Theoretical study of the effective modulus of a composite

considering the orientation distribution of the fillers and the

weakened interface

Sangryun Lee1 and Seunghwa Ryu*,1

Affiliations

1
Department of Mechanical Engineering, Korea Advanced Institute of Science and

Technology (KAIST), 291 Daehak-ro, Yuseong-gu, Daejeon 305-701, Republic of Korea

*
Corresponding author e-mail: [email protected]

Keywords
Homogenization; Micromechanics; Orientation average; Weakened interface
Abstract

In the manufacturing process of a filler-reinforced composite, the fillers in the matrix are

aligned due to the shear flow occurring during the drawing stage, and the interface between the

matrix and the fillers form various imperfections that lead to debonding and slip under

mechanical loading. Hence, there have been numerous micromechanics studies to predict

effective moduli of the composites in the presence of partial alignment of fillers and interface

imperfections. In this study, we present an improved theory that overcomes two limitations in

the existing micromechanics based approaches. First, we find that the interface damage tensor,

which has been developed to model the weakened interface between matrix and fillers, has

singularities that cause non-physical predictions (such as infinite or negative effective moduli).

We correct the mathematical mistakes to remove singularities and derive analytic expressions

of the damage tensor for ellipsoidal inclusions. Second, we reveal that the previous theory on

the effective moduli with axisymmetric filler orientation distribution fails because the

longitudinal and transverse moduli do not converge in the limit of random orientation

distribution. With appropriate corrections, we derive an analytic expression for the orientation

average of arbitrary transversely isotropic 4th order tensor under general axisymmetric

orientation distribution. We apply the improved method to compute the effective moduli of a

representative composite with non-uniform filler orientation and interface damage.

Nanocomposites (such as polymer reinforced with carbon nanotubes (CNTs), graphene,

and nanowires), which exhibit remarkable mechanical, electrical, and thermal properties

compared to the properties of the bare matrix, have attracted considerable attention because of

their numerous potential applications, including flexible sensors[1, 2], lithium batteries[3-5],

optical device[6, 7], and flexible energy storage devices[8, 9]. To effectively design and make

use of the nanocomposites, it is essential to accurately predict the effective properties of the

composite as a function of shape, volume fraction, and orientation distribution of fillers.

For example, the piezoresistivity of the metal nanowire-polydimenthylsiloxane

(PDMS) composites has been utilized for the highly stretchable strain sensors[2]. The effective

electrical property of such composites has been analyzed by solving three dimensional electric

circuit constructed from a percolation network of thousands of nanowires[1]. For other

examples, the finite element method (FEM) was used to compute the effective moduli of

composites at the bulk scale[10-12], whereas molecular dynamics (MD) simulations were used

at the atomic scale[13-15]. The aforementioned examples require computationally expensive

and time-consuming calculations of a large simulation cell involving many fillers for serving

as the representative volume element (RVE).

To compute the effective properties more efficiently, many micromechanics-based

approaches have been used, such as the Eshelby method, the Mori-Tanaka (MT) method, and

the Self-Consistent (SC) method. Eshelby was the first to introduce the Eshelby tensor (S) and

to solve the single inhomogeneity in the matrix by converting the problem into an equivalent

eigenstrain problem[16]. The S tensor, which is a 4th order tensor that depends on the matrix

properties and the shape of fillers, relates the eigenstrain of equivalent inclusion (fillers) and

the constrained strain of matrix. As extensions of the approach, the Eshelby method, the MT
method, and the SC method are devised; these methods are mean-field homogenization

schemes to treat the multiple inhomogeneity problems[17]. The MT method is the most popular

method because it provides more accurate predictions on the effective properties than the

Eshelby method and has an explicit and closed-form solution, whereas the SC method relies

on implicit equations. However, the original MT method suffers from two limitations: first, it

does not account for the imperfections in the filler-matrix interface, such as debonding and slip;

second, it is only applicable when the fillers in the matrix are perfectly aligned in one direction.

