A Multi-Scale Model For Coupled Heat Conduction and Deformations of Viscoelastic Functionally Graded Materials

Composites: Part B 40 (2009) 511521
Contents lists available at ScienceDirect
Composites: Part B
journal homepage: www.elsevier.com/locate/compositesb
A multi-scale model for coupled heat conduction and deformations

of viscoelastic functionally graded materials
Kamran A. Khan, Anastasia H. Muliana *
Department of Mechanical Engineering, Texas A&M University College Station, TX 77843-3123, USA
a r t i c l e
i n f o
Article history:
Received 4 December 2008
Received in revised form 25 January 2009
Accepted 6 February 2009
Available online 7 May 2009
Keywords:
B. Creep
A. Particle reinforcement
C. Micro-mechanics
B. Thermomechanical Functionally graded
material (FGM)
a b s t r a c t
An integrated micromechanical-structural framework is presented to analyze coupled heat conduction
and deformations of functionally graded materials (FGM) having temperature and stress dependent viscoelastic constituents. A through-thickness continuous variation of the thermal and mechanical properties of the FGM is approximated as an assembly of homogeneous layers. Average thermo-mechanical
properties in each homogeneous medium are computed using a simplied micromechanical model for
particle reinforced composites. This micromechanical model consists of two isotropic constituents. The
mechanical properties of each constituent are timestresstemperature dependent. The thermal properties (coefcient of thermal expansion and thermal conductivity) of each constituent are allowed to vary
with temperature. Sequentially coupled heat transfer and displacement analyses are performed, which
allow analyzing stress/strain behaviors of FGM having time and temperature dependent material properties. The thermo-mechanical responses of the homogenized FGM obtained from micromechanical
model are compared with experimental data and the results obtained from nite element (FE) analysis
of FGMs having microstructural details. The present micromechanical-modeling approach is computationally efcient and shows good agreement with experiments in predicting time-dependent responses
of FGMs. Our analysis forecasts a better design for creep resistant materials using particulate FGM
composites.
2009 Elsevier Ltd. All rights reserved.
1. Introduction
Functionally graded materials (FGMs) are composite materials
in which the physical and mechanical properties of the materials
vary spatially along specic directions over the entire domain.
Changes in the composition of the constituents result in a nonuniform microstructure leading to gradual variations of the macroscopic material properties. Structures made of FGMs are often
subjected to high temperature gradient loadings. Under such conditions, the properties of the constituents can vary signicantly
with temperature accompanied by a non-negligible time-dependent response. For example, FGMs composed of metal and ceramic
constituents tend to creep at high temperatures. In addition, nonuniform temperature elds and mismatch in the properties of the
constituents in FGMs generate thermal/residual stresses that affect
overall performance of FGMs. Therefore, understanding nonlinear
thermo-viscoelastic behavior of FGMs plays a signicant role in
evaluating the performance of structures made from such materials. This study investigates the non-linear viscoelastic behavior of
FGMs during transient heat conduction process. The thermal and
* Corresponding author. Tel.: +1 979 458 3579; fax: +1 979 845 3081.
E-mail address: amuliana@neo.tamu.edu (A.H. Muliana).
1359-8368/$ - see front matter 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compositesb.2009.02.003
mechanical properties of the constituents are allowed to change

with time, stress and temperature.
Extensive numerical and analytical models have been developed to study the macroscopic thermal, elastic and inelastic behaviors of FGMs. Noda [17] and Tanigawa [27] provided detailed
reviews on thermo-elastic and thermo-inelastic studies in FGMs
having temperature dependent/independent material properties.
Limited analytical models have been developed to study the linear
viscoelastic macroscopic behavior of FGMs, e.g., Yang [30], Paulino
and Jin [20] and Mukherjee and Paulino [13]. Finite element (FE)
formulations have also been developed to study the thermomechanical behavior of cylinders and plates having graded material properties. Examples are given in Reddy and Chin [22], Praveen
et al. [21], Reddy [23] and Shabana and Noda [26]. Few experimental studies have been performed to determine the variation of the
thermal as well as mechanical properties of FGMs and are mainly
limited to thermo-elastic behavior, e.g., Zhai et al. [32,33] and Parameswaran and Shukla [19].
The microstructural details of FGMs are complex and the distribution of the inclusions in FGMs vary from one manufacturing process to others. This drives the modelers to create idealized
inclusion distributions and particle geometries in order to analyze
the thermo-mechanical microscopic responses of FGMs. Simulating the microscopic responses of FGMs using detailed FE modeling
512
K.A. Khan, A.H. Muliana / Composites: Part B 40 (2009) 511521
of the graded microstructure with some idealized geometry is

computationally expensive and inadvisable for practical applications. Micromechanical models have an advantage over other modeling techniques because they can give macroscopic properties of
non-homogeneous material while recognizing the microscopic
properties of each constituent.
Several micromechanical-modeling approaches have been used
to study the elastic behavior of composites including FGMs. Detailed discussions of available micromechanical models to obtain
effective mechanical properties of the composites can be found
in Nemat-Nasser and Hori [18]. Micromechanical models like the
self-consistent method (SCM), Mori-Tanaka (MT), and Method of
Cells (MOC) have been used for evaluating the properties of FGMs.
In all these models, the macroscopic material point of a heterogeneous composite material is dened by a representative volume
element (RVE) consisting of the constituents of the composite,
which is a statistical representation of the microstructure in the
neighborhood of the RVE. The average properties of RVE represent
the overall properties of the homogenized composites. The FGM
was represented as a piece-wise layered material with uniform
effective properties. The homogeneous macroscopic properties in
each layer were evaluated using the micromechanical model. The
self-consistent scheme and Mori-Tanaka method have been widely
used by the authors to analyze the behavior of FGM, e.g., Zhai et al.
[32,33], Reiter et al. [24], Tsukamoto [29] and Zhang et al. [34]. A
detailed review of the micromechanical-modeling approaches
used by the researchers to study the behavior of FGM can also be
found in Gasik [7]. Buryachenko and Rammerstorfer [3] developed
a micromechanical model based on the multiparticle effective eld
method (MEFM), which constitutes the Green function and the theory of function of random variables to evaluate the thermo-elastic
responses of FGM. Aboudi et al. [2] developed a generalized higher
order micromechanical theory using MOC. The coupling effects at
the micro and macro levels were considered to analyze the thermo-mechanical behavior of the FGM graded in three directions.
Another approach to idealize the graded microstructure of FGM
is done by dening RVE in which functional spatial variations of
the inclusions are assumed to analyze the thermo-mechanical
behavior of the FGM. Grujicic and Zhang [9] used the Voronoi cell
FE method (VCFEM), and Yin et al. [31] introduced pair wise particle interactions and Eshelbys equivalent inclusion solution to evaluate the effective thermo-elastic properties of graded RVE.
Recently, Fang and Hu [6] formulated the effective thermal conductivity of functionally graded brous composite using the nonFourier heat conduction equation. Reiter et al. [24] and Dao et al.
[4] performed thermo-elastic FE analysis of FGM models. The
FGM consists of an idealized geometry of inclusions graded linearly
was considered. The results were also computed of FGM represented by a piece-wise layered model. They concluded that microstructural details should be considered to predict the residual
a) Functionally graded
material idealization
b) Microstructure
details
stresses at the interface of the inclusion and matrix. Muliana [14]

