Professional Documents
Culture Documents
Composites: Part B
journal homepage: www.elsevier.com/locate/compositesb
a r t i c l e
i n f o
Article history:
Received 18 February 2011
Received in revised form 24 August 2011
Accepted 30 November 2011
Available online 10 December 2011
Keywords:
A. Fibers
A. Carbon ber
B. Elasticity
B. Mechanical properties
C. Finite element analysis (FEA)
a b s t r a c t
This paper compares the storage moduli and loss factors for three different types of short ber reinforced
nanocomposites. Comparisons have been made for different ber volume fractions (Vf) as well as for different ber aspect ratios (l/d). Results show that staggered array of bers give higher values of storage
modulus as compared to regular array. Aspect ratio has little effect on E022 and G012 , except at low aspect
ratios. The storage moduli show an increasing trend with increasing ber volume fraction.
2011 Elsevier Ltd. All rights reserved.
1. Introduction
Damping is the inherent material property that dissipates energy during cyclic loading. Finite element modeling (FEM) can be
used for numerical computation of bulk properties based on the
geometry, properties, and volume fraction of constituent phases.
FEM involves discretization of a material representative volume
element (RVE) into elements for which the elastic solution leads
to determination of stress and strain eld. The coarseness of the
discretization determines the accuracy of the solution. Nano-scale
RVEs of different geometric shapes can be chosen for simulation of
mechanical properties. Finite element modeling is performed to
predict the response of a structure under prescribed boundary conditions and time independent applied loads, when linear response
behavior can be assumed with reasonable accuracy. The desired response quantities are generally, displacements, stresses, strains,
reactions, and energy. The basic equation for linear static analysis
may be written in the form;
Kfug fpg
where [K] is the linear stiffness matrix for the structure (known), {u}
is the nodal displacement vector (unknown) and {p} is the load vector (known).
Most of the work has been carried out for longitudinal storage
moduli and loss factors but little has been stated about transverse
and shear properties of discontinuous ber reinforced nanocomposites. In this study, transverse and shear properties of short ber
Corresponding author.
E-mail address: sumit_sharma1772@yahoo.com (S. Sharma).
1359-8368/$ - see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compositesb.2011.11.059
2. Literature review
In the approach, suggested by Mori and Tanaka [1] each inclusion behaves as an isolated inclusion, subject to the averaged stress
elds acting on it from all the other inclusions. The stresses, acting
on an inclusion and caused by the presence of other inclusions, are
superimposed on the applied stress. The idea of Mori and Tanaka
was to combine the Eshelby approach and the effective eld concept. This is done by dening the strain concentration tensor,
which relates the strain in the inclusion to the unknown strain in
the matrix instead of the applied strain, as in the case of the dilute
distribution model.
Benveniste [2] expanded the relations suggested by Mori
Tanaka and provided a general method for determining the effective properties of ber reinforced composites.
Chandra et al. [3] have provided comprehensive review of
damping involving macro-mechanical, micromechanical and visco-elastic approaches. According to them, the various composite
damping mechanisms are: visco-elastic nature of matrix or ber
materials, damping due to inter-phase, visco-plastic damping and
thermo-elastic damping. Chandra et al. [4] have predicted damping
coefcients of two-phase continuous ber reinforced composites.
They studied the effect of shape of ber cross section and ber volume fraction on various damping coefcients using visco-elastic
correspondence principle. They also determined loss factors for
composites reinforced with cylindrical continuous ber using 2D
micromechanical FEM/strain energy approach. They concluded
that longitudinal loss factor is independent of shape of ber and
478
software, is shown in Fig. 2. Similarly, the third type of arrangement of bers is shown in Fig. 3. The models are drawn for different ber volume fractions and ber aspect ratios. Each time, when
we consider a new ber volume fraction or a new ber aspect ratio,
a new model has to be drawn as we cannot use the same model for
different ber volume fractions and aspect ratios. In this study the
Z-direction represents the ber direction i.e. (1) direction and X
and Y-directions are transverse directions which refer to (3) and
(2), respectively. With these two types of arrangements, modeling
is done on NISA. Different cases are considered in this modeling.
These are listed below:
(i) In the rst case, length of the bers is kept constant and
diameter is varied.