To take into account the reduced effective modulus from the weakened interface, Qu

et al. introduced the interface spring model, which assumes a virtual linear spring at the

interface to allow for a displacement jump across the interface[18]. The 4th order interfacial

tensor (R tensor) (H tensor in Qu et al.[18]) was defined to represent the degree of interface

damage as a function of the filler shape and the compliance of the interface springs. The

modified Eshelby tensor is then obtained by using the R tensor, which eventually leads to the

change in the effective modulus prediction. Barai et al. first derived the analytic expression of

the R tensor for the axisymmetric ellipsoidal fillers[19]. Because the shape of most fillers

(such as spherical nanoparticles or carbon nanotube) can be approximated as axisymmetric

ellipsoids with varying aspect ratio, the approach has been used to predict the effective moduli

of various composites[14, 20, 21]. However, we find that the mathematical errors and the

resulting singularity in R tensor have not been recognized and mentioned in the previous

studies14, 20, 21.

In addition, in the manufacturing process of filler-reinforced composite fillers, the

fillers in the matrix are aligned because of the shear flow during the drawing stage[22-24].

Hence, it is crucial to predict the effective modulus as a function of the filler orientation

distribution[25]. Odegard et al. suggested the orientation average of 4th order stiffness tensor
using the 3-1-3 Euler angle and showed that the effective moduli become anisotropic when the

fillers are partially or fully aligned[26]. However, we find that their prediction on the effective

moduli with axisymmetric filler orientation distribution fails because the longitudinal and

transverse moduli do not converge in the limit of random orientation distribution. Meantime,

the orientation average scheme has been applied to the study of the effective moduli of

composites with various types[15, 27, 28].

In this work, we propose an improved micromechanics model to correct the two

problems regarding the interface damage and the orientation average. First, we obtain the

analytic expressions by solving the integral form of R tensor analytically for axisymmetric

ellipsoidal fillers and validate our results against the numerical integration results. We also

confirm that our expression satisfies two limiting cases, i.e., aspect ratio of 1 and infinity, where

analytic results are readily available[18]. We demonstrate that the singularities in effective

modulus prediction can be removed when the correct R tensor is used. Second, instead of the

3-1-3 Euler angle, we use polar and azimuthal angles on the unit sphere to represent the filler

orientation and derive algebraic expressions for the general transversely isotropic 4th order

tensor under axisymmetric filler orientation distribution. We confirm that the longitudinal and

transverse elastic moduli from our model converge to the moduli of randomly oriented case of

which the analytic expression exits[17]. In summary, we derive analytic expressions for the

general axisymmetric orientation distribution of axisymmetric ellipsoidal fillers with a

damaged interface described by the linear spring model, which can be widely used to describe

composites that include particles and fillers at various aspect ratios.

Figure 1 (A) Schematic of RVE with partially aligned fillers. (B) Interface spring model to

simulate the damaged interface between matrix and filler. (C) The local coordinates (x , x , x )

and global coordinates (X , X , X ) displayed with the spherical coordinates. (D) Modulus in

the longitudinal and transverse directions. The dashed lines are the results from known analytic

expressions. The parameters used in calculation are listed in Table 1 (E) Previous and corrected

Q1122 component, one of the damaged interface properties components, with respect to the

aspect ratio of the filler. The triangular symbol is the known value at the aspect ratio of 1

reported by Qu et al.[18].

We adopt Hill’s notation to calculate the effective modulus of RVE, as shown Fig. 1(A).

To perform the complex calculations of 4th order tensors, such as double-contraction and
inverse conveniently, we follow the Walpole’s scheme[17]. In what follows, the capital sized

alphabets represent 4th order tensors, and the colons denote double contraction operations. The

effective stiffness tensor of the composite in the original MT approach is expressed as below.

: : (1)

where and are the stiffness tensors of the matrix and filler, respectively. I is the 4th

order identity tensor. and refer to the volume fraction of matrix and fillers, respectively;

thus, 1. is the strain concentration tensor defined in Eq. (2) and is expressed in

terms of Eshelby tensor ( ) and stiffness tensors.