presented a micromechanical model for predicting effective thermal properties and viscoelastic responses of FGM. A simplied
micromechanical model of particle reinforced composites is used
to obtain effective properties at each material point in the FGM.
Available micromechanical and FE based studies on FGM are limited to thermo-elastic behaviors. There is clearly a need to understand the non-linear thermo-viscoelastic behavior of FGMs with
non-constant properties of the constituents.
The present study addresses coupled heat conduction and
deformation of viscoelastic FGM using a micromechanical-modeling approach. A through-thickness continuous variation of the
thermal and mechanical properties of the FGM is modeled as an
assembly of homogeneous layers. Average thermo-mechanical
properties in each homogeneous medium are dened using previously developed micromechanical models for particle reinforced
composites of Muliana and Kim [15], and Khan and Muliana [12].
This micromechanical model consists of two constituents, inclusions and matrix. Both inclusions and matrix constituents are assumed to have timestresstemperature dependent moduli. The
thermal properties (coefcient of thermal expansion and thermal
conductivity) of each constituent are allowed to vary with temperature. Sequentially coupled heat transfer and displacement analyses are performed, which allow analyzing stress/strain behaviors of
FGM having timetemperature dependent properties. Experimental data available in the literature are used to verify the model.
Numerical simulations are also performed to analyze non-linear
thermo-viscoelastic responses of homogenized FGM using a micromechanical model and comparisons are made with the results obtained from FE analysis of two dimensional (2D) FGM models
having microstructural details (i.e., heterogeneous FGM). In the
2D FE model of heterogeneous FGM, the inclusions are idealized
as circles and their volume fraction vary from one end of geometric
model to the other. Comparisons of results show that the present
micromechanical model is capable of predicting the non-linear viscoelastic responses of FGMs.
2. Modeling of functionally graded material

In this study, a FGM consisting of two constituents whose
material properties change with time, stress, and temperature, is
considered. The FGM graded in one direction is idealized as a
piece-wise homogeneous medium whose macroscopic properties
are evaluated using a micromechanical model. The variation in
properties from a series of homogeneous layers along the graded
direction is shown in Fig. 1(a). The FGM is approximated as an
assembly of a ctitious layered medium to facilitate the process
of integration of micromechanical model with FE package, i.e.,
ABAQUS [1]. The variations in material properties with locations
c) Microstructure
Idealization
d) RVE
Fig. 1. Illustration of modeling approach for FGM using a micromechanical model.
e) Unit Cell
513
are incorporated by introducing the multiple integration points

along the graded directions, i.e., thickness direction. Each integration point represents a material property of a discretized area.
Thus, the overall through-thickness FGM responses show zigzag
(discontinuous) variations. By increasing the number of integration
points along the graded directions, the discontinuities in the FGM
can be minimized. It is also possible to analytically solve the equations that govern the conduction of heat and deformation in the
FGM body. The developed micromechanical model of particle reinforced composites can be used as material parameters in the governing equations.
Each layer of the FGM is composed of a homogeneous matrix
and spherical inclusion particles. The spherical particles are assumed to be uniformly distributed in each macroscopic layer.
The particles in the microstructure are idealized as cubes and distributed uniformly in three dimensional periodic arrays. A representative volume element (RVE) consisting of one particle
embedded in cubic matrix is considered. Due to the three plane
symmetry, one-eight unit-cell is assumed to consist of four sub
cells. The rst subcell contains a particle constituent, while subcells 2, 3, and 4 represent the matrix constituents, as shown in
Fig. 1(e). Perfect bonds are assumed at the subcells interfaces. Periodic boundary conditions are imposed on the RVE. Micromechanical relations are formulated in terms of incremental average eld
quantities, i.e., stress, strain, heat ux and temperature gradient,
in the subcells. Stress, temperature and time dependent constitutive models are used for the isotropic constituents. Temperature
dependent thermal properties are considered for particle and matrix constituents. The effective properties of the unit-cell dene the
macroscopic properties at a material point in the homogeneous
layer which in turn represents the effective response of each layer.
The present micromechanical model is compatible with general
displacement based FE software to perform the thermo-mechanical analyses of FGM structures.
2.1. Constitutive model for isotropic constituents
At the constituent level, the modied viscoelastic constitutive
model of Schapery [25] is used for each subcell. The generalized
three dimensional constitutive equations with stress and temperature dependent behavior for non-aging materials can be written as:
etij eij t g 0 r t ; T t Sijkl 0rtkl g 1 r t ; T t
DSijkl wt ws
Z
0
dn
arn aT
ws ws
Z s
0
dn
arn aT
where utj
@T t
;
@xj
where qti and utj are the heat ux and temperature gradient. K tij is
the temperature dependent thermal conductivity. K tij is also called
the consistent tangent thermal conductivity, which varies with
temperature at current time t.
2.2. Effective thermo-viscoelastic properties
By satisfying the displacement and traction continuities at the
interfaces during thermo-viscoelastic deformations, the expression
for the effective time dependent stiffness matrix and coefcient of
thermal expansion are formulated. This formulation leads to an
effective timetemperaturestress-dependent coefcient of thermal expansion. The effective thermal conductivity is formulated
by imposing heat ux and temperature continuities at the subcell
interfaces.
The method of volume averaging is used to evaluate the effective response of a unit-cell (micromechanical model). The average
stresses and strains are dened by:
N Z
N
1X
1X
a a
a
r
xk dV a
V a rij and
ij
V a1 V a
V a1
N Z
N
1X
1X
a
eij
eija xka dV a
V a eij ;
V a1 V a
V a1
r ij
4
5
where an over bar indicates average material quantities. The superscript a denotes the subcells number and N is the number of suba
a
cells. Stress rij and strain eij are the average stress and strain in
each subcell. The unit-cell volume V is:
N
X
V a ;
N 4:
a1
The superscript t indicates a variable at time t. Sijkl 0 are the components of the instantaneous elastic compliance, DSijkl wt ws are
the components of the transient compliances, and aij are the components of coefcient of thermal expansion (CTE) tensor. The
parameters T t and T 0 are the current and reference temperatures.
The linear coefcient of thermal expansion, a, also varies with temperatures. w is the reduced-time (effective time) given by:
wt wt
qti K tij utj ;

Z t
s ; T s rskl
d g 2 r
dDT s
aij T t
ds:
ds
dsds
0
for solving the stress tensor using the incremental strains. An iterative scheme is included in order to minimize residual from the linearization. A detailed discussion about the non-linear parameters
and recursive algorithm formulation of Eq. (1) can be found in
Haj-Ali and Muliana [10] and Muliana and Khan [16]. An outline
of the recursiveiterative algorithm for the nonlinear isotropic viscoelastic material is given Appendix A.
The conduction mode of heat transfer is considered. The Fourier
law of heat conduction with temperature dependent thermal conductivity is used and can be expressed as:
where g 0 ; g 1 ; g 2 in Eq. (1) and a in Eq. (2) are nonlinear parameters

and dened as functions of current temperature T t and effective
t . For isotropic materials, the total strain can be separated
stress r
into deviatoric and volumetric parts. A recursiveiterative method
developed by Muliana and Khan [16] is used to solve the deviatoric
and volumetric components of the mechanical strains. The formulation is derived with a constant incremental strain rate during each
time increment, which is compatible with a displacement based FE
analysis. Linearized trial stress tensors are used as starting points
For non-linear stressstrain constitutive relations, the total average

of the stresses and strains at current time t are solved incrementally, which are:
r tij r ijtDt Dr tij

