropy of fibres in
which both fibre’s longitudinal modulus (Ef1) and
Materials
transverse modulus (Ef2) are required in order to pre- Needle-punched nonwoven kenaf fibres (KF) with a
dict the composites’ tensile modulus.10 The model stitching density of 50 cm2 and area density of
basically treats a randomly oriented fibre composite 1100 g/m2 were supplied by Kenaf Natural Fibre
as a quasi-isotropic laminate. Industries. The average density of a single KF extracted
While it is easy to measure the Ef1 directly, charac- from the nonwoven mat is 1.4 g/cm3. Bisphenol A
terizing the Ef2 has proven to be challenging. Mounting epoxy resin CP 210DF part A was used with an
and testing fibres in any direction other than the longi- amine hardener CP 210DF part B supplied by Camel
tudinal is difficult. Nevertheless, Cichocki Jr. and Polymer Sdn. Bhd. The mixing ratio of the epoxy-hard-
Thomason11 reported that it is possible to predict the ener was 2:1, respectively. The epoxy resin system has a
Ef2 of jute with the help of a number of micromecha- blend viscosity and density of 1.61 Pa.s and 1.13 g/cm3,
nical models. However, the scope of the aforemen- respectively.
tioned study was limited to the thermoelastic
anisotropy of the fibre. As such, the obtained modulus
was not incorporated in any models to predict the
Determination of fibre length and diameter
mechanical behaviour of composites. The fibre’s length distribution was characterised
Tensile strength is an essential element that defines according to ISO 6989 Method A. The characterisation
the mechanical properties of materials. It is important used 500 individual fibres extracted from the nonwoven
to note that reinforcing the fibre and its geometric non- mat for length distribution measurement. The fibre’s
homogeneity will have a greater impact on its strength diameter distribution was measured under an
than on the composites’ stiffness. Furthermore, mater- Olympus BX61 optical microscope. The diameters of
ial failure usually starts at the stress and strain concen- 200 individual fibres were recorded.
tration points, amplifying this effect on the fibre’s
strength.12
In addition to tensile strength and modulus,
Fibre tensile properties
Poisson’s ratio is another important element that has The tensile strength,  f, and the longitudinal tensile
been overlooked in many studies on the mechanical modulus, Ef1, of single KF were evaluated directly fol-
behaviour of composites. Predictions of strength and lowing the ASTM C 1557 standard by using the mini-
Poisson’s ratio are equally important, but there have ature tensile tester model Lex810 Dia-stron (UK)
been few micromechanical studies directed towards equipped with a load cell of 19.6 N. The displacement
nonwoven natural fibre composites. Hence, there is a speed of the upper clamp was set to 3 mm/min. The  f
need to find a good micromechanical model to repre- and Ef1 of 20 single KF extracted from the nonwoven
sent both the tensile properties and Poisson’s ratio of mat was measured. Prior to the test, the fibres were
nonwoven composites. dried in a circulating oven at 60 C for 4 h.
The objective of this research is to evaluate the Since the Ef2 is difficult to measure, a simplified
performance and applicability of several models in pre- method based on the method presented by Cichocki
dicting the elastic constants (tensile modulus and and Thomason11 was used in which composites consist-
Poisson’s ratio) and tensile strength of nonwoven ing of unidirectional KF in an epoxy matrix (UKFE
KF/epoxy (NKFE) composites over a range of fibre composites) were produced as shown in Figure 1. Single
volume fraction (Vf). In order to yield better elastic KF from the nonwoven mat were first wrapped around
constant predictions, KF elastic anisotropy was taken a thin metal bar with width and length of 25 and
into account in the Tsai–Pagano model. The Tsai– 50 mm, respectively. They were then secured into
Hahn model13 was used to predict the fibre transverse place with a cellophane tape. Furthermore, they were
modulus, Ef2 and that was, in turn, used in the calcu- carefully and uniformly wrapped to a thickness of
lation of composites’ tensile modulus and Poisson’s approximately 2 mm to prevent variations in fibre con-
ratio. The estimation of elastic constants by the Tsai– centration and orientation.
Pagano model was then compared with experimental The fibre-metal bar assemblage was then submerged
results and prediction by the Manera14 and Cox- into a 400-ml beaker filled with freshly mixed epoxy
Krenchel15,16 models. Lastly, predictions of tensile and hardener. The beaker containing the epoxy-hard-
strength were made using models by Hirsch17 and ener mixture and fibre-metal bar assemblage was
Bowyer–Bader,18 and the estimates produced by each quickly placed in a vacuum chamber calibrated to
model were compared to experimental results and 0.1 atm. The vacuum was then released to assist in the
discussed. infiltration of the KF and removal of air bubbles from