∗
≡ : , : : , ≡ : (2)

∗
Here, is the eigenstrain, and     are the averaged strain values within the matrix

and the fillers, respectively. The Eshelby tensor (S) for the filler whose shape is axis-symmetry

ellipsoid is taken from Qiu et al.[17].

As mentioned previously, the original MT method is only applicable when the fillers

are completely aligned and have perfect bonding with the matrix. The modified Mori-Tanaka

(mMT) approach must be employed to account for the imperfect bonding at the interface (slip

or debonding). To model the weakened interface bonding, we consider the displacement jump

across the interface by adopting the linear spring model[18], given by Eq. (3).

Δ , (3)

where the tensor refers to the compliance of the interface spring in the tangential  and

normal  directions. After solving the eigenstrain problem with the displacement jump, we

obtain the modified Eshelby tensor  as follows:

: : : . (4)
 In this work, we limit our focus on the axisymmetric ellipsoidal fillers with two semi-

axial lengths of a along the x direction and a along the x and x directions. Due to

the nonphysical overlapping arising from a nonzero , the tensor is typically expressed by

considering only tangential spring compliance as

where

3
16

and

3
.
4

with , , , cos sin , and  .

The non-zero components of and Q can be obtained from the integral given below:

3 1 1 1 3 2 1
,   
2 1 1 2 12 1 2 1

1 1
,
4 2

3 2 1 3 1 3 2 1 3
/
,    /
2 1 1 4 1 1

9 2 5 4 1 1
/
, .
8 2 1 2 1 3

Other components can be obtained by using the symmetry condition in the 2-3 plane as well as

the minor and major symmetry of and Q tensors. In the limit of zero spring compliance,

the modified Eshelby tensor becomes the original Eshelby tensor because becomes zero

tensor, and hence, mMT method becomes MT method. The effective moduli in the mMT
scheme can be obtained by replacing the original Eshelby tensor with the modified Eshelby

tensor , as presented below[18]:

: : : : (5)

where is the modified strain concentration tensor, : : .

We find that there exists a mathematical error in Q1122 component and typographical

errors in the Q1111 and Q2222 components in the previous study[19] and that these errors were

applied to the several follow-up studies[14, 19-21].We plot the Q1122 component in terms of the

aspect ratio  in Fig. 1(E). The uncorrected Q1122 diverges as the aspect ratio goes to unity and

overestimates the value for the entire range of . Because the R tensor plays a critical role in

determining modified Eshelby tensor and effective modulus, we expect that the effective

modulus with non-corrected R tensor will have singularities at some values of the aspect ratio.

We confirm that the newly obtained analytic expression of Q1122 predicts the same result with

the numerical integration with Gaussian quadrature. In addition, our expression matches with

the readily available analytic expression at two limiting case of 1 and ∞ obtained

by Qu et al.[18]. We present the validation of all other nonzero components in Appendix A.

The derived mathematical expression on the effective modulus is only applicable for

the composites with completely aligned fillers. However, usually, fillers are randomly oriented

or partially aligned in the manufacturing processes[23, 24, 29]. Because the originally

randomly oriented fillers in the liquid-state matrix are partially aligned along the drawing axis

in most manufacturing processes and experiments, we limit our focus on the axisymmetric

orientation distribution. Following the previous studies, we define the orientation averaged

Mori-Tanaka (oaMT) as Eq. (6).

: : : : (6)
where the operator < > denotes the orientation average of each tensor. When the shape of fillers

is an axisymmetric ellipsoid, the orientation average of an arbitrary 4th order tensor X can be

defined as

,
(7)

where and are azimuthal and polar angle with respect to the global coordinate (see Fig.

1(C)). ′ is the tensor transformed to global coordinate system. Following the coordinate

transformation rule of 4th order tensor with rotation matrix c, can be expressed as Eq.(8).

cos sin 0
    sin cos cos cos sin (8)
sin sin cos sin cos

The axisymmetric filler orientation distribution can be expressed as a function of θ

only. The axisymmetric orientation distribution function, , can be categorized into three

types: 3D random, normal distribution, and aligned.

