See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/336376084

A finite volume method for an analysis of transient heat

conduction in functionally graded materials

Conference Paper · October 2019

CITATIONS READS

0 404

3 authors, including:

Passakorn Vessakosol J. Charoensuk

Prince of Songkla University King Mongkut's Institute of Technology Ladkrabang
17 PUBLICATIONS 115 CITATIONS 51 PUBLICATIONS 307 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Biomass combustion in pulverised form View project

hydrogen production View project

All content following this page was uploaded by J. Charoensuk on 10 October 2019.

The user has requested enhancement of the downloaded file.

 Copyright © 2009 ICCES Proceedings of ICCES’09, pp.4-19

A finite volume method for an analysis of transient heat

conduction in functionally graded materials
P. Vessakosol1 , J. Chareonsuk12 , T. Teschareon1

Summary
For a non-homogeneous medium, the governing equation for heat conduction is
nonlinear in general which difficult to be solved by analytical approach. In the
present work, we propose a numerical approach based on finite volume method
for transient temperature analysis in functionally graded material (FGM), a new
class of composite material, with a simple exponential variation of thermal non-
homogeneities in one coordinate. Numerical calculations are carried out for a
FGM strip and a FGM hollow circular cylinder. The effect of the material non-
homogeneity parameter on the temperature response is discussed.
Keywords: Transient temperature; Functionally graded materials; Finite vol-
ume method; Heat conduction; Time-stepping technique

Introduction
Composite materials have turn out to be preferable materials for developing effi-
cient machine components in several industrial areas including aeronautics, aerospace,
civil and mechanical engineering etc. A composite material may consists of two or
more constituents that possess different attributes. For instance, ceramic based
composite, usually made of ceramic and metal, has been developed for thermal
protection component of space vehicles. Ceramic has excellent properties of high
thermal resistance and corrosion resistance whereas metal can give high mechani-
cal strength and fracture toughness. Under very high temperature and large thermal
gradients, thermal stresses due to the abrupt transition from one material to another
material in conventional layered composites can lead to structural failure like de-
lamination and cracking. These problems can be reduced by properly smooth out
the transition of material properties corresponding to functionally graded material
(FGM). FGM is a new generation of composites where the volume fraction of its
constituents varies gradually, giving a non-uniform microstructure in material with
continuously graded properties. As a result, there is a growing interest in devel-
oping a comprehensive understanding of the behavior of FGMs for microstructural
architecture and design of the overall geometry. A comprehensive review of de-
sign, processing, and modeling as well as applications can be found in the books
by Suresh and Mortensen [1] and by Miyamoto et al. [2].
1 Department of Mechanical Engineering, Faculty of Engineering, King Mongkut’s Institute of