Dt
etij et
Detij
ij
7
8
The superscript t Dt represents the converged eld quantities

from the previous steps, which are stored as history variables. Detij
tij are the incremental average strains and stresses of the
and Dr
unit-cell at current time. The unit-cell length scale is assumed much
smaller compared to the macroscopic geometry of FGM, so that the
total temperature in each unit-cell of a material point is dened by:
T t;a T tDt;a DT t;a
and DT t;a DT t
The micromechanical model is designed to be compatible with displacement based FE software, in which the incremental strains are
chosen as independent variable. We introduce a strain interaction
t;a
matrix Bt;a , which relates the subcell average strains, Deij , to
t

the unit-cell average strain, Deij , and it is written as:
a
a;t
Det;
Bijkl Detkl
ij
10
514
Using the incremental strains in Eq. (10), constitutive equation for

each subcell, and volume average of the incremental stress, the scalar components of the incremental effective stress are:
The average heat ux equation for a homogeneous composite medium is expressed by the Fourier law of heat conduction as:
N
h
i
1X
a;t
a;t
a;t
tij
Dr
V a C ijkl Bklrs Detrs akl DT t
V a1

tkl DT t
C tijkl Detkl a
ti K tij u
tj
q
11
From the above equation, the effective tangent stiffness matrix, C tijrs ,
is:
C tijrs
N
1X
a;t a;t
V a C ijkl Bklrs ;
V a1
12
a;t
where Bklrs is a scalar component of the fourth order interaction

tensor, which can be obtained by satisfying the micromechanical
relations and the constitutive equations. The micromechanical relations within the four subcells are derived by assuming perfect
bonding at the interfaces of the subcells and imposing displacement
compatibility and traction continuity at the subcell interfaces. Dea;t
tailed formulations of Bklrs and C tijrs are described in Muliana and
Kim [15]. From Eq. (11), the effective coefcient of linear thermal
expansion is given as:
a tij
1;t N
X
C ijkl
a;t
a;t
V a C klmn amn
;
13
a tij a t dij
1;t h
i
C ijkl
A;t
4;t
3 3;t 3;t
C klmn amn V 4 C klmn a4;t
V A C klmn aA;t
mn V
mn ;
V
14
where the total volume of subcells 1 and 2 in Eq. (14) is

A;t
A;t
V A V 1 V 2 , and where aij and C ijkl in Eq. (14) are the effective thermal expansion and stiffness expressions for subcells 1 and
A;t
A;t
2. The scalar components of aij and C ijkl can be expressed in the
following equations:
aA;t
aA;t dij
ij
X 1;t
ijkl
where
X tijkl
i
1 h 1 1;t
2 2;t
; and
V
a
V
a
ij
ij
VA
15
16
1 h
VA
11;t
C ijkl
21;t
C ijkl
17
The effective CTE in Eq. (14) depends on the moduli and CTE of each
constituent. Thus, for the stress, temperature and time dependent
constituent mechanical and thermal properties, the effective CTE
also varies with stress, temperature and time. A detailed formulation for the effective coefcient of thermal expansion can be found
in Appendix B.
2.3. Effective thermal conductivity
A volume averaging method based on spatial variation of the
temperature gradient in each subcell is adopted to determine the
effective thermal conductivity of particle reinforced composites.
The average heat ux and temperature gradient are:
N Z
N

1X
1X
a
a
a
qi xk dV a
V a qi ; and
V a1 V a
V a1
N Z
N

1X
1X
a
i
u
uia xka dV a
V a u i :
V a1 V a
V a1
i
q
dT t
:
dx j
20
It is noted that the components of the conductivity tensor, K tij ,

vary with temperature as the thermal conductivity for each constituent is allowed to vary with temperature. The micromechanical
relations within the four subcells in Fig. 1(e) are derived by assuming perfect bonding at the interfaces of the subcells. Using the heat
conduction equation for each subcell and the effective heat ux
relation, the tangent effective thermal conductivity matrix of the
composite can be expressed as:
K tik
4
1X
a;t
a;t
V a K ij Mjk ;
V a1
21
where, the Ma;t matrix is the concentration tensor that relates the
average subcell temperature gradient with the overall temperature
gradient across the unit-cell. A detailed formulation of Ma;t tensor
and the effective thermal conductivity can be found in Appendix B.
In the following sections, Eqs. (12), (14) and (21) are used to
determine the variations of effective time dependent stiffness matrix, coefcient of linear thermal expansion, and thermal conductivity along the graded direction of FGMs.
a1
The previously developed micromechanical relations for the effective viscoelastic response of particle reinforced composite (Muliana
and Kim [15]) is modied to determine the effective CTE. Using the
micromechanical relations and thermo-viscoelastic constitutive
relations for the particle and matrix subcells, the effective CTE for
the isotropic nonlinear responses can be expressed as:
A;t
C ijkl
tj
where u
18
19
3. Numerical simulations and discussion

A micromechanical model of particulate composite having time,
temperature and stress dependent material properties at the constituent level is utilized for predicting the thermo-mechanical
behavior of FGM. To demonstrate the capability of the proposed
micromechanical model, the thermo-mechanical responses of
FGM from the micromechanical model are compared with ones
from existing experimental data, and FE model of FGMs having
microstructural details.
3.1. Effective thermo-mechanical properties
The experimental data of Zhai et al. [32,33] is used to compare
the variations of elastic modulus in the FGMs. TiC=Ni3 Al FGMs
were prepared with Ni3 Al particles dispersed in the continuous
TiC matrix. The elastic properties of the constituents are shown
in Table 1. Comparisons of the predicted elastic modulus distribution along the gradation direction with the experimental data are
shown in Fig. 2. The proposed model provides relatively good
agreement with the experimental data. The responses are characterized at xed temperatures.
The FGMs consisting of metal-matrix systems are widely used
in high temperature applications. The metallic components having
a high coefcient of thermal expansion (CTE) are generally doped
with constituents having a low CTE to tailor the overall CTE. Thus,
the FGM can be used in applications requiring low CTE and high
thermal conductivity. The CTE is one of the main features of FGM
needed to be analyzed when designing FGM for high temperature
applications. Geiger and Jackson [8] measured the CTE distribution
in Al/Si FGM with a volume fraction of Si particles that varies from
0% to 40%. The material properties of Al and Si are given in Table 1.
Comparisons of the effective CTE obtained from the micromechanical model with the ones obtained from the experimental data are
shown in Fig. 3. The results are found to be in good agreement with
the experimental data. Geiger and Jackson [8] also measured the
thermal conductivity of the Al-6061(T6)/SiC FGM with a SiC particle volume fraction varies from 0% to 60%. The material properties
of Al-6061(T6) and SiC are given in Table 1. Fig. 4 shows the comparison of experimental data and the results obtained from the
515