1∶ 

exp ∶    (9)

∶ 

When in the normal distribution goes to zero or infinite, the distribution converges to

random or fully aligned distributions, respectively. We note that previously used 3-1-3 Euler

angle set cannot describe the axisymmetric distribution along the X axis due to the

geometrical constraint, while several studies have adopted the effective modulus from the 3-1-

3 Euler angle set to describe the composites with axisymmetric filler orientation distribution[15,

27, 28]. When k goes to zero, i.e., for a random orientation distribution, the composite must
behave as an isotropic material. However, as shown in Fig. 1(D), the longitudinal and

transverse Young’s modulus from the previous study do not converge to the same value in the

random orientation limit. In contrast, the oaMT with our angle set predicts that both

longitudinal and transverse moduli approach the modulus of the composite with randomly

oriented fillers, whose analytic expression is available.

The evaluation of the orientation average in Eq. (7) is rather difficult because it

involves complex transformations of 4th order tensor to global axis. Hence, no analytic

expression exists on the orientation average for general axisymmetric distribution, whereas

analytic expressions were derived for the two limiting cases of the fully random and fully

aligned distributions. To facilitate the use of the oaMT, we derive algebraic expressions of the

orientation averaged transverse isotropic 4th order tensor under general axis-symmetry

distribution of in Appendix B. The six independent components of orientation averaged

4th order transversely isotropic tensor can be written as linear combinations of eight constants

that are easily obtainable via simple numerical integrals. We expect that our orientation average

algebraic expression can be widely applicable to the problems, including 4th order tensors.
 3. Results and Discussions

Figure 2 (A) Effective Young’s modulus and (B) shear modulus of a composite with randomly

oriented fillers. (C) Longitudinal Young’s modulus and (D) transverse Young’s modulus of a

composite whose orientation distribution of fillers is aligned. The insets in each figure show

the results near the aspect ratio of 1 to highlight that our model satisfies the limit case with the

known analytic expression of spherical fillers. The tangential spring compliance of 10 nm/GPa

with zero normal spring compliance is used.

LaRC-CP2 polymer composite. The data are based on other research studies[19, 26]. The

volume fraction of CNT is 0.01 (1%), and the tangential compliance at weakened interface

between the polymer and CNT is set 10 nm/GPa[19].

Material properties Matrix phase Carbon nanotube (CNT)

Young’s modulus (E) 0.85 GPa

Poisson’s ratio ν 0.4

Longitudinal Young’s modulus E 1.06 TPa

Transverse bulk modulus 271 GPa

Transverse shear modulus 17 GPa

In-plane shear modulus 442 GPa

In-plane Poisson’s ratio 0.162

We apply our new oaMT with the damaged interface to predict the effective moduli of

CNT-reinforced LaRC-CP2 polymers. The material constants used in our calculations are listed

in Table 1. Following previous studies[19], we only consider the tangential part in the linear

spring model (see Fig. 1 (B)) and set the normal compliance to zero 0 because finite

normal spring compliance may cause a nonphysical configuration, such as the overlap of the

filler and matrix volumes. To avoid such a problem, it is necessary to set the spring compliance

to be a non-linear function of the displacement jump, which makes the evaluation of effective

moduli highly difficult[30]. Because this issue is an independent topic in the micromechanics

community, we will not consider the normal compliance in our work. Indeed, the tangential

spring model is sufficient to model the interfacial slip.

First, we compare the effective moduli of composites with the previous R tensor and

the corrected R tensor. We correct two simple typographical errors in the Q1111 and Q2222
components, but leave the mathematical error in the Q1122 component when considering the

previous R tensor. Fig. 2 depicts the elastic moduli as functions of the filler aspect ratio for

the two limiting cases of fully random and fully aligned distributions. Figs. 2 (A and B) show

the Young’s modulus and the shear modulus of the composite when the fillers are randomly

oriented. Under perfect interfacial bonding condition, both elastic moduli increase with the

aspect ratio of fillers at a given volume fraction (Fig. 2 (A and B)). The moduli dramatically

rise when the aspect ratio reaches approximately 100 and saturate when the aspect ratio exceed