Technology Ladkrabang, Bangkok 10520, Thailand

2 Corresponding Author; Tel.: 662 3264197, Fax.: 662 3264198, Email: [email protected]
A finite volume method 5

Since FGMs are frequently used for structures under thermal load, it is indis-
pensable to analyze their thermal properties. In the past few years, transient heat
conduction behavior in FGMs has been investigated using analytical and numerical
approach. Most of analytical investigations determined the heat conduction solu-
tions by assuming the exponential variation of material constants with Cartesian
coordinates under stationary thermal loading conditions ([3-7]). Although analyti-
cal approach can provide closed-form solutions to the partial differential equation
that governs the transient heat conduction problems in FGMs, it is restricted to
numerous assumptions such as simple geometries (e.g. plate, cylinder or sphere),
certain types of gradation of material properties (e.g. exponential or power-law
distribution), specific types of boundary conditions (e.g. isothermal or adiabatic)
and special loading cases (e.g. uniform or sinusoidal loading). Hence, the solution
of heat conduction differential equation in FGMs requires numerical approach due
to the high mathematical complexity. The well known numerical methods includ-
ing the finite element method (FEM) and boundary element method (BEM) have
been used. In the former method, Wang et al. [8] derived a finite element for-
mulation for determining the one-dimensional transient temperature field in FGMs
plate, cylinder and sphere. They also gave the expressions for the displacement
and thermal stresses in hollow and solid cylinders after the temperature field was
determined. Wang and Tian [9] extended the work of Wang et al. [8] for solv-
ing the two-dimensional transient heat conduction considering FGM plate under a
sudden internal heat generation. In previous FEM papers, FGM, a kind of non-
homogeneous material, was defined by a number of homogeneous elements which
their constant material properties evaluated at their own centroids. This attempt
leads to additional discretization error in the modeling of material property field
because a step-wise constant approximation was employed instead of a continuous
one in each individual element. A few studies in developing new finite element
formulations based on graded material properties were carried out ([10-11]). These
formulations are called graded finite elements by both researchers and only applied
for the plane elasticity problems. Nonetheless, they may be applied for other appli-
cations such as the heat conduction problem.
Secondly, the BEM concept was extensively used to investigate many heat con-
duction problems in graded materials. Gray et al. [12] developed a Galerkin BEM
formulation by deriving the Green’s function for steady state heat conduction prob-
lem in exponentially graded material. Sutradhar et al. [13] extended the work of
Gray et al. [12] to transient heat conduction for exponentially graded materials
in three dimensions using Laplace transform Galerkin boundary element method
(LTBEM). Sladek et al. [14-16] presented a meshless local boundary integral equa-
tion (LBIE) formulation for transient heat conduction considering exponential ma-
6 Copyright © 2009 ICCES Proceedings of ICCES’09, pp.4-19

terial variation. Suthradhar and Paulino [17] proposed a simple three-dimensional

BEM for transient heat conduction in FGMs where the non-homogeneous problem
can be transformed into the homogeneous diffusion problem. By means of vari-
able transform technique, this work can handle three classes of material gradation
(quadratic, exponential and trigonometric) with multi-dimensional variation. How-
ever, in order to solve such problems by the BEM concept, the thermal conductivity
and specific heat of material must have the same form to preserve the constant ther-
mal diffusivity.
Beside the FEM and the BEM, the finite volume method (FVM) is an interest-
ing method for solving non-linear problems. FVM was originally developed from
conservation law of physics. By means of: i) volumetric integration, ii) the use of
divergence theorem and iii) numerical differentiation, the material properties as-
sociated with normal temperature gradient at the control surfaces are incorporated
into the discretized model. These values are obtained in terms of node values rather
than shape function as used in FEM. Hence, the properties at the control surfaces
are related to the space coordinates in a different manner. FVM formulation is also
not restricted to the assumption of constant thermal diffusivity of ideal FGM that
often used in the BEM formulation. The detailed procedures of FVM for analyzing
the transient heat conduction in homogeneous medium as well as computational
fluid dynamics can be found, for instance, in the book by Versteeg and Malalasek-
era [18]. Generally, the material coefficients in standard FEM codes are usually as-
sumed to be constant within each element (step-wise constant material properties)
while, in this paper, the FVM provides the linear-piecewise material properties.
Therefore, the discretization errors in material properties of FVM formulation are
expected to be smaller than the FEM formulation. In this paper, we intend to ap-
ply the FVM concept to analyze the two-dimensional transient heat conduction in
functionally graded composite. The derivation and implementation of a finite vol-
ume method for the solution of simple initial-boundary value problems in FGMs is
given. A commonly used time-stepping scheme known as Crank-Nicolson method
is selected to perform time integration and a central difference scheme is utilized to
evaluate the space integrals. The exponential spatial variation of material constants
in FGMs is chosen to demonstrate the effect of nonhomogeneity on the tempera-
ture solutions. The accuracy and the efficiency of the proposed method are verified
in the following numerical examples by comparing solutions with those obtained
from exact solutions.
The paper is organized as follows. Section 2, basic governing equations are
described. Section 3 gives the numerical implementation of the finite volume/finite
difference method in rectangular and polar coordinate systems for the general two-
dimensional thermal conduction problems. Section 4, some numerical examples
A finite volume method 7

are examined. Finally, conclusions are summarized in section 5.