Table 1
Mechanical and physical properties of materials used in FGM.
Material
Young modulus (E), GPa
Poisson ratio t
Linear thermal expansion a 106 ; 1=K
Thermal conductivity (K), W/m/K.
Ni3 Al
TiC
Al-6061 (T6)
SiC
Si
Al
199
460
70.3
400
112.4
72
0.295
0.19
0.34
0.20
0.42
0.33
11.90
7.20
23.40
3.4
3.0
23.6
173
120
100
234
500
200
Effective Thermal Conductivity (W/m.K)
Effective Young Modulus (GPa)
TiC filled with Ni 3 Al Particles
400
300
200
Experimental Data Zhai et. al. (1993)
180
Km = Thermal Conductivity of Al
160
140
120
Experimental Data Geiger and Jackson (1989)
Micromechanical Model
Present Model
100
0
0.2
0.4
0.6
0.8
100
0
Volume Fraction (VF)
0.1
0.2
0.3
0.4
0.5
0.6
Fig. 2. Comparison of Youngs modulus for FGM consisting of TiC and Ni3 Al.
micromechanical model for the thermal conductivity variations

along the graded direction. Comparing the experimental data with
the results obtained from the micromechanical model, as shown in
Figs. 24, it is suggested that the proposed micromechanical model
is capable of predicting the thermo-mechanical behaviors of FGM
along the graded direction.
3.2. Sequentially coupled heat conduction and deformations of FGM
Experimental data of FGMs having timetemperature dependent behavior is currently not available. FE analysis is performed
to determine the effects of timetemperature dependent constituent properties on the thermo-viscoelastic behaviors of FGMs. All
3E-005
Aluminum with Silicon inclusions
-1
Coefficient of Thermal Expansion (K )
Kp/Km = 120:173
Aluminum Filled with Silicon Carbide Particles

Kp = Thermal Conductivity of SiC
2E-005
Fig. 4. Comparison of the thermal conductivity of FGM consisting of Al and SiC

particles.
simulations are performed using the ABAQUS FE software. The results obtained from the micromechanical model are compared
with the ones from FE analysis of the FGM model having microstructural details. The FGM panel of 16 mm length 10 mm
height 1 mm depth is studied. The volume fraction of inclusions
varies along the length direction. A 2D FE model of FGM having a
gradation of the particle in one direction is shown in Fig. 5(a)
and (b). The particles in the form of circles are dispersed randomly
with a gradient of volume fractions of particle from 0% to 40%. In
Fig. 5(a), large diameter particles are distributed while Fig. 5(b)
contains small size particles. Small size particles show more uniform distribution as compare to the ones with large size particle.
Fig. 5(c) illustrates the simplied piece-wise homogenized model
with sixteen (16) layers. Each layer represents the macroscopic
material point with homogeneous properties varying with the gradient of volume fraction of the particles. The heat transfer thermal
analysis is rst performed to obtain the temperature distribution
along the graded direction. Using the temperature distribution,
the stress analyses are carried out to determine the timetemperature dependent deformations of FGM along the graded directions.
In order to obtain the temperature proles, the equation governing
the heat conduction in an FGM body needs to be solved. This equation is written as:

1E-005
qcxk T qi;i i; k 1; 2; 3
Experimental Data Geiger and Jackson (1989)
0.1
0.2
0.3
0.4
0.5

Fig. 3. Comparison of the coefcient of thermal expansion for FGM consisting of Al
with Si inclusions.
22
where qcxk is the effective heat capacity that depends on the composition, density, and specic heat of the two constituents in the
FGM body. The effective heat capacity is obtained using a volume
average method.
The FGM consisting of Ti6Al4V and ZrO2 is rst considered.
The temperature dependent mechanical and physical properties
of these materials are given in Table 2. The properties are taken
from Praveen et al. [21]. First, a transient heat transfer analysis is
516
(a)
(b)
(c)
Fig. 5. Illustration of the geometry of the nite element models for a volume fraction that varies from 0% to 40%. (a) Coarse and (b) ne microstructural details; (c) piece-wise
homogeneous macroscopic layers.
1200
Ti6Al-4v with Zr0 2 inclusions
Coarse Microstructural Model

Fine Microstructural Model
1000
Temperature (K)
performed by applying the uniform temperature of 1000 K at one

end. The entire FGM is initially at constant temperature of 300 K.
After 159 s, the temperature distributions reach to steady state
condition. Temperature prole is plotted along the graded direction. The results obtained from our micromechanical formulation
are compared with the ones obtained from FE model having coarse
and ne microstructural details. Comparison of the results in Fig. 6
shows that proles obtained from ne microstructural details and
micromechanical model are in good agreement, while results obtained from coarse microstructural details show some deviations.
The deviation is mainly due to uneven and sparse distribution of
inclusions. In all further analyses, FE model having ne microstructural details will be considered. The computational (CPU) time taken to analyze the FE model of FGM having ne microstructural
details is 118 s which is about 13 times higher than the one taken
by the analysis using our micromechanical formulation.
Next, the stress analyses are performed based on the temperature distribution obtained from the heat conduction analyses. A
uniaxial stress of 10 MPa is applied along the graded direction.
The temperature distribution obtained from transient heat transfer
analysis is considered as a eld dependent temperature loading.
The elastic properties of the constituents change with temperature
as shown in Table 2. Fig. 7 shows the proles of the displacement
elds. The results obtained from FE analysis of FGM models having
ne microstructural details and the ones using a micromechanical
model are in good agreement. In the above analyses, the effect of
temperature on the deformation is incorporated through the temperature dependent elastic properties while the effect of thermal
expansion is neglected. However, with such a high temperature
changes, the effects of free thermal expansion of each constituents
can be very signicant. Mismatches in the thermal expansion coefcient of the constituents can generate thermal stresses. Fig. 8
shows the results of variations of displacement along the graded
direction with temperature dependent thermal expansion and
elastic properties. The FE model with microstructural details incorporates thermal stresses due to mismatches in the thermal expansion coefcients. At the beginning of the heat transfer analysis,
there is a high rate of change of the temperature gradient which
causes generation of high thermal stresses at the interfaces of
the constituents. These stress elds in turn affect the displacement
elds of the entire FGM. The thermal stress effect is currently not
being included in the present micromechanical model, which is
t =15s
800
t =10s
t =5s
600
t =3s
t =1s
400
Volume fraction varies from 0 to 40%
with maximum at distance 16 mm.
Steady State Time = 159 seconds
200
0
12
16
Distance (mm)
Fig. 6. Temperature proles at different times along the graded direction of FGM
having constituents with temperature dependent thermal conductivities, for Ti
6Al4V/ZrO2.
shown by the deviation in the two responses in Fig. 8. As time progresses, the temperature gradient decreases which reduces thermal stresses and for a zero temperature gradient both results
agree quite well.
Next, coupled heat conduction and deformation analysis is performed to investigate the effect of viscoelastic constituents on the
overall thermo-mechanical responses of FGMs. The time-dependent behavior of metal-matrix composites is of importance at high
temperatures. The creep behavior of Al with silicon carbide inclusions is thus numerically studied as an example. The temperature
dependent elastic modulus of aluminum is taken from Kaufman
[11]. The properties of SiC are taken from Geiger and Jackson [8].
The temperature dependent mechanical and physical properties
of Al and SiC are given in Table 3. The present model requires
the creep parameters which can be obtained from a series of experimental data performed at constant stress and different temperatures. Because of the unavailability of such data, creep properties
of aluminum at 573 K and 28.5 MPa are taken from the experimental work of Tjong and Ma [28]. The time dependent and non-linear
temperature dependent parameters of Al are given in Table 4.
Table 2
Temperature dependent mechanical and physical properties of materials of Ti6Al4V and ZrO2 .
Property
Young modulus (E), Pa
Poisson ratio t
Coefcient of thermal expansion a 106 ; 1=K
Thermal conductivity (K), W/m/K
Specic heat (C), J/kg K
Density q kg=m3
T is temperature in K.
Ti6Al4V
11
Zirconia ZrO2
6
1:23 10 56:457 10 T
0.3
7:58 106 4:927 109 T 2:388 1012 T 2
1.2095 + 0.01686T
625:2969 0:264T 4:49 104 T 2
4429
2:44 1011 334:28 106 T 295:24 103 T 2 89:79T 3