(ii) In the second case, diameter of the bers is kept constant
and length is varied. Since, length of the bers is varying
gap between the bers is also varying.
Matrix
Fibers
Matrix
Fibers
Fig. 2. Arrangement of bers in RVE of Type 2. The two bers are touching the front and rear faces of RVE.
479
Matrix
Fibers
Fig. 3. Arrangement of bers in RVE of Type 3. The two bers are not touching the front and rear faces of RVE, hence, bers are shown as hidden.
Table 1
Boundary conditions for different types of loading.
Ux
Transverse loading (22)
Constraint
Force/displacement
Uy
Uz
0
0.001
Longitudinal shear loading (12) (Bottom face xed and top face given displacement in z direction)
Constraint
0
0
0
Force/displacement
0
0
0.001
Tx
Ty
Tz
0
0
0
0
0
0
0
0
0
0
0
0
g12) the boundary conditions are as shown in the Fig. 5. For nding
g21, the bottom of the RVE is constrained in all directions i.e., all the
nodes in the bottom face of the RVE have all degrees of freedom
equal to zero, and the nodes on the top face are given a small displacement, Uz = 0.001, in the z direction, which is the ber direction
97
97
Uy = 0.001
200
(a)
85
10
85
Uz = 0.001
200
(b)
Fig. 5. Boundary conditions for longitudinal shear loading (12).
Fig. 6. Different ways in which bers can be arranged in the RVE. (All dimensions in
nm).
480
or 1 direction. Here, g31 = g12, because of transverse isotropy. Fibers in RVE of Fig. 3 can be arranged in different ways. Two of these
are as shown in Fig. 6. While making the models on NISA, several
arrangements are used.
These are explained in the following sections.
(i) Type 1: This is a staggered arrangement in which the RVE
consists of one full central ber and the other bers, on
the top and bottom faces of the RVE are half of the full ber
i.e., they are semi-circles. On the top face, there are two
semi-circular bers and similarly on the bottom face, there
are two other semi-circular bers. This type of RVE is shown
in Fig. 1.
(ii) Type 2: In this type of model, there are two central bers separated by some inter-ber spacing as shown in Fig. 2.
(iii) Type 3: In this type, there are two central bers separated by
some inter-ber spacing as well as there is some distance
between the rst ber and the front face of the RVE and
between the second ber and the back face of the RVE as
shown in Fig. 3.
4. Theoritical formulation
The following theoretical formulation is used for predicting the
properties of Vapour Grown Carbon Fiber/Polypropylene (VGCF/
pp) nanocomposites:
Storage modulus E0
2W
V e2
Loss factor g
gf W f gm W m
W
Thus, the FEM results obtained through NISA software are used
in Eqs. (2)(4) for determining the moduli and corresponding loss
factors.
5. Modeling for E22
Boundary conditions for the case of transverse loading (22), are
shown in Table 1. Fig. 7 shows the boundary conditions on NISA
software. Bottom face of the RVE is xed and the top face is given
a small displacement of 0.001 in Y-direction. In Fig. 7, red1 colour
shows the xed position and the yellow colour shows that a small
displacement is given to the top face.
Fig. 8 shows the strain energy plot for transverse loading (22),
for a ber volume fraction of 0.03 and an aspect ratio of 10, for
Type 1 RVE. Here, there are 12,312 elements in the model. Number
of elements in the ber and matrix, respectively, is 4104 and 8208.
Displaced shape is shown in Fig. 9. Strain energy shared by ber
and matrix is shown below:
W f 7:51 108
W m 9:79 105
W m 6:59 107
W m 1:37 104
V 7:70 103
Fig. 11 shows another case of transverse loading where the model belongs to the Type 3. Between the front face and the rst ber
there is a spacing of 2 nm. Then, there is a spacing of 2 nm between
the rst and the second bers followed by a further spacing of 2 nm
between the second ber and the back face of the RVE. The effect of
increase in this inter-ber spacing on the modulus and the loss factor can be estimated. RVE consists of 2160 elements in the ber and
7560 elements in the matrix. The total number of elements in the
model is 9720 and the total number of nodes is 10,447. The strain
energy shared by ber and the matrix are as shown below:
W f 6:66 107
W m 1:37 104
Total volume of the RVE; V 7:70 103
6. Modeling for G31
The boundary conditions for the case of longitudinal shear loading (31) are shown in Table 1. Fig. 12 shows the boundary conditions on NISA. Left face of the RVE is xed and the right face is
given a small displacement (0.001) in the z-direction. Here, yellow
coloured dots represent xed state and red coloured dots represent
a small displacement in the z-direction.