Governing Equations in rectangular and polar forms

Suppose in the rectangular Cartesian coordinate (x, y), there is a solid occupying
a space Ω which is surrounded by a surface Γ. The temperature inside the solid
may vary from point to point, and from time to time. Let T = T (x, y;t) be the
temperature which is assumed to be a continuous function of the coordinates (x, y)
and time t. The thermal conductivity and the specific heat of the solid material
are also assumed to be two different continuous functions of the coordinates (x, y)
represented by k = k (x, y) and c = c (x, y) respectively. From the basic law of heat
conduction and the conservation of thermal energy, the general differential equation
can be written as
   
∂ ∂T ∂ ∂T ∂T
k + k + Q = ρc (1)
∂x ∂x ∂y ∂y ∂t

where ρ is the density constant, and Q is the internal heat generation per unit vol-
ume.
The thermal conduction equation (1) must be solved for prescribed boundary
and initial conductions. The initial condition specifies the temperature distribution
at time zero. This is,
T (x, y; 0) = T̄0 (x, y). (2)

The frequently encountered boundary conditions are specified temperature and

surface heat flow:
T (x, y;t) = T̄ , on boundary ΓT (3a)
∂ T (x, y;t)
q (x, y;t) = −k (x, y) = q̄, on boundary Γq (3b)
∂n
where Γq + ΓT = Γ, and the bar over the quantity of interest means that it assumes
a prescribed value, n is the coordinate normal to the exterior of surface Γ. Eq. (3a)
indicates that the temperature on the boundary ΓT is known. Eq. (3b) indicates that
on the boundary Γq the thermal flux is specified.
Now, we consider the transient temperature within a solid that may be con-
veniently determined if the solution procedures are performed on the polar coor-
dinates (r, θ ). The temperature is assumed to be a continuous function of spatial
positions (r, θ ) and time t. By using standard calculus technique, the thermal con-
duction equation (1) can be rewritten as
   
1 ∂ ∂T 1 ∂ k ∂T ∂T
rk + + Q = ρc (4)
r ∂r ∂r r ∂θ r ∂θ ∂t
8 Copyright © 2009 ICCES Proceedings of ICCES’09, pp.4-19

where ρ is the density constant; Q is the internal heat generation per unit volume; k
and c are the thermal conductivity and specific heat functions of the spatial position
(r, θ ) respectively.
Similarly, the temperature distribution at time zero is,

T (r, θ ; 0) = T̄0 (r, θ ) . (5)

The boundary conditions may be given as follows;

T (r, θ ;t) = T̄ , on boundary ΓT (6a)

∂ T (r, θ ;t)
q (r, θ ;t) = −k (r, θ ) = q̄, on boundary Γq (6b)
∂n
Likewise, Γq + ΓT = Γ, n is the coordinate normal to the exterior of surface Γ. Eq.
(6a) and Eq. (6b) are specified.

Finite volume/finite difference method

In this section, the FV/FD method is used to determine the transient temperature
response in FGMs. The FV/FD method involves two essential procedures: (a)
using finite volume space discretization to obtain a first-order system of differential
equations, and (b) finding transient response via finite difference technique.