0.3
1:28 105 19:07 109 T 1:28 1011 T 2 8:67 1017 T 3
1:7 2:17 104 T 1:13 105 T 2
487:3427 0:149T 2:94 105 T 2
5700
517
Table 4
Prony series coefcients and non-linear temperature dependent parameters for Al.
0.0008
Ti6Al-4v with Zr02 inclusions
Displacement (mm)
0.0006
Detail Microstructural Model

t =50s
0.0004
kn s1
Dn 106 MPa1
1
2
3
4
5
6
1
101
102
103
104
105
0.1
0.15
20
30
160
1100

2
0
g T0 exp 0:36 TT
;
T0
t =40s
g T1 g T2 aT 1
t =30s
0.0002
t =20s
t =5s
0
0
12
16
Distance (mm)
Fig. 7. Variations of displacement eld at different times along the graded direction
of FGM having constituents with temperature dependent elastic properties, for Ti
6Al4V/ZrO2.
0.2
Displacement (mm)
0.15
Ti6Al-4v with Zr0 2 inclusions
t =159s
t =50s
0.1
t =40s
t =30s
0.05
t =20s
0
0
12
16
Distance (mm)
Fig. 8. Variations of displacement eld at different times along the graded direction
of FGM having constituents with temperature dependent elastic and thermal
properties, for Ti6Al4V/ZrO2.
Moreover, DiCarlo and Yun [5] reported that the SiC does not show
any creep up to 1073 K. This temperature is far above the temperatures considered in this study. Therefore, SiC is assumed to behave linearly elastic. The axial creep data of pure aluminum is
shown in Fig. 9. The axial creep of composites and FGMs consisting
of aluminum and SiC is computed using the micromechanical model. A constant stress of 28.5 MPa is applied at one end followed by
holding for 12,000 s at a constant temperature of 573 K. The
boundary conditions of the specimens are shown in Fig. 9. Composites having a uniform distribution and linear gradation of SiC along
the graded direction are considered. The results of axial creep

deformations (measured at point B) in Fig. 9 show that, the graded
material shows more creep resistance than the composites having
uniform distribution of the SiC particles.
A sequentially coupled analysis of Al/SiC FGM is then performed
to analyze its timetemperature dependent behavior. The Al is assumed to have timetemperature dependent properties while SiC
behaves linearly elastic. Initially the entire FGM is assumed to be
at 300 K. Transient heat transfer analysis is performed by applying
a constant temperature of 573 K at one end. The analysis reached
steady state heat transfer conditions after 14.8 s. Fig. 10 shows
the temperature prole at different times along the graded direction. The results obtained from heat transfer analyses of FGM models having ne microstructural details and the ones using
micromechanical model are in good agreement. Next, the stress
analyses are performed based on the temperature distribution obtained from the thermal analyses. A constant stress of 28.5 MPa is
applied in the graded direction. The stress is held constant for up to
2000 s. Creep deformations at different times are plotted along the
graded direction. Though, steady state time is reached after 14.8 s,
because of the presence of viscoelastic Al, the deformation continues to grow under a constant stress of 28.5 MPa at a temperature of
573 K. Fig. 11 shows that the results obtained from FE analysis of
FGM models having ne microstructural details and the ones using
a micromechanical model are in good agreement. The comparisons
of these results are strong evidence that the present micromechanical model is capable of predicting non-linear viscoelastic behavior
of FGM with a reasonable level of uncertainty 10%.
4. Conclusion
The coupled thermo-viscoelastic analysis of FGM is performed
using a micromechanical-modeling approach. The proposed model
has a capability to analyze the heat conduction and thermo-viscoelastic deformations of FGMs having timetemperature and stress
dependent eld dependent properties. When the gradients of temperature in the FGM are substantially large, the effect of thermal
stresses due to the mismatch in the coefcient of thermal expansions of the constituents on the overall mechanical responses of
FGM is signicant. The thermal stresses are localized at the interface between inclusion and matrix, which can potentially cause
debonding. In the present micromechanical model, the effect of
Table 3
Temperature dependent mechanical and physical properties of materials of Al and SiC.
Property
Aluminum (Al)
Silicon carbide (SiC)
Young modulus (E), MPa

Poisson ratio t
Thermal conductivity (K), W/m/K
Specic heat (C), J/kg K
Density q; kg=m3
Coefcient of thermal expansion a 106 ; 1=K
65144 73:432T 0:1618T 2

0.33
235 0:0305T 0:0003T 2 6E 07T 3 3E 10T 4
900
2700
2 105 6 109 T 3 1012 T 2 1 1014 T 3
406,783 22.61T
0.2
183.78 0.1569T
750
3210
3 106 3 109 T 6 1013 T 2
T is temperature in K.
518
ature applications can be obtained by proper distribution of the

particles along the graded direction of the composites. The present
micromechanical-modeling approach is computationally efcient
and quite accurate in predicting time-dependent responses of
FGMs.
0.01
Pure Aluminum Experimental Creep Data ( Tjong and Ma,1999)
Axial Creep Deformation (mm)
25% inclusions by volume graded linearly with a maximum of 40%.
0.008
25% inclusions by volume distributed uniformly

x
0.006
Acknowledgements
A
This research is sponsored by the National Science Foundation

(NSF) under Grant No. 0546528 and the Air Force Ofce of Scientic Research (AFOSR) under Grant number FA9550-09-1-0145.
0.004
T(K)
(MPa)
T= 573
0.002
=28.5
Appendix A
Al with SiC inclusions
12000
12000
0
0
4000
8000
12000
Time (seconds)
Fig. 9. Comparison of axial creep deformations for Al, Al/SiC composite and FGM.
1
3
1
1
t ; T t J0 Stij g 1 r
t; Tt
etij g 0 r
2
2
t =3s
500
t =2s
t =0.5s
400
t =0.1s
Volume fraction varies

from 0 to 40% with
maximum at distance
16 mm.
Steady State Time = 14.8 seconds
300
4
12
16
Distance (mm)
J 0 21 tD0
DJ
Fig. 10. Temperature proles at different times along the graded direction of FGM
having constituents with temperature dependent thermal conductivities, for Al/SiC.
0.04
wt
=28.5
t
A
t =2000s
0.02
t =1000s
t =350s
0.01
t =14.8s
t =0.5s
0
0
DBw 31 2tDDw
A:2
Here D0 and DD are the instantaneous elastic and transient compliances under uniaxial (extensional) creep loading. The uniaxial transient compliance, DD, is expressed in terms of Prony series as:
DDw
N
X