Downloaded from jrp.sagepub.com at TEXAS SOUTHERN UNIVERSITY on June 6, 2016


Andre et al. 3

Figure 1. Schematic representation of the UKFE composite fabrication used to determine the Ef2 of KF.

the epoxy-hardener mixture. It was repeated until no


more air bubbles could be observed in the epoxy
mixture.
After approximately 15 min, the assemblage was
removed from the beaker. Another metal bar was
clamped and compressed against the assemblage and
it was allowed to cure for 2 h at 80 C. The cured
UKFE composite was then separated from the metal
bar and cut, perpendicular to the fibre direction, into Figure 2. Strain gage placement on the tensile test specimen.
rectangular shapes, measuring approximately
20  4  1 mm3. Post-curing was carried out after
machining for another 4 h at 60 C to remove excess
moisture and allow for the epoxy to fully cure. A KF mat. The composites then underwent post-curing at
Densitometry Model Precisa XT 220 A was used to 60 C for 24 h. They were finally cut into tensile and
determine the densities of the UKFE composite sam- flexural test samples by using a bandsaw.
ples. These data were then used to estimate the volume
fraction of the fibres in the composites.
Tensile properties of epoxy and NKFE composites
A Mettler Toledo DMA 861 dynamic mechanical
analyser was utilised to measure the elastic constants The tensile strength and modulus of the epoxy and
of samples mounted with a gauge length of 15 mm NKFE composites were evaluated according to the
over a range of temperatures. Dynamic tensile dis- ASTM D 3039 standard by using the INSTRON
placement was superimposed sinusoidally with a dis- 5969 universal testing machine equipped with an
placement amplitude of 3 mm at a frequency of 10 Hz extensometer. The test machine was also equipped
under tensile mode. After reaching 50 C, the tem- with a 50 kN load cell and operated at a crosshead
perature was raised to 100 C at 2 C/min intervals. speed of 2 mm/min. Each of the specimens was also
The Tsai–Hahn model14 was then applied to calculate attached with a triaxial strain gage supplied by Showa
Ef2 of the KF. Measuring Instruments Co. Ltd. The strain gage was
carefully attached in the middle of the cleaned surface
of the specimen by using a cyanoacrylate-type adhesive
Composite preparation (see Figure 2). The gage length was 5 mm with a resist-
The NKFE composites were produced using a Hypaject ance of 120
. The UCAM-550 A series Fast Data
MKV-type resin transfer moulding machine with two Logger by Kyowa Electronic Instruments was used to
reciprocating pumps. The mixing ratio of epoxy to measure the x, y, and z-axis strain during the tensile
hardener was 2 to 1. The mixtures were shot into a test. The values were, in turn, used to calculate the
mould containing 0.15, 0.31, and 0.42 Vf of nonwoven Poisson’s ratio of epoxy and its composites.

Downloaded from jrp.sagepub.com at TEXAS SOUTHERN UNIVERSITY on June 6, 2016


4 Journal of Reinforced Plastics and Composites 0(0)

Void content incorporated the properties of the composite and


The void content, VV, was calculated from the equation matrix were used. In this study, a semi-empirical
below: model developed by Tsai and Hahn13 was employed
to help with the prediction of Ef2. This model was
theoretical  experimental developed under the assumption that the transverse
VV ¼ ð1Þ
theoretical stress in the matrix was less than the stress carried by
the fibre. The formulation is defined as
where  
1 Vf Vm
E22 ¼ þ ð3Þ
theoretical ¼ Vf f þ Vm m ð2Þ Vf Vm Ef2 Em