Finite volume formulation

Since the thermal energy is conserved everywhere in the solution domain, the dif-
ferential thermal equation (1) can be integrated over the control volume (Ω) and
over a time interval from t to t + Δt as

 
t+Δt    
t+Δt    
t+Δt
∂ ∂T ∂ ∂T
k dΩdt + k dΩdt + Q dΩdt
∂x ∂x ∂y ∂y
t Ω t Ω t Ω
 
t+Δt
∂T
= ρc dΩdt. (7)
∂t
t Ω

Following two types of coordinate systems given earlier, the present FVM for-
mulation demonstrates two types of control volume for numerical solutions. Re-
ferring to the two-dimensional Cartesian mesh in Fig. 1, a general nodal point is
identified by P and its neighbor in a two-dimensional geometry, the nodes to the
west, east, south and north are identified by W, E, S and N respectively. The four
side faces are labeled w, e, s and n which stand for west, east, south and north. The
A finite volume method 9

width and height of the control volume are Δx and Δy respectively. If the tempera-
ture at a node is assumed to prevail over the whole control volume, the right hand
side can be written as

 t+Δt

∂T
ρc dtdΩ = ρ c (TP − TPo ) ΔxΔy. (8)
∂t
Ω t

Figure 1: two-dimensional Cartesian

mesh
Figure 2: two-dimensional polar mesh
In Eq. (8), superscript ‘o’ refers to temperature at time t; temperature at time
levelt + Δt are not superscripted. After substituting Eq. (8) into Eq. (7) and apply-
ing central differencing to the diffusion terms on the left hand side of Eq. (7), we
have

 
t+Δt   
t+Δt 
TE − TP TP − TW TN − TP TP − TS
ke Δy − kw Δy dt + kn Δx − ks Δx dt
Δx Δx Δy Δy
t t

t+Δt
+ QΔxΔy dt = ρ c (TP − TPo )ΔxΔy. (9)
t

Another type of control volume which is refered to as a two-dimensional polar

coordinate system is shown Fig. 2. Integration of equation (4) over the control
10 Copyright © 2009 ICCES Proceedings of ICCES’09, pp.4-19

volume (Ω) and over a time interval from t to t + Δt gives

 
t+Δt    
t+Δt    
t+Δy
1 ∂ ∂T 1 ∂ k ∂T
rk dΩdt + dΩdt + QdΩdt
r ∂r ∂r r ∂θ r ∂θ
t Ω t Ω t Ω
 
t+Δt
∂T
= ρc dΩdt (10)
∂t
t Ω

With the temperature at a node prevailing over the whole control volume, the
right hand side can be written as

 t+Δt
  
∂T re + rw
ρc dtdΩ = ρ c (TP − TP )
o
ΔrΔθ . (11)
∂t 2
Ω t

Similar technique is used for Eq.(11) and Eq.(10), thus we have

 
t+Δt 
TE − TP TP − TW
ke reΔθ − kw rw Δθ dt
Δr Δr
t
 
t+Δt   
kn TN − TP ks TP − TS re + rw
+ Δr − Δr dt + Q ΔrΔθ
rn Δθ rs Δθ 2
t
 
re + rw
= ρ c (TP − TPo ) ΔrΔθ . (12)
2

Finding transient response via finite difference

The generalized trapezoidal family of methods is used in order to write the integral
IT of temperature TP with respect to time as


t+Δt
IT = TP dt = [β TP + (1 − β ) TPo ] Δt . (13)
t

A weighting factor β is bounded between 0 and 1. If β = 0, the algorithm is the

Euler forward method; if β = 0.5, the algorithm is the Crank-Nicolson; if β = 2/3,
the algorithm is known as the Galerkin method; if β = 1, the algorithm is the Euler
backward method. The time-marching algorithm (Eq. (13)) is unconditionally
stable for β ≥ 0.5. Using Eq. (13) for TN , TS , TE and TW , Eq. (9) may be re-written
A finite volume method 11

as

ρ c (TP − TPo ) ΔxΔy kΔyβ kΔy (1 − β ) o kΔyβ

= (TE − TP )+ (TE − TPo )− (TP − TW )
Δt Δx Δx Δx
kΔy (1 − β ) o kΔxβ kΔx (1 − β ) o
− (TP − TWo ) + (TN − TP ) + (TN − TPo )
Δx Δy Δy
kΔyβ kΔy (1 − β ) o
− (TP − TS ) − (TP − TSo ) + QΔxΔy. (14)
Δx Δx