Dn 1 expkn wt
A:3
n1
0.03
2000
B0 31 2tD0
21 tDDw
Steady State Time = 14.8 seconds
(MPa)
h
i
s ; T s Ssij
d g 2 r
where eM;t
and eT;t
ij
ij are the total mechanical and thermal strains,
respectively. The superscript t indicates a variable at time t. The
parameters J0 and B0 are the instantaneous elastic shear and bulk
compliances, respectively. The terms DJ and DB are the time-dependent shear and bulk compliances, respectively. The corresponding
linear elastic Poissons ratio, t, is assumed to be time independent,
which allows expressing the shear and bulk compliances as:
t =1s
DJwt ws
ds
ds

Z t
s
s
; T rskk
d g r
1
1
etkk g 0 r t ; T t B0 rtkk g 1 r t ; T t DBwt ws 2
ds
ds
3
3
0
A:1
Temperature (K)
1
3
etij etij etkk dij aT t T 0 dij

t
eM;t
etij etkk dij ; eT;t
ij
ij aT T 0 dij
600
Axial Creep Deformation (mm)
This appendix outline the basic concepts of solving the Eq. (1)
using the recursiveiterative algorithm. For isotropic materials,
the total strain in Eq. (1) can be written as:
12
16
Distance (mm)
Fig. 11. Variations of axial creep deformations of FGM having constituents with
temperature dependent elastic and thermal properties, for Al/SiC.
stress concentration is not incorporated, which is shown to be a

limitation of the current model. Based on creep analysis of Al/SiC,
it is concluded that better creep resistant material for high temper-
A recursiveiterative method developed by Muliana and Khan [16]

is used to solve the deviatoric and volumetric components of the
mechanical strains in Eq. (A.1). Outline of the recursiveiterative
algorithm for the nonlinear isotropic viscoelastic material is given
in Fig. A1.
tDt
Dt
The parameters qij;n
and qt
kk;n are hereditary integral (history
state variables) stored from the last converged step at time
t Dt. The parameters qtij;n and qtkk;n ; n 1 . . . N are the hereditary
integrals at current time for every term in the Prony series in the
form of deviatoric and volumetric strains. The superscript tr
means trial value of that variable. The g b b 0; 1; 2 represent
the nonlinear parameters g 0 ; g 1 and g 2 , given in Eq. (1). The
parameters J t and Bt are the effective shear and bulk compliances
at the current time, respectively. The initial approximation (trial)
incremental stress tensor is determined using the trial nonlinear
parameters, incremental deviatoric stress tensor, DSt;tr
ij , and volu. The trial current stress tensor is formed
metric stress tensor, Drt;tr
kk
based on the given variables and history variables from the previous converged step. An iterative scheme is then employed to nd
the correct stress tensor for a given strain tensor. The correct stress
tensor at current time is solved by minimizing a residual tensor,
519
Fig. A1. Recursiveiterative algorithm for the nonlinear isotropic viscoelastic material.
given in 3.3 of Fig. A1. Finally, the consistent tangent stiffness matrix is dened by taking the inverse of the partial derivative of the
incremental strain with respect to the incremental stress at the end
of the current time step. The consistent tangent stiffness, C tijkl , at
the converged state, are:
C tijkl
@drtij
M;t
kl
@de
"
@Rtij
@drtkl
#1
;
Rtij
! 0
A:5
Eq. (A.5) denes material properties at current time t for each subcell
in the micromechanical model. The components of the consistent
tangent stiffness tensor vary with time, temperature, and stress.
embedded in a cubic matrix (Fig. B1). Periodic boundary conditions

(BCs) are imposed to the RVE. After solving boundary value problems (BVPs) for the RVE with the prescribed periodic BCs, the effective eld quantities and material properties are obtained using the
volume averaging relations. Detailed description about the periodic structure can be found in Nemat-Nasser and Hori [18].
Let xi be the local coordinate system of the RVE. Let ai denote
the dimensions of the cube in local coordinate system, i.e.,
xi i 1; 2; 3 that can be reduced to ai a. In this study, we assumed that the component of displacement uxi at any pointxk inside
the RVE can be written as follows:
x;
uxi xk ui
Appendix B
This appendix outlines the basic concepts require in formulating the effective thermal properties of the homogenized composite
medium having spherical particle inclusions. The detailed micromechanical formulations are presented in the manuscript Khan
and Muliana [12] which is currently under review.
B.1. Periodic boundary conditions
To reduce complexity in the micromechanical formulation, a
composite having a periodic microstructure is assumed. A representative volume element (RVE) is then dened by a single particle

Xi;i xk x
~xi xk
u
u
k
B:1
The superscript x is for local eld quantities while overbar and X is

x;0
is the component of displacefor global eld quantities. Where ui
ment at a reference point in the ith direction, which can be picked
~ xi xk is the component of the displacearbitrarily, e.g., point (*); u
ment inside the RVE due to the prescribed macroscopic boundary
condition. In this study, due to the nature of the prescribed boundary condition, i.e., prescribed displacement on the boundary, the
~ i xk vanishes. Thus, the periodic boundary convolume integral of u
ditions are given by:

a
Xi;i x0k xaki
uxi x0k uxi xki u
txi x0k
a
t xi xki
txi
x
ij nj
B:2
B:3
520
mined for every subcell. This requires forming 24 equations. The

rst sets of equations are determined from the strain compatibility
equations which are given as:
x2
a1
(*)
f Re g AM
1
a2
121
8 1 9
e >
>
>
>
> 2 >
i <
e =
1224
x1
>
e3 >
>
>
>
;
: 4 >
x3
Fig. B1. Representative volume element of the periodic microstructure.
where t xi represents the traction on the surfaces of the RVE, with nj

a
being the corresponding unit normal vectors. xok and xki are the coordinates of arbitrary points on faces xi 0 and xi ai .
For heat conduction equation, the similar procedure is adopted
and the total temperature eld T x at any point xk inside RVE can be
written as follows:

T x xk T x; T i;i xk xk Te xk
B:4
This form of expansion of the temperature eld gives a similar set of

periodic boundary conditions for temperature and heat ux, as previously obtained for displacement eld and traction.
a
0
i
T x x0k T x xki T X
i;i xk xk
B:5
a
qxi xki ni
B:6
B.2. Formulation of effective coefcient of thermal expansion
The micromechanical relations within the four subcells in