Fig. 13 shows the strain energy plot for Type 1 RVE with a ber
volume fraction of 0.03 and an aspect ratio of 10. Here, the RVE
consists of 12,312 elements and the number of elements in the ber and matrix, respectively, is 4104 and 8208. Displaced shape is
shown in Fig. 14. Strain energy shared by ber and the matrix are
as shown below:
W f 4:42 108
1
For interpretation of color in Figs. 15 and 724, the reader is referred to the web
version of this article.
W m 1:65 107
481
Fig. 8. Strain energy plot for transverse loading for Type 1 RVE with Vf = 0.03 and (l/d)=10.
After
Deformation
Before
Deformation
Fig. 9. Displaced shape for transverse loading for Type 1 RVE with Vf = 0.03 and (l/d) = 10.
V 7:69 103
Fig. 15 shows the strain energy plot, for shear loading (31), for
Type 2 RVE, having an inter-ber spacing of 6 nm, with ber volume fraction of 0.03 and an aspect ratio of 97. The RVE consists
of 5832 elements and 6403 nodes. There are 2160 elements in
the ber and 3672 elements in the matrix. Strain energy shared
by ber and matrix for shear loading (31) is shown below:
W f 5:80 109
W m 4:92 105
482
Fig. 10. Strain energy plot for transverse loading for Type 2 RVE for Vf = 0.03 and (l/d) = 97 with inter-ber spacing of 6 nm.
Fig. 11. Strain energy plot for transverse loading for Vf = 0.03 and ber aspect ratio (l/d) = 97 with inter-ber spacing of Type 3.
W f 5:89 109
W m 4:88 105
Total volume of the RVE; V 7:69 103
7. Comparison of results
483
Fig. 13. Strain energy plot for shear loading (31) for Type 1 RVE for Vf = 0.03 and (l/d) = 10.
Before
Deformation
After
Deformation
Fig. 14. Displaced shape for shear loading (31) for Vf = 0.03 and (l/d) = 10.
in l/d, for l/d > 100, is very slow and become insensitive to l/d at
large aspect ratios. At l/d = 100, the percentage difference between
the E022 values for FEM models of Type 1 and 2, is 0.00358%, while at
l/d = 1000, the difference reduces to 0.0000579%. Thus, it can be
seen that at large l/d ratios, E022 is insensitive to the variation in l/
d. From l/d = 5 to l/d = 100, the percentage decrease in E022 for Type
1 RVE is 0.069% and from l/d = 100 to l/d = 1000, percentage decrease in E022 reduces to 0.0013%. A probable reason for rapid fall
of storage moduli till l/d = 100 is, increasing the ber length decreases the number of ber ends and thus the amount of matrix
between the ber ends. Since, the amount of matrix is xed, it
moves towards the space between the bers and increases the inter-ber spacing and thus decreases the values of E022 .
7.1.2. Variation of G012
Fig. 19 shows the variation of G012 with Vf for l/d = 19. The trend
is similar to that of Fig. 17. Storage modulus (G012 ) increases 25.53%
for Type 1 RVE, for an increase in ber volume fraction from Vf = 0
to Vf = 0.06. Similarly, for an increase in ber volume fraction from
Vf = 0.06 to Vf = 0.16, G012 increases 35.59%. At Vf = 0.06, the
484
Fig. 15. Strain energy plot for shear loading (31) for Type 2 RVE having Vf = 0.03 and (l/d) = 97 with inter-ber spacing of 6 nm.
Fig. 16. Strain energy plot for shear loading (31) for Vf = 0.03 and ber aspect ratio (l/d) = 97 with inter-ber spacing of Type 3.
difference between G012 values for Type 1 and Type 2 RVEs is 3.62%.
This difference increases to 3.75% at Vf = 0.16. Similarly, the difference between G012 values for Type 1 and Type 3 RVEs at Vf = 0.06 is
6.20% and this difference increases to 6.25% at Vf = 0.16.