The symbol theoretical and experimental represent the where  denotes the stress partitioning factor which is
theoretical and experimental density of the NKFE com- obtained by curve-fitting the equation to the elastic
posites while f and m denote the density of fibre and modulus of composite, E22, at a range of fibre volume
matrix. Vf and Vm represent the volume fraction of fibre fraction, Vf. The value of  ¼ 0.5 was employed in this
and matrix, respectively. study following an acceptable curve fit of the predicted
transverse and shear moduli data to the experimental
data for a unidirectional glass fibre/epoxy composite.13
Surface morphology The fibre concentration was determined to be
The surface morphology of the NKFE composites was 0.3  0.02 Vf, assuming zero void contents in the com-
observed using SEM Model Carl Zeiss Leo Supra posites. Figure 3 shows the elastic modulus of UKFE
50 VP. All samples were first coated with a thin layer composites perpendicular (90 ) from the fibre direction
of gold by using VG Microtech–Polaron Sputter from 10 C to 50 C, which were obtained by the
Coater to render them electrically conductive. dynamic mechanical analysis method. Gradual decre-
ments of the composite’s elastic modulus can be
observed as the temperature increased.
Results and discussion Figure 3 also shows the predicted Ef2 calculated using
equation (3). Results obtained from the DMA test and
Properties of single kenaf fibre
the prediction by Tsai–Hahn model provides some insight
Table 1 summarises the physical and tensile properties into the relationship between elastic modulus of fibre and
of single KF extracted from the nonwoven kenaf mat. temperature. It can be seen that the calculated transverse
The average longitudinal tensile strength of KF was modulus of the fibre, Ef2, decreased as the temperature
514.59  141.06 MPa; it was in the range of previous rose. Due to the hygroscopic nature of natural fibre, it
reported studies.19,20 may contain some residual water which is known to
The anisotropic properties of KF can be proven by increase its stiffness.22 At elevated temperature, the
comparing the fibre’s longitudinal Young’s modulus, water moisture is gradually removed which in turn,
Ef1, with its transverse Young’s modulus, Ef2. The dif-
ference between these two moduli will indicate the
extent of the anisotropic behaviour of the fibre.
The average Ef1 obtained from the miniature tensile
test was 26.06  6.63 GPa. The obtained Ef1 value was
within the range of experimental values presented in the
literature.19–21
In order to evaluate the transverse Young’s modu-
lus, Ef2 of KF, micromechanical models that

Table 1. Physical and tensile properties of single KF.

Mean length (mm) 40.10  17.40


Mean diameter (mm) 63.79  37.37
Aspect ratio 628.63  151.10
Longitudinal tensile modulus (GPa) 26.06  6.63 Figure 3. The elastic modulus of UKFE composite perpen-
Longitudinal tensile strength (MPa) 514.59  141.06 dicular to the fibre direction (EC) and the predicted KF transverse
modulus by Tsai–Hahn model.

Downloaded from jrp.sagepub.com at TEXAS SOUTHERN UNIVERSITY on June 6, 2016


Andre et al. 5

Table 2. Experimental tensile properties, Poisson’s ratio, and void content of epoxy and NKFE composites.

Fibre volume fraction, Vf Tensile modulus (GPa) Tensile strength (MPa) Poisson’s ratio Void content (%)
28
0 1.93 (0.12) 43.18 (1.49) 0.39 (0.041) –
0.15 3.21 (0.16) 31.86 (1.16) 0.39 (0.004) 2.08 (0.33)
0.31 3.97 (0.10) 38.45 (1.38) 0.41 (0.004) 3.14 (0.44)
0.42 5.50 (0.27) 50.17 (0.82) 0.42 (0.018) 5.17 (0.30)
( ) ¼ Standard deviation.

decreased the fibre’s elastic modulus.23 Nevertheless, at


Prediction of composites’ tensile modulus
room temperature, the predicted transverse modulus of
the single KF was 2.50 GPa. This also shows that single
Various micromechanical models have been developed
KF was highly anisotropic due to the large difference
and proposed to predict the modulus of fibre-reinforced
between Ef1 and Ef2.
polymer composites. Nonwovens are basically made up
of a loose matting structure which consists of fibre
Effect of fibre loading mechanical interlocking and hollow cylinders produced
by the needle punching process. However, assuming the
Table 2 presents the obtained experimental variation of
nonwoven to be a randomly oriented fibre composite,
tensile properties of epoxy and NKFE composites at
the established micromechanical models might still be
different fibre volume fractions, Vf. The rigidity of the
useful to predict and validate experimental data. In
composite materials increased with the increment of the
order to predict the tensile modulus of the NKFE com-
nonwoven KF loading. The tensile modulus increased
posites, models by Tsai–Pagano, Manera, and Cox-
by 66% with the addition of 0.15 Vf of nonwoven mats.
Krenchel have been used.
Further addition of nonwoven mats, 0.31 and 0.42 Vf,
increased the tensile modulus by 106% and 185%, Before any comparison can be done, it is vital to
respectively. The addition of stiffer fibres that exhibited understand the motivation behind the selection of
higher modulus than the polymer matrix inhibited the each model in this study. The Tsai–Pagano model was
movements of the polymer chain. This, in turn, selected as it took into account the fibre’s elastic anisot-
improved the stiffness of the resulting composites. ropy. As for the model by Manera, it is basically based
The tensile strength reduced by 26% upon incorpor- on the classical laminate analogy and shares the same
ation of 0.15 Vf. At low fibre loading, the amount of invariants properties with the Tsai–Pagano model.14
fibre is not sufficient in providing high strength to the Therefore, a comparison between these two models is
epoxy matrix, and thus the nonwoven KF act as relevant to observe the performance of the model that
flaws.24,25 Nevertheless, as Vf increased, the properties includes or does not include the fibre’s elastic anisot-
of the KF became more significant, and consequently ropy. On the other hand, the Cox-Krenchel model was
the tensile strength of the composite was improved. The based on the modification of the rule of mixture,
highest tensile strength was attained at 0.42 Vf with whereby the dimension and orientation factor of the
16% improvement from the tensile strength of neat fibre phase was included. The relevance behind the
epoxy. Meanwhile, the Poisson’s ratio exhibited small selection of this model was to compare a model
changes upon incorporation of nonwovens, wherein the that was based on the laminate and invariants theory
highest value was attained at 0.42 Vf. (Tsai–Pagano) and a model that was based on other
Table 2 also shows the void content of the NKFE consideration (Cox-Krenchel). Apart from that, it has
composites at different Vf. The void content shows an been proven previously that the model by Cox-
increasing trend against Vf. The highest void content Krenchel has produced good tensile modulus
was 5.2%, observed at 0.42 Vf. There are several rea- prediction.29–32
sons why voids are formed in the composites. First is
The model proposed by Tsai and Pagano10 was used
the inability of the matrix to displace all the air which is
extensively to predict the elastic modulus of composites
trapped within the nonwovens during fibre impregna-
consisting of fibres randomly oriented in a plane. The
tion26; this happened when the resin was shot into the
equation is as follows
mould containing the nonwoven KF. Second is the
incomplete wetting of fibres by the resin due to the
poor adhesion between the hydrophobic matrix and 3 5
EC ¼ E11 þ E22 ð4Þ
hydrophilic fibres.27 8 8