Fig. 1(e) are derived by assuming perfect bond along the interfaces
of the subcells and imposing displacement compatibility and traction continuity at the subcells interface. The homogenized incremental strain relations for the particle reinforced composites are
given as:
1;t
ij
V de
V
V 1 V 2
for i j; i; j 1; 2; 3
2;t
ij
de
3;t
ij
de
4;t
ij
de
h
fRr g AM;t
2
121
8 1 9
>
>
>e >
=
< e2 >
i>
>
e3 >
>
>
>
;
: 4 >
1224
The residual vector Rr results from satisfying traction continuity

relations. For linear elastic constituents, the components of Rr are
zero. The matrix O is the zero matrix and the components of matrix
M;t
and DM
AM
1 are given as follows:
1 ; A2
V 1
I
V A
6 33
6
6
6 0
0
6
M
A1 6 33 33
6 0
0
6 33 33
6
4
0
V 1 I
A;t
drij
d tij
1;t
drij
1;t
ij
r dr
3;t
4;t
2;t
drij
dr
2;t
ij
33
AM;t
2
C1
ax
6 33
6
6
6 0
6 33
6
6
6 0
6 33
6
6
4
0
33
33
33
33
33
33
33
33
V 2 I
33
33
33
33
V 3 I
33
7
7
7
0 7
33 7
7
0 7
7
33 7
5
V 4 I
33
33
33
33
33
33
33
C2
ax
33
33
C
Csh
33
33
33
1
Csh
33
33
Csh
33
33
1
sh
33
33
33
33
33
33
33
33
33
3
C sh
33
33
33
33
33
0
0
33
for i j
33 7
7
7
0 7
33 7
7
7
0 7
7
33 7
7
4 5
C sh
33
B:14
where:
V 4 drij
33
B:13
The homogenized stresses are written as:
V 3 drij
V 2
I
V A
33
33
for ij
B:8
A;t
B:12
126 61
33
tij V 1 V 2 drij
dr
O feg
241
B:7
2;t
3;t
4;t
tij V 1 dc1;t
dc
V 2 dcij V 3 dcij V 4 dcij
ij
B:11
61
where Re is the residual vector arising from imposing strain compatibility relations. In the case of linear elastic responses are exhibited for all subcells, the vector Re is zero. The second sets of
equations are formed based on traction continuity relations. The
equations based on the traction continuity relations within
subcells:
33
dtij
126
241
a3
qxi x0k ni
h i
DM
feg
1
B:9
C 1111
C 1122
C 3311
C 2222
C 3322
6
Cax 4 C 2211
C 1133
C 1212
7
6
C 2233 5 Csh 4 0
C 3333
0
0
C 1212
0
7
0
5
C 1212
B:15
dr
3;t
ij
4;t
ij
dr
for ij
B:10
Using the thermo-viscoelastic constitutive relations for the particle

and matrix subcells, volume averaging schemes for the incremental
stress and strain, and micromechanical relations in Eqs. (B.7)
(B.10), the effective CTE is obtained. The effective CTE in Eq. (14) requires dening the effective tangent stiffness matrix. Formulations
of the effective tangent stiffness matrix, which also require formulating the strain interaction matrix Ba;t , are given in Muliana and
Kim [15].
In order to formulate the strain interaction matrix Ba;t , introduced in Eq. (10), the micromechanical relations and the constitutive equations are imposed. The micromechanical models consists
of four subcells with six components of strains need to be deter-
33
DM
1
6
6 I
6 33
6
6
6 I
6 33
4
0
33
33
7
0 7
7
7
0 7
33 7
5
I
33 7
B:16
33
a;t
The B
matrices in Eq. (10) are then formed using Eqs. (B.11) and
(B.12), which in linearized relations are.
i
Ba;t
24x6
"
AM
1
AM;t
2
24x24
1
"
DM
1
O
24x6
B:17
Once the Ba;t matrices are determined, the effective homogenized

stresses and stiffness matrix can be solved using Eqs. (11) and (12),
respectively.
B.3. Formulation of effective thermal conductivity
The homogenized temperature gradient and heat ux relations
for unit-cell are summarized as follows:
1 h
ti
du
ti
dq
A;t
2;t
v 1 du1;t
v 2 dui
i
3;t
dui
i
1 h A A;t
4;t
v dqi v 3 dq3;t
v 4 dqi
i
V
1;t
dqi
dqi
2;t
dqi
4;t
dui
B:18
B:19
B:20
where the total volume of subcells 1 and 2 in Eqs. (B.18) and (B.19)
2
is V A V 1 V : .
We introduce a concentration tensor that relates the average
subcells temperature gradient with the overall temperature gradient across the unit-cell. Let Ma;t be the concentration tensor of the
temperature gradient. The temperature gradient in each subcell is
expressed by:
a;t
dui
a;t
tj
Mij du
B:21
a;t
To formulate the M
matrix, the micromechanical relations and
the constitutive equations are imposed. The present micromodel
consists of four (4) subcells with three (3) components of heat ux
need to be determined for every subcell. This requires forming
twelve (12) equations based on the temperature and heat ux continuities at the interface of each subcell which are written as:
8
1;t 9
dui >
>
>
>
>
>
>
=
< du2;t >
i
ti g
A1
D1 fdu
3;t
>
> du
912 >
93
31
>
>
>
i
>
>
:
4;t ;
dui
B:22
121
8
1;t 9
dui >
>
>
>
>
>
>
>
=
t < du2;t
i
ti g
A2
O fdu
3;t >
>
13
d
u
31
>
112 >
>
>
i
>
>
:
4;t ;
dui
B:23
121
By substituting Eq. (B.21) to Eqs. (B.22) and (B.23), the Ma;t matrix
can be determined, which is:
h
i A1 D
1
1
Ma;t
At2
O
121
124
B:24
41
The matrix O is the zero matrix and the components of matrix A1 ; At2
and D1 are given as follows:
3
2
V 1
I VV A I
0
0
V A
33
33 7
6 33
33
7
6
7
6
A1 6 0
0
I
0 7
6 33 33 33 33 7
5
4
0
0
0
I
33
33
33
33
h 1;t
i
2;t
I
0
0
At2 K33I K
33
33
33
3
2
I
6 33 7
6 I 7
6 33 7
7
6
D1 6
7
6 I 7
6 33 7
5
4
I
2
33
B:25
B:26
B:27
521
References
[1] ABAQUS, Hibbitt, Karlsson and Sorensen Inc., Users Manual, Version, 6.5;
2005.
[2] Aboudi J, Pindera MJ, Arnold SM. Higher-order theory for functionally graded
materials. Composites B 1999(30):777832.
[3] Buryachenko VA, Rammerstorfer FG. Local effective-thermoelastic properties of
graded random structure matrix composites. Arch Appl Mech 2001;71:24972.
[4] Dao M, Gu P, Maewal A, Asaro RJ. A micromechal study of residual stress in
functionally graded materials. Acta Mater 1997;45(8):326576.
[5] DiCarlo JA, Yun HM. Non-oxide (silicon carbide) bers, handbook of ceramic
composites. US: Springer; 2005.
[6] Fang XQ, Hu C. Dynamic effective thermal properties of functionally graded
brous composites using non-Fourier heat conduction. Comput Mater Sci
2008;42:194202.
[7] Gasik MM. Micromechanical modeling of functionally graded materials.
Comput Mater Sci 1998;13:4255.
[8] Geiger AL, Jackson M. Low expansion metal-matrix composites for electronic
packaging. Microelectronic packaging technology: materials and processes. In:
Proceedings of the 2nd ASM international electronic materials and processing
congress, Philadelphia, PA, USA, vol. 442447; 1989. p. 93102.
[9] Grujicic G, Zhang Y. Determination of effective elastic properties of
functionally graded materials using Voronoi cell nite element method.
Mater Sci Eng A 1998;251:6476.
[10] Haj-Ali RM, Muliana AH. A multi-scale constitutive framework for the
nonlinear analysis of laminated composite materials and structures. Int J
Solids Struct 2004;41:346190.
[11] Kaufman JG. Properties of aluminum alloys: tensile, creep, and fatigue data at
high and low temperatures. ASM International; 2000.
[12] Khan KA, Muliana AH. Effective thermal properties of viscoelastic composites
having eld-dependent constituent properties. Acta Mech 2009. doi:10.1007/
s00707-009-0171-6.
[13] Mukherjee S, Paulino GH. The elastic-viscoelastic correspondence principle for
functionally graded materials. J Appl Mech 2003;70:35963. Revisited.
[14] Muliana AH. A micromechanical model for predicting thermal properties and
thermo-viscoelastic responses of functionally graded materials. Int J Solids
Struct 2009;46:191125.
[15] Muliana AH, Kim JS. A concurrent micromechanical model for nonlinear
viscoelastic behaviors of particle reinforced composites. Int J Solids Struct
2007;44(21):6891913.
[16] Muliana AH, Khan KA. A time-integration algorithm for thermo-rheologically
complex polymers. Comput Mater Sci 2008;41(4):57688.
[17] Noda N. Thermal stresses in materials with temperature dependent properties.
Appl Mech Rev 1991;44:38397.
[18] Nemat-Nasser S, Hori M. Micromechanics: overall properties of heterogeneous
materials. 2nd ed. Elsevier; 1999.
[19] Parameswaran V, Shukla A. Processing and characterization of a model
functionally graded material. J Mater Sci 2000;33:219.
[20] Paulino GH, Jin ZH. Correspondence principle in viscoelastic functionally
graded materials. ASME J Appl Mech 2001;68:12932.
[21] Praveen GN, Chin CD, Reddy JN. Thermoelastic analysis of functionally graded
ceramic-metal cylinder. J Eng Mech 1999;125(11):125967.
[22] Reddy JN, Chin CD. Thermomechanical analysis of functionally graded
cylinders and plates. J Therm Stress 1998;21:593626.
[23] Reddy JN. Analysis of functionally graded plates. Int J Numer Methods Eng
2000;47:66384.
[24] Reiter T, Dvorak GJ, Tvergaard V. Micromechanical models for graded
composite materials. J Mech Phys Solids 1997;45(8):1281302.
[25] Schapery RA. On the characterization of nonlinear viscoelastic materials.
Polym Eng Sci 1969;9(4):295310.
[26] Shabana YM, Noda N. Numerical evaluation of the thermomechanical effective
properties of a functionally graded material using the homogenization
method. Int J Solid Struct 2008;45:3494506.
[27] Tanigawa Y. Some basic thermoelastic problems for nonhomogeneous
structural materials. Appl Mech Rev 1995;48:287300.
[28] Tjong SC, Ma ZY. High-temperature creep behaviour of powder-metallurgy
aluminum composites reinforced with SiC particles of various sizes. Compos
Sci Technol 1999;59:111725.
[29] Tsukamoto H. Analytical Method of inelastic thermal stresses in a functionally
graded material plate by a combination of micro and macromechanical
approaches. Composites B 2003;34:5618.
[30] Yang YY. Time dependent stress analysis in functionally graded materials. Int J
Solid Struct 2000;37:7593608.
[31] Yin HM, Paulino GH, Buttlar WG, Sun LZ. Micromechanics-based thermoelastic
model for functionally graded particulate materials with particle interactions. J
Mech Phys Solids 2007;55:13260.
[32] Zhai PC, Zhang QJ, Jiang CR. Micromechanical design of ceramic/metal
functionally gradient composite materials. In: Proceeding of ICCM-9, July
1216, 1993, Madrid, Spain, vol. 442447. p. 95102.
[33] Zhai PC, Jiang CR, Zhang QJ. Application of three-phase micro-mechanical
theories to ceramic/metal functionally gradient materials. Ceram Trans
1993;34:449.
[34] Zhang M, Zhai PC, Zhang QC. Effective thermal conductivity of functionally
graded composite with arbitrary geometry of particulate. Key Eng Mater
2005;297-300:15228.