Eq. (14) is solved with the Crank-Nicolson by settingβ = 0.5. Then, the dis-
cretized unsteady heat conduction equation may be expressed in the compact form
as
       
TE + TEo TW + TWo TN + TNo TS + TSo
aP TP = aE + aW + aN + aS
2 2 2 2
a E aW a N a S
+ aoP − − − − T o + b (15)
2 2 2 2 P
where aP = 12 (aE + aW + aN + aS ) + aoP , (16)

ρ cΔxΔy
aoP = , (16)
Δt
Δy
aE = ke , (17)
Δx
Δy
aW = kw , (18)
Δx
Δx
aN = kn , (19)
Δy
Δx
aS = ks , (20)
Δy
b = QΔxΔy. (21)

In polar coordinate system, after using Eq. (13) for TN , TS , TE and TW , we

substitute them into Eq. (12). The resulting equation is in the same form as Eq.
(15). All coefficients can be summarized as follows:
1
aP = (aE + aW + aN + aS ) + aoP , (22)
2
 
ρ cΔrΔθ re + rw
aP =
o
, (23)
Δt 2
12 Copyright © 2009 ICCES Proceedings of ICCES’09, pp.4-19

Δθ
aE = ke re , (24)
Δr
Δθ
aW = kw rw , (25)
Δr
kn Δr
aN = , (26)
r n Δθ
ks Δr
aS = , (27)
r s Δθ
 
re + rw
b = QΔrΔθ . (28)
2

Numerical examples
To verify the numerical implementation, the following examples are presented:

(1) Finite strip problem

(2) Hollow cylinder problem

The first example is a finite strip with a unidirectional variation of the ther-
mal conductivity and diffusivity. On both opposite sides parallel to the y-axis, two
different temperatures are prescribed. The other sides of the strip are insulated.
Two different exponential parameters of material variation are selected in numer-
ical calculation. The special case of exponential parameter γ = 0 corresponds to
a homogeneous material is used for comparison the numerical and analytical so-
lutions. In the last example, an infinitely long and thick-walled hollow cylinder is
considered. The inner and outer surfaces are kept to different temperatures. Due
to symmetry in geometry and boundary conditions, a quarter of the cross-sectional
area of the hollow cylinder is sufficient for analyzing this problem. Both examples
we use the zero temperature as the initial condition and the time interval (Δt) of 1
sec.

A finite strip problem

Fig. 3 illustrates a square plate with a side-length of a = 0.04 m. The finite volume
mesh consists of sixteen pieces of square control volumes. The left surface of the
plate is maintained at the temperature T = 0, while the right surface (at x = a)
has T = 1. The remaining two surfaces are insulated (zero normal heat flux). In
homogeneous case, analytical solution [19] for the temperature is known as:
 
x 2 ∞ cos(nπ ) nπ x α n2 π 2 t
T (x,t) = + ∑ sin exp − (29)
a π n=1 n a a2
A finite volume method 13

Figure 3: Boundary conditions, control volumes and node locations for a finite strip

 )90[D 
 )90[D 
 )90[D 
7HPSHUDWXUH7

)90[D 

H[DFW[D 
 H[DFW[D 
 H[DFW[D 

 H[DFW[D 





      
7LPHW VHF

Figure 4: Time variation of temperature in a finite strip at four positions along

y/a = 0.375

The temperature comparison between the analytical and the numerical solu-
tions along the x-axis at different time level are depicted in Fig. 4. The numerical
and analytical results are in excellent agreement.
To characterize the response of the FGM, a unidirectional variation of the ther-
mal conductivity (k) and diffusivity (α = k/ρ c) are considered. The same expo-
nential spatial variation is taken

k = k0 eγ x, α = α0 eγ x (30)
14 Copyright © 2009 ICCES Proceedings of ICCES’09, pp.4-19