Downloaded from jrp.sagepub.com at TEXAS SOUTHERN UNIVERSITY on June 6, 2016


6 Journal of Reinforced Plastics and Composites 0(0)

where the E11 and E22 are the respective longitudinal with the properties’ laminates with infinite number of
and transverse moduli for a unidirectional aligned short layers oriented in every direction.
fibre composite having similar aspect ratio and fibre The Cox-Krenchel model utilised the modified rule
volume fraction, Vf. of mixture and is expressed as follows
In conjunction with the Tsai–Pagano model, the
equation proposed by Halpin-Tsai33 was used to deter- EC ¼ l o Ef Vf þ Em Vm ð8Þ
mine both E11 and E22. The mathematical expressions
are as follows where l is the fibre length distribution factor and o is
 the fibre orientation distribution factor; the effect of
    Ef1 voids has been neglected.15 The fibre orientation
1 þ 2ld 1 Vf Em  1
E11 ¼ Em where 1 ¼  ð5Þ factor was based on Krenchel’s16
1  1 V f Ef1
þ 2lEm d
o ¼ cos4 ðo Þ ð9Þ

  1
Ef2
1 þ 22 Vf Em where o is the fibre orientation limit angle. Proposed
E22 ¼ Em where 2 ¼  ð6Þ
1  2 V f E f2
þ2 by Thomason and Vlug, the orientation factor, o, for
Em
laminates containing an in-plane random fibre orienta-
tion is 0.375.35 On the other hand, the l was calculated
The combination of these models took into consid- based on the Cox shear lag theory, which is given by15:
eration the fibre loading and both of the fibres’ elastic
l
modulus: longitudinal and transversal. They have, tanh 2
therefore, been utilised in this study to predict the l ¼ 1  l
ð10Þ
2
NKFE composites’ modulus under variations of non-
woven KF loading. vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
The physical properties as shown in Table 1, as well 2u
u Em
¼ t  ð11Þ
as the estimated fibre transverse modulus, were applied d E ð1   ÞLn 
f1 M i Vf
into the Halpin-Tsai and then Tsai–Pagano equations
to predict the elastic modulus of the NKFE composites.
Table 3 summarises the E11 and E22 under different where the shear-parameter  represents the coefficient
fibre loading calculated by using equations (5) and (6). of stress concentration rate at fibres’ end, vM is the
Manera has proposed a simplified equation to pre- Poisson’s ratio of matrix, and l is the length of fibre.
dict the elastic properties of randomly oriented short This study assumes a square fibre packing, and thus
fibre composites.14 The proposed equation by Manera i ¼ 4.
is as follows Figure 4 shows the experimental and predicted
  tensile modulus of NKFE composites over a range of
16 8 Vf. Predictions of tensile modulus by all three models
EC ¼ Vf Ef1 þ 2Em þ Em ð7Þ
45 9

The model is a result of the usage of the classical


laminate analogy alongside the invariant properties of
composite defined by Tsai–Pagano, and micromechanic
formulation proposed by Puck.34 The model also
assumes that the mechanical properties of randomly
oriented discontinuous fibre composites are the same

Table 3. E11 and E22 at different Vf .