Fig. 20 shows the variation of G012 with l/d for Vf = 0.03. Here
also, G012 decreases rapidly with l/d till l/d < 100 and then becomes
almost constant. For Type 1 RVE, G012 decreases 0.609% for an increase in ber aspect ratio from l/d = 5 to l/d = 100. Whereas, from
l/d = 100 to l/d = 1000, the decrease in G012 reduces to 0.001%. At l/
d = 100, the difference between the G012 values for FEM models of
Type 1 and 2, is 0.0377%. Whereas, at l/d = 1000, the difference reduces to 0.0328%, again showing that l/d has little effect on shear
storage modulus. Similarly, the difference between G012 values for
Type 1 and Type 3 RVEs at l/d = 100 is 0.09428% and this difference
485
Fig. 17. Variation of E022 with Vf for l/d = 19 for three different types of RVEs.
Fig. 19. Variation of G012 with Vf for l/d = 19 for three different types of RVEs.
0.5340
0.5335
Type 1
0.5330
0.5325
0.5320
Type 2
0.5315
0.5310
0.5305
Type 3
0.5300
0.5295
0
100 200 300 400 500 600 700 800 900 1000
Fig. 18. Variation of E022 with l/d for Vf = 0.03 for three different types of RVEs.
G012
486
0.060020
0.060000
Type 1
0.059980
0.059960
0.059940
Type 2
0.059920
0.059900
Type 3
0.059880
0.059860
0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18
0.0599690
0.0599685
Type 1
0.0599680
0.05999565
0.05999560
Type 1
0.05999555
0.0599675
Fig. 21. Variation of g22 with Vf for l/d = 19 for three different types of RVEs.
0.0599670
0.0599665
Type 2
0.0599660
0.0599655
0.0599650
Type 3
0.05999550
0.0599645
0.05999545
0.05999540
Type 2
0.0599640
0
100 200 300 400 500 600 700 800 900 1000
0.05999535
0.05999530
Fig. 24. Variation of g12 with l/d for Vf = 0.03 for three different types of RVEs.
0.05999525
Type 3
0.05999520
0.05999515
0
100 200 300 400 500 600 700 800 900 1000
Fig. 24 shows the variation of g12 with l/d for Vf = 0.03. For Type
1 RVE, g12 increases 0.006% for an increase in ber aspect ratio
from l/d = 5 to l/d = 100. Whereas, from l/d = 100 to l/d = 1000, g12
increases 0.00018%, for Type 1 RVE. At l/d = 100, the percentage difference between the g12 values for FEM models of Type 1 and 2, is
6.138 104% while at l/d = 1000, the difference reduces to
5.921 104%. Similarly, at l/d = 100, the percentage difference between the g12 values for FEM models of Type 1 and 3 is 9.5 104%
while at l/d = 1000, the difference reduces to 9.25 104%%. The
results obtained show that the trend of variation of g12 with l/d
is same as that of variation of g22 with l/d.
8. Conclusions
(v)
487
Decrease in damping with increase in ber volume fraction is more for the bers having large aspect ratios as
compared to the bers having small aspect ratios.
References
[1] Mori T, Tanaka K. Average stress in matrix and average elastic energy of
materials with mis-tting inclusions. Acta Materials 1973;21:5714.
[2] Benveniste Y. A new approach to the application of MoriTanakas theory in
composite materials. Mechanics of Materials 1987;6:14757.
[3] Chandra R, Singh SP, Gupta K. Damping studies in ber reinforced composites- a
review. Composite Structures 1999;46:4151.
[4] Chandra R, Singh SP, Gupta K. Micromechanical damping models for berreinforced composites: a comparative study. Composites: Part A
2002;33:78796.
[5] Chun Hway Hsueh. Youngs modulus of unidirectional discontinuous ber
composites. Composites Science and Technology 2000;60:267180.
[6] Tucker CL, Liang E. Stiffness predications for unidirectional short-ber
composites: review and evaluation. Composite Science and Technology
1999;59:65571.
[7] Wang WX, Dongmei Luo, Takao Y. Effects of the distribution and geometry
of carbon nanotubes on the macroscopic stiffness and microscopic stresses
of nanocomposites. Composite Science and Technology 2007;67:
294758.