1000. The same trend can be found in the longitudinal and transverse Young’s modulus in

another limit case of fully aligned fillers (Fig. 2 (C and D)). When the interfacial imperfection

is taken into account, the effective modulus should become lower than the perfect bonding

condition. As shown in the Fig. 2, the effective moduli obtained from the previous R tensor

are lower than those with perfect bonding in most of the range. However, they have singularities

or negative values at few specific aspect ratios due to the presence of a singularity in the

wrongly obtained Q1122 component. The singularity in the effective moduli can only be found

when the aspect ratio is less than 100, and the number of singularities change with material

properties of each phases and the volume fraction of fillers. The insets in Fig. 2 show that

effective modulus at the aspect ratio of unity, i.e., 1. The effective moduli based on

previous R tensor does not converge to the effective moduli of composite with spherical fillers,

i.e., = 1, whose analytic expression is available[18]. In comparison, the effective moduli

obtained from the corrected R tensor are continuous functions of the aspect ratio, are always

lower than the effective moduli with perfect bonding condition, and converge to the known

values for spherical fillers. We suspect that these problems have not been identified because

previous studies consider the composites having fillers in the range of very high aspect ratio (>

1,000) where the previous R tensor does not show a singularity[14, 19-21]. In addition, for

large aspect ratio, the difference between the results with previous and corrected R tensors
becomes rather negligible because the Q1122 component becomes nearly zero in the limit of

large aspect ratio (Fig. 1(E)). However, we note that the use of the corrected R tensor is crucial

in the small aspect ratio range (< 100).

Figure 3 (A) Longitudinal Young’s modulus and (B) transverse Young’s modulus with a

normal distribution of fillers, exp , as functions of k and aspect ratio under

fixed volume fraction (1%). (C) Longitudinal Young’s modulus and (D) transverse Young’s

modulus with respect to the degree of orientation (k) and the volume fraction of the fillers under

fixed aspect ratio (1000).

Second, we compute the effective modulus when the fillers are partially aligned with

the orientation distribution of exp . We can study the effect of alignment by

easily evaluated from the linear combinations of eight constants, as described in detail in

Appendix B. Figs. 3 (A and B) show the change of longitudinal and transverse Young’s

modulus with k and aspect ratio at a fixed volume fraction (1%). Because the moduli are

calculated from the corrected R tensor, they are free of singularities. When the degree of filler

alignment reduces (i.e., when k decreases), the difference between two moduli diminishes and

they eventually converges at 0. We find that both moduli increase with the aspect ratio.

In contrast, when k increases, the longitudinal modulus increases whereas the transverse

modulus decreases, as expected. We also demonstrate the combined effect of volume fraction

and orientation distribution in Figs. 3(C and D). Both moduli increase when the volume fraction

increases, and they approach the modulus of matrix in the zero volume fraction limit. We note

that the contour plot depicted in Fig. 3 can be easily obtained for any arbitrary axisymmetric

orientation distribution of axisymmetric ellipsoidal fillers by computing the eight constants in

Appendix B, which will facilitate the use of the micromechanics approach in designing and

predicting the performance of the composites.

We computed the effective moduli of composites as functions of the aspect ratio of fillers,

degree of orientation, and interfacial damage. With appropriate corrections in the previous

interfacial damage tensor and orientation average, the spurious singularities in the effective

composite modulus predictions are removed, and the longitudinal and transverse moduli

converges to a single value as the orientation distribution of fillers becomes random. We

demonstrated that the magnitude of moduli can be tuned by either the aspect ratio or the volume

fraction and the directional elastic properties can be tuned by controlling the orientation

distribution of fillers. To facilitate the application of our method, we derived an analytic

expression for the orientation average for arbitrary axisymmetric filler orientation distribution.

We believe that our study can provide a comprehensive picture and guidance in the effort to

tune the effective moduli of composite materials.