Fibre volume fraction, Vf

Elastic modulus 0.15 0.31 0.42

E11 (GPa) 5.26 8.92 11.51


E22 (GPa) 2.00 2.09 2.15 Figure 4. Experimental and predicted tensile modulus of NKFE
composites at different Vf.

Downloaded from jrp.sagepub.com at TEXAS SOUTHERN UNIVERSITY on June 6, 2016


Andre et al. 7

displayed a linear increment as a function of Vf.


Comparing the theoretical and experimental values, it
can be seen that the model by Manera has failed in
predicting the tensile modulus accurately. There is a
significant overestimation by the model at all Vf. The
model deviates 20%, 45% and 31% at 0.15, 0.31, and
0.42 Vf, respectively. Based on previous study, the
model only works well when Em is in the range of 2
to 4 GPa and the fibre has a high aspect ratio: l/d > 300.
Both epoxy and nonwoven KF did not fulfil both
requirements by Manera.14 Furthermore, the model
that was meant for glass fibre disregards the elastic
anisotropy of fibre in the derivation. Unlike glass
fibre, natural fibre has been proven to have large vari- Figure 5. SEM image of NKFE composites (voids are indicated
ations in its properties due to varied (a) content of cel- by the arrows).
lulose and (b) fibre’s dimensions.11,36 Moreover, in the
previous part of this paper, the fibre that was utilised in
this study has proven to be elastically anisotropic. All
of these have influenced the ability of the model to pro-
Prediction of Poisson’s ratio
duce an accurate prediction of tensile modulus. To predict the Poisson’s ratio vC of the NKFE compos-
Cox-Krenchel’s model has produced a very good ites by using the Tsai–Pagano model, the following
prediction at each Vf. It falls under the experimental equation was used with the assumptions that the com-
tensile modulus with very small deviations: <9%. posites have a random distribution of fibres10,33:
This is due to the inclusion of both fibre orientation
and length fibre in the model, which are both EC
C ¼ 1 ð12Þ
important factors influencing the tensile properties 2GC
of composites. A previous study stated that the
improvement of prediction models could probably in which the EC was obtained from the previous pre-
be reached if factors such as fibre angle and length diction; the shear modulus, GC, was estimated by
distribution corrections were included.9 This is a another equation proposed by Tsai–Pagano.10,33 It is
noteworthy finding as the Cox-Krenchel’s model expressed as:
involves simple calculations and the required data
are obtainable from a simple test. 1 1
GC ¼ E11 þ E22 ð13Þ
For the Tsai–Pagano model, at 0.15 Vf, the predic- 8 4
tion deviated 5.6% from the experimental result. At
0.31 and 0.42 Vf, the deviations were 14.6% and where again, the E11 and E22 are longitudinal and trans-
2.8% above the experimental values. The model that verse elastic moduli of unidirectional aligned fibre com-
was used in conjunction with the Halpin-Tsai model posite calculated from equations (3) and (4). The
has yielded better tensile modulus prediction compared predicted GC of the composites was 1.19, 1.69, and
to the prediction by Manera. This proves that taking 2.03 GPa at 0.15, 0.31, and 0.42 Vf, respectively.
into consideration the fibre’s elastic anisotropy gives Meanwhile, both models by Cox-Krenchel and
out better prediction. However, comparison between Manera put constraints on the Poisson’s ratio and
the model and the Cox-Krenchel model reveals that were set to be vC ¼ 1/3 at every Vf.14,37
the latter predicted tensile modulus more accurately. Figure 6 shows the experimental and predicted
The previous study shows that fibre stiffness was deter- Poisson’s ratio, vC. It can be observed that the experi-
mined by spiral angles of the crystalline fibrils and the mental vC lies between the models by Tsai–Pagano and
content of cellulose in the fibre.36 The inconsistencies of Manera/Cox-Krenchel but closer to the former.
these two in the KF and the presence of void in the Moreover, based on equation (12) by Tsai–Pagano,
composites as shown in Table 2 and Figure 5 might the calculated vC at 0.15, 0.31, and 0.42 Vf were
be responsible for the small discrepancies in the predic- 0.395, 0.423, and 0.438. Likewise, at Vf of 0.15, the
tion as it is not included in the model. Nonetheless, the predicted vC deviates by 2.3% above the experimental
differences between the deviations of Tsai–Pagano and value. In addition, the predictions of vC further deviate
Cox-Krenchel models from the experimental values are as Vf increased, i.e. by 3.2% and 4.3% at 0.31 and 0.42
not significant and both could still predict the tensile Vf, respectively. As for the prediction by Manera and
modulus fairly well. Cox-Krenchel model, the deviations from experimental