A Multi-Scale Model For Coupled Heat Conduction and Deformations of Viscoelastic Functionally Graded Materials

Enviado por

Dados do documento

Título original

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

A Multi-Scale Model For Coupled Heat Conduction and Deformations of Viscoelastic Functionally Graded Materials

Enviado por

Direitos autorais:

Formatos disponíveis

Composites: Part B 40 (2009) 511521

Contents lists available at ScienceDirect

A multi-scale model for coupled heat conduction and deformations

mechanical properties of the constituents are allowed to change

K.A. Khan, A.H. Muliana / Composites: Part B 40 (2009) 511521

of the graded microstructure with some idealized geometry is

stresses at the interface of the inclusion and matrix. Muliana [14]

2. Modeling of functionally graded material

Fig. 1. Illustration of modeling approach for FGM using a micromechanical model.

K.A. Khan, A.H. Muliana / Composites: Part B 40 (2009) 511521

are incorporated by introducing the multiple integration points

etij eij t g 0 r t ; T t Sijkl 0rtkl g 1 r t ; T t

qti K tij utj ;

where g 0 ; g 1 ; g 2 in Eq. (1) and a in Eq. (2) are nonlinear parameters

For non-linear stressstrain constitutive relations, the total average

r tij r ijtDt Dr tij

The superscript t Dt represents the converged eld quantities

T t;a T tDt;a DT t;a

K.A. Khan, A.H. Muliana / Composites: Part B 40 (2009) 511521

Using the incremental strains in Eq. (10), constitutive equation for

where Bklrs is a scalar component of the fourth order interaction

where the total volume of subcells 1 and 2 in Eq. (14) is

It is noted that the components of the conductivity tensor, K tij ,

3. Numerical simulations and discussion

K.A. Khan, A.H. Muliana / Composites: Part B 40 (2009) 511521

Young modulus (E), GPa

Linear thermal expansion a 106 ; 1=K

Thermal conductivity (K), W/m/K.

Effective Thermal Conductivity (W/m.K)

Effective Young Modulus (GPa)

TiC filled with Ni 3 Al Particles

Volume Fraction (VF)

Volume Fraction (VF)

micromechanical model for the thermal conductivity variations

Coefficient of Thermal Expansion (K )

Aluminum Filled with Silicon Carbide Particles

Fig. 4. Comparison of the thermal conductivity of FGM consisting of Al and SiC

Volume Fraction (VF)

K.A. Khan, A.H. Muliana / Composites: Part B 40 (2009) 511521

Coarse Microstructural Model

performed by applying the uniform temperature of 1000 K at one

Steady State Time = 159 seconds

2:44 1011 334:28 106 T 295:24 103 T 2 89:79T 3

K.A. Khan, A.H. Muliana / Composites: Part B 40 (2009) 511521

Ti6Al-4v with Zr02 inclusions

Detail Microstructural Model

Steady State Time = 159 seconds

the graded direction are considered. The results of axial creep

Silicon carbide (SiC)

Young modulus (E), MPa

65144 73:432T 0:1618T 2

K.A. Khan, A.H. Muliana / Composites: Part B 40 (2009) 511521

ature applications can be obtained by proper distribution of the

Axial Creep Deformation (mm)

25% inclusions by volume graded linearly with a maximum of 40%.

25% inclusions by volume distributed uniformly

This research is sponsored by the National Science Foundation

Al with SiC inclusions

Al with SiC inclusions

Detail Microstructural Model

Volume fraction varies

Steady State Time = 14.8 seconds

Detail Microstructural Model