Acknowledgements

This work is supported by the Basic Science Research Program (2013R1A1A1010091) of the

National Research Foundation of Korea (NRF), funded by the Ministry of Science, ICT &

Future Planning.
Appendix A. Components of weakened interface property tensor, R

Figure A1 (A) P1111component; (B) P1212component; (C) P2222component; (D) Q1111component;

(E) Q2222component.
Appendix B Orientation average with the rotation matrix

The orientation average of the transversely isotropic 4th order tensor under arbitrary axis-

symmetry distribution function,  , is calculated as

  2 , , , ,2 ,2 → 2 , , , ,2 ,2

Where the tensors are expressed following Hill’s notation for the transversely isotropic 4th order

tensor. In the case of the transversely isotropic 4th order material stiffness tensor, the six elastic

constants can be determined by following Hooke’s law, which is expressed as

2 2 ,

2 ,          2

2 ,          2

The orientation averaged 4th order transversely isotropic tensor, , can be calculated by

using eight constants from numerical integrations.

1
2 2 4
4

1
4
2

1
4
2

1
4

1
2 6 4 4
8

1
5 4 8 4 4
16

With
 sin ,      cos sin ,      cos sin

cos sin ,      sin ,      sin

cos2 ,      cos4

When is the normal distribution with almost zero k, the orientation average expression

in the Appendix reproduces the analytic results of the random orientation case, as already

derived in some research studies[14, 17, 19].