Downloaded from jrp.sagepub.com at TEXAS SOUTHERN UNIVERSITY on June 6, 2016


8 Journal of Reinforced Plastics and Composites 0(0)

Figure 6. Experimental and predicted Poisson’s ratio of non- Figure 7. Experimental and theoretical tensile strength of
woven KF/epoxy composites at different Vf. epoxy and its composites at different Vf.

from subcritical and supercritical fibres as well as that


vC were 13.7%, 11.9%, and 8.5% at 0.15, 0.31, and 0.42 of the matrix. Accordingly, the equation for tensile
Vf, accordingly. In effect, the smaller percentage of strength is given as
deviations from the experimental values compared to
the Manera and Cox-Krenchel model validated that the C ¼ f K1 K2 Vf þ m Vm ð15Þ
Tsai–Pagano model was able to predict the elastic con-
stants of the NKFE composites under different Vf. This where K1 is the fibre orientation and K2 is the fibre
also indicated that a good prediction of composites’ length factor. For fibres with l > lc,
tensile modulus and Poisson’s ratio is possible by
taking into account the anisotropic nature of KF. l  lc
K2 ¼ ð16Þ
2l
Prediction of tensile strength
and for fibre with l < lc
Two pertinent models have been utilised to predict
the tensile strength of NKFE composites at various l
K2 ¼ ð17Þ
Vf,: the Hirsch’s model and Bowyer–Bader’s model. 2lc
These models have been used extensively in previous
literatures and have produced reliable predictions.38,39 in which l and lc represent the fibre length and critical
In particular, the first model proposed by Hirsch17 is fibre length. For randomly oriented fibre composites,
based on the combination of parallel (rule of mixture) K1 is 0.2 based on previous research.40 By using the
and series (inverse rule of mixture) models. According interfacial shear strength (IFSS) of KF obtained from
to the model, the tensile strength of the composite is the study by Park et al.41 alongside the information in
given as Table 1, the estimated critical length of KF was
1.64 mm. As the length of the KF is significantly
  m f larger than its critical fibre length, K2, therefore,
C ¼ x m Vm þ f Vf þ ð1  xÞ ð14Þ
m Vf þ f Vm becomes l/2l, thus equivalent to 0.5. The agreement
between theoretical and experimental values in previous
where  C,  m, and  f are the tensile strengths of the studies38,39 further corroborates the previous statement.
composite, matrix (epoxy), and KF, respectively. Vm From Figure 7, it can be seen that Hirsch’s model
and Vf are the volume fraction of matrix and fibre, overestimates the tensile strength of the NKFE
respectively, and x is the parameter which determines composites at every Vf. Specifically, the theoretical ten-
the stress transfer between the matrix and fibre. sile strength deviates by 39%, 53%, and 41% at 0.15,
For randomly oriented fibre composites, x is 0.1 0.31, and 0.42 Vf, respectively. Unlike in previous stu-
based on agreement between experimental and theoret- dies,35,36 this model has failed to represent the tensile
ical values.38 strength of the composites. Meanwhile, the tensile
The second model, which was used to predict the strength predicted by the Bowyer–Bader’s model was
strength, is by Bowyer and Bader.18 According to the much closer to the experimental values. Furthermore,
model, the tensile strength of short fibre-reinforced the deviation of the models at various Vf ranged from
thermoplastic composites is the sum of contributions 0.9% to 18.6%. The smaller deviation is due to the