[1] Lee S, Amjadi M, Pugno N, Park I, Ryu S. Computational analysis of metallic nanowire-elastomer
nanocomposite based strain sensors. AIP Advances. 2015;5(11):117233.
[2] Amjadi M, Pichitpajongkit A, Lee S, Ryu S, Park I. Highly Stretchable and Sensitive Strain Sensor
Based on Silver Nanowire–Elastomer Nanocomposite. ACS Nano. 2014;8(5):5154-63.
[3] Cui G, Gu L, Zhi L, Kaskhedikar N, van Aken PA, Müllen K, et al. A Germanium–Carbon
Nanocomposite Material for Lithium Batteries. Advanced Materials. 2008;20(16):3079-83.
[4] Huang H, Yin SC, Kerr T, Taylor N, Nazar LF. Nanostructured Composites: A High Capacity, Fast
Rate Li3V2(PO4)3/Carbon Cathode for Rechargeable Lithium Batteries. Advanced Materials.
2002;14(21):1525-8.
[5] Oh S-M, Oh S-W, Yoon C-S, Scrosati B, Amine K, Sun Y-K. High-Performance Carbon-LiMnPO4
Nanocomposite Cathode for Lithium Batteries. Advanced Functional Materials. 2010;20(19):3260-5.
[6] Lü C, Cheng Y, Liu Y, Liu F, Yang B. A Facile Route to ZnS–Polymer Nanocomposite Optical
Materials with High Nanophase Content via γ-Ray Irradiation Initiated Bulk Polymerization.
Advanced Materials. 2006;18(9):1188-92.
[7] Lü C, Guan C, Liu Y, Cheng Y, Yang B. PbS/Polymer Nanocomposite Optical Materials with High
Refractive Index. Chemistry of Materials. 2005;17(9):2448-54.
[8] Nyholm L, Nyström G, Mihranyan A, Strømme M. Toward Flexible Polymer and Paper-Based
Energy Storage Devices. Advanced Materials. 2011;23(33):3751-69.
[9] Pushparaj VL, Shaijumon MM, Kumar A, Murugesan S, Ci L, Vajtai R, et al. Flexible energy storage
devices based on nanocomposite paper. Proceedings of the National Academy of Sciences.
2007;104(34):13574-7.
[10] Banerjee S, Sankar BV. Mechanical properties of hybrid composites using finite element method
based micromechanics. Composites Part B: Engineering. 2014;58:318-27.
[11] Sodhani D, Reese S. Finite Element-Based Micromechanical Modeling of Microstructure
Morphology in Filler-Reinforced Elastomer. Macromolecules. 2014;47(9):3161-9.
[12] Sun H, Di S, Zhang N, Wu C. Micromechanics of composite materials using multivariable finite
element method and homogenization theory. International Journal of Solids and Structures.
2001;38(17):3007-20.
[13] Yang BJ, Shin H, Lee HK, Kim H. A combined molecular dynamics/micromechanics/finite element
approach for multiscale constitutive modeling of nanocomposites with interface effects. Applied
Physics Letters. 2013;103(24):241903.
[14] Yang S, Yu S, Ryu J, Cho J-M, Kyoung W, Han D-S, et al. Nonlinear multiscale modeling approach
to characterize elastoplastic behavior of CNT/polymer nanocomposites considering the interphase
and interfacial imperfection. International Journal of Plasticity. 2013;41:124-46.
[15] Alian AR, Kundalwal SI, Meguid SA. Interfacial and mechanical properties of epoxy
nanocomposites using different multiscale modeling schemes. Composite Structures. 2015;131:545-
55.
[16] Eshelby JD. The Determination of the Elastic Field of an Ellipsoidal Inclusion, and Related
Problems. Proceedings of the Royal Society of London A: Mathematical, Physical and Engineering
Sciences. 1957;241(1226):376-96.
[17] Qiu YP, Weng GJ. On the application of Mori-Tanaka's theory involving transversely isotropic
spheroidal inclusions. International Journal of Engineering Science. 1990;28(11):1121-37.
[18] Qu J. The effect of slightly weakened interfaces on the overall elastic properties of composite
materials. Mechanics of Materials. 1993;14(4):269-81.
[19] Barai P, Weng GJ. A theory of plasticity for carbon nanotube reinforced composites. International
Journal of Plasticity. 2011;27(4):539-59.
[20] Shokrieh MM, Ghajar R, Shajari AR. The effect of time-dependent slightly weakened interface
on the viscoelastic properties of CNT/polymer nanocomposites. Composite Structures.
2016;146:122-31.
[21] Pan Y, Weng GJ, Meguid SA, Bao WS, Zhu ZH, Hamouda AMS. Interface effects on the
viscoelastic characteristics of carbon nanotube polymer matrix composites. Mechanics of Materials.
2013;58:1-11.
[22] Fan Z, Advani SG. Characterization of orientation state of carbon nanotubes in shear flow.
Polymer. 2005;46(14):5232-40.
[23] Fu SY, Lauke B, Mäder E, Yue CY, Hu X. Tensile properties of short-glass-fiber- and short-carbon-
fiber-reinforced polypropylene composites. Composites Part A: Applied Science and Manufacturing.
2000;31(10):1117-25.
[24] Pötschke P, Brünig H, Janke A, Fischer D, Jehnichen D. Orientation of multiwalled carbon
nanotubes in composites with polycarbonate by melt spinning. Polymer. 2005;46(23):10355-63.
[25] Buck F, Brylka B, Müller V, Müller T, Weidenmann KA, Hrymak AN, et al. Two-scale structural
mechanical modeling of long fiber reinforced thermoplastics. Composites Science and Technology.
2015;117:159-67.
[26] Odegard GM, Gates TS, Wise KE, Park C, Siochi EJ. Constitutive modeling of nanotube–reinforced
polymer composites. Composites Science and Technology. 2003;63(11):1671-87.
[27] Pan J, Bian L, Zhao H, Zhao Y. A new micromechanics model and effective elastic modulus of
nanotube reinforced composites. Computational Materials Science. 2016;113:21-6.
[28] Nguyen HG, Ortola S, Ghorbel E. Micromechanical modelling of the elastic behaviour of polymer
mortars. European Journal of Environmental and Civil Engineering. 2013;17(2):65-83.
[29] Zaixia F, Zhangyu, Yanmo C, Hairu L. Effects of Pre-stretching on the Tensile Properties of Knitted
Glass Fiber Fabric Reinforced Polypropylene Composite. Journal of Thermoplastic Composite
Materials. 2006;19(4):399-411.
[30] Wang J, Duan HL, Zhang Z, Huang ZP. An anti-interpenetration model and connections between
interphase and interface models in particle-reinforced composites. International Journal of
Mechanical Sciences. 2005;47(4–5):701-18.