Downloaded from jrp.sagepub.com at TEXAS SOUTHERN UNIVERSITY on June 6, 2016


Andre et al. 9

introduction of fibre orientation and length factor in 2. Kamath M and Bhat G. Cotton fiber nonwovens for
the equation, which are important factors affecting automotive composites. Int Nonwovens J 2005; 14:
the mechanical properties of composites. Nevertheless, 34–40.
the deviation of Bowyer–Bader’s model is still signifi- 3. Sreekumar PA, Albert P, Unnikrishnan G, et al.
cant, especially at higher Vf. At higher Vf, there would Mechanical and water sorption studies of ecofriendly
Banana fiber-reinforced polyester composites fabricated
be more fibre-rich area which would act as stress
by RTM. J Appl Polym Sci 2008; 109: 1547–1555.
concentration point. This, together with the poor
4. Idicula M, Sreekumar PA, Joseph K, et al. Natural fiber
fibre-matrix interaction, would consequently affect the hybrid composites-a comparison between compression
tensile strength of the composites. Unfortunately, these molding and resin transfer molding. Polym Compos
are not accounted for in the model, which explains the 2009; 30: 1417–1425.
deviations in the experimental values. 5. Sun Z, Zhao X, Wang X, et al. Predicting the elastic
properties of sisal fiber reinforced polypropylene compos-
ites by a new method based on generalized method of
Conclusion cells and laminate analogy approach. Compos Sci
Several conclusions were drawn from this study. First, the Technol 2014; 91: 45–49.
addition of nonwoven KF to the epoxy matrix resulted in 6. Ntenga R, Béakou A, Atangana Atéba J, et al.
the increase of tensile modulus. However, the tensile Estimation of the elastic anisotropy of sisal fibres by an
strength of NKFE composites decreased upon incorpor- inverse method. J Mater Sci 2008; 43: 6206–6213.
7. Epaarachchi J, Ku H and Gohel K. A simplified empir-
ation of 0.15 Vf but exhibited improvements at higher Vf.
ical model for prediction of mechanical properties of
Consequently, the highest tensile properties were attained
random short fiber/vinylester composites. J Compos
at 0.42 Vf. Second, the Tsai–Hahn model has allowed the
Mater 2009; 44: 779–788.
determination of transverse modulus Ef2 of single KF. 8. Islam MA and Begum K. Prediction models for the elas-
Additionally, it was noted that KF is highly anisotropic tic modulus of fiber-reinforced polymer composites: an
with Ef1 and Ef2 values of 26.06 and 2.50 GPa, respect- analysis. J Sci Res 2011; 3: 225–238.
ively. Third, by taking the anisotropic nature of KF into 9. Facca AG, Kortschot MT and Yan N. Predicting the
account, the Tsai–Pagano model in conjunction with the elastic modulus of natural fibre reinforced thermoplas-
Halpin-Tsai model has yielded fair predictions of EC. tics. Compos A Appl Sci Manuf 2006; 37: 1660–1671.
Likewise, the Tsai–Pagano model has produced the best 10. Tsai S and Pagano N. Invariant properties of composite
prediction of vC, further validating the assumption that materials. In: Composite materials workshop, 1968, pp.
the consideration of KF elastic anisotropy in a model will 233–253. St. Louis, MO: Technomic Publishing
yield a good prediction of the composite’s elastic con- Company.
stants. Moreover, a comparative study between the tensile 11. Cichocki FR and Thomason JL. Thermoelastic anisot-
ropy of a natural fiber. Compos Sci Technol 2006; 62:
strengths obtained from experimental and selected models
669–678.
revealed that the Bowyer–Bader model produced better
12. Gibson RF. Principles of composite material mechanics,
predictions due to the inclusion of fibre length and orien- 3rd ed. Boca Raton, FL: CRC Press, 2011.
tation factors. 13. Hahn HT and Tsai SW. Introduction to composite mater-
ials, 1st ed. Boca Raton, FL: CRC Press, 1980.
Declaration of conflicting interests 14. Manera M. Elastic properties of randomly oriented short
The author(s) declared no potential conflicts of interest with fiber-glass composites. J Compos Mater 1977; 11:
respect to the research, authorship, and/or publication of this 235–247.
article. 15. Cox HL. The elasticity and strength of paper and other
fibrous materials. Br J Appl Phys 1952; 3: 72–79.
16. Krenchel H. Fibre reinforcement: Theoretical and prac-
Funding
tical investigations of the elasticity and strength of fibre-
The author(s) disclosed receipt of the following financial sup- reinforced materials. Copanhagen: Akademisk Forlag,
port for the research, authorship, and/or publication of this 1964.
article: The authors would like to acknowledge the financial 17. Hirsch T. Modulus of elasticity of concrete affected by
support given by Universiti Sains Malaysia and the Ministry elastic moduli of cement paste matrix and aggregate. ACI
of Higher Education, Malaysia by providing RUC Grant J Proc 1962; 59: 427–452.
(Grant no.: 1001/PBAHAN/814134) and LRGS Grant 18. Bowyer WH and Bader MG. On the re-inforcement of
(Grant no.: 1001/PKT/8640012), respectively. thermoplastics by imperfectly aligned discontinuous
fibres. J Mater Sci 1972; 7: 1315–1321.
References 19. Shibata S, Cao Y and Fukumoto I. Flexural modulus of
1. Thilagavathi G, Pradeep E, Kannaian T, et al. Development the unidirectional and random composites made from
of natural fiber nonwovens for application as car interiors biodegradable resin and bamboo and kenaf fibres.
for noise control. J Ind Text 2010; 39: 267–278. Compos A Appl Sci Manuf 2008; 39: 640–646.

Downloaded from jrp.sagepub.com at TEXAS SOUTHERN UNIVERSITY on June 6, 2016


10 Journal of Reinforced Plastics and Composites 0(0)

20. Akil HM, Omar MF, Mazuki AAM, et al. Kenaf fiber based on flax fibres and polypropylene. Appl Compos
reinforced composites: a review. Mater Des 2011; 32: Mater 2000; 7: 351–372.
4107–4121. 32. Houshyar S, Shanks RA and Hodzic A. The effect of
21. Ochi S. Mechanical properties of Kenaf fibers and Kenaf/ fiber concentration on mechanical and thermal properties
PLA composites. Mech Mater 2008; 40: 446–452. of fiber-reinforced polypropylene composites. J Appl
22. Folster TH and Michaeli W. Flax—a renewable source of Polym Sci 2005; 96: 2260–2272.
reinforcing fibre for plastics. Kunststoffe Ger Plast 1993; 33. Halpin J and Kardos J. The Halpin-Tsai equations: a
83: 687–691. review. Polym Eng Sci 1976; 16: 344–352.
23. Hornsby PR, Hinrichsen E and Tarverdi K. Preparation 34. Puck A. Zur Beanspruchung und Verformung von GFK-
and properties of polypropylene composites reinforced Mehrschichtenverbund-Bauelementen: Teil 1. Grundlagen
with Wheat and Flax straw fibres part I fibre character- der Spannungs-und Verformungsanalyse. Teil 2. Spannungs-
ization. J Mater Sci 1997; 2: 443–449. und Verformungsanalyse an GFK-Wickelrohren unter
24. Uma Devi L, Bhagawan SS and Thomas S. Mechanical Überdruck. Teil 3. Versuche an Mehrschichtenverbunden.
properties of pineapple leaf fiber-reinforced. J Appl Sci Hanser, 1967.
1996; 64: 1739–1748. 35. Thomason JL and Vlug MA. Influence of fibre length
25. Al-Khanbashi A, Al-Kaabi K and Hammami A. Date and concentration on the properties of glass fibre-rein-
palm fibers as polymeric matrix reinforcement: fiber char- forced polypropylene: 1. Tensile and flexural modulus.
acterization. Polym Compos 2005; 26: 486–497. Compos A Appl Sci Manuf 1996; 27: 477–484.
26. Jawaid M, Khalil HPSA, Bakar AA, et al. Chemical 36. Bledzki AK and Gassan J. Composites reinforced with
resistance, void content and tensile properties of oil cellulose based fibres. Prog Polym Sci 1999; 24: 221–274.
palm/jute fibre reinforced polymer hybrid composites. 37. Pan N. The elastic constants of randomly oriented fiber
Mater Des 2011; 32: 1014–1019. composites: a new approach to prediction. Sci Eng
27. Madsen B, Thygesen A and Lilholt H. Plant fibre com- Compos Mater 1996; 5: 63–72.
posites—porosity and volumetric interaction. Compos Sci 38. Kalaprasad G, Joseph K, Thomas S, et al. Theoretical
Technol 2007; 67: 1584–1600. modelling of tensile properties of short Sisal fibre-
28. Mahjoub R, Yatim JM, Sam ARM, et al. Characteristics reinforced low-density polyethylene composites. J Mater
of continuous unidirectional Kenaf fiber reinforced
Sci 1997; 2: 4261–4267.
epoxy composites. Mater Des 2014; 64: 640–649.
39. Venkateshwaran N and Elayaperumal A. Banana fiber
29. Pervaiz M, Sain M and Ghosh A. Evaluation of the influ-
reinforced polymer composites—a review. J Reinf Plast
ence of fibre length and concentration on mechanical per-
Compos 2010; 29: 2387–2396.
formance of Hemp fibre reinforced polypropylene
40. Curtis PT, Bader MG and Bailey JE. The stiffness and
composite. J Nat Fibers 2006; 2: 67–84.
strength of a polyamide thermoplastic reinforced with
30. Thomason JL and Groenewoud WM. The influence of
glass and carbon fibres. J Mater Sci 1978; 13: 377–390.
fibre length and concentration on the properties of glass
41. Park JM, Son TQ, Jung JG, et al. Interfacial evaluation
fibre reinforced polypropylene: 2. Thermal properties.
of single Ramie and Kenaf fiber/epoxy resin composites
Compos A Appl Sci Manuf 1996; 27: 555–565.
using micromechanical test and nondestructive acoustic
31. Garkhail SK, Heijenrath RWH and Peijs T. Mechanical
properties of natural-fibre-mat-reinforced thermoplastics emission. Compos Interf 2012; 13: 105–129.

Downloaded from jrp.sagepub.com at TEXAS SOUTHERN UNIVERSITY on June 6, 2016

Vous aimerez peut-être aussi