where k0 = 17 W /(m · deg) , α0 = 0.17×10−4 m2 /s and γ is known as the exponen-

tial parameter. Here γ = 0 corresponds to a homogeneous material with constant
thermal conductivity (k0 ) and diffusivity (α0 ). Two different exponential param-
eters γ = 0.2 and 0.5 cm−1 are chosen in numerical calculation of the FGM and
compared with the results of the homogeneous material. The time history of tem-
perature at the position x/a = 0.125 for three γ -values is shown in Fig. 5. It is
clear that the material nonhomogeneity (i.e. the exponential parameter) has an ef-
fect on the temperature distribution within the plate. Both transient and steady state
temperature levels at the considered position increase as the exponential parameter
increases. The steady-state temperature profiles along x-direction of the domain
are plotted in Fig. 6 for the four material variations. The overall temperature level
at interior positions increases with increasing γ -values.



7HPSHUDWXUH7

 γ=0
γ=0.2
 γ=0.5




        

7LPHW VHF

Figure 5: Time variation of temperature at position x/a = 0.125 of functionally

graded finite strip


7HPSHUDWXUH7

γ=0.0

γ=0.2
 γ=0.5
γ=1.0



     
[D

Figure 6: Temperature distribution along x-direction for a functionally graded finite

strip under steady state condition
A finite volume method 15

A long hollow cylinder problem

The last example is an infinitely long thick-walled hollow cylinder with prescribed
constant temperature on inner and outer surfaces. The problem of interest and
corresponding FVM mesh is shown in Fig. 7, where the inner radii R1 = 0.08 m
and the outer radii R2 = 0.1 m are selected. The temperature is specified along
the inner radius as T = 0 and outer radius as T = 1. All the other surfaces are
insulated. For comparison purpose, a homogeneous material is first considered.
The analytical solution [19] for the homogenous material can be found as
∞
ln (r/R1 ) J 2 (R1 ωn )U0 (rωn )
T (r,t) = −π ∑ 2 0 exp −αωn2 t , (31)
ln (R2 /R1 ) J
n=1 0 (R ω
1 n ) − J0
2 (R ω )
2 n

where
U0 (rωn ) = J0 (rωn )Y0 (R2 ωn ) − J0 (R2 ωn )Y0 (rωn ), (32)
and ωn are roots of the following transcendental equation

J0 (R1 r)Y0 (R2 r) − J0 (R2 r)Y0 (R1 r) = 0, (33)

with J0 (x) and Y0 (x) are Bessel functions of first and second kind and zeroth order.
The thermal conductivity and diffusivity are taken as

k = k0 eγ (r−R1 ) , α = α0 eγ (r−R1 ) , (34)

where k0 = 17 W /(m · deg) , α0 = 0.17×10−4 m2 /s and γ is known as the exponen-

tial parameter. Here γ = 0 corresponds to a homogeneous material with constant
thermal conductivity (k0 ) and diffusivity (α0 ). The time history of temperatures at
the position r/R1 = 1.025 is shown in Fig. 8. Again, the finite volume method
gives good agreement with the analytical solutions. Fig. 9 shows the temperature
distributions along radial direction at four different time levels.
To illustrate the influence of material gradation on the temperature variation,
three different exponential parameters γ = 0.2, 0.4 and 0.6 cm−1 are selected for
thermal conductivity and diffusivity in FGM. Numerical results for time variation
of the temperature in the FGMs and the homogeneous material at a selected position
are shown in Fig. 10. The results shown in Fig. 11 also suggest that the increasing
of exponential parameter of FGM raises the spatial variation of the temperature
response.

Conclusions
The understanding of temperature distribution and related heat fluxes in FGMs
under severe thermal environment is important to several industrial applications.
16 Copyright © 2009 ICCES Proceedings of ICCES’09, pp.4-19

Figure 7: Boundary conditions, control volumes and node locations for a hollow
cylinder



7HPSHUDWXUH7




)90 [

H[DFW


    
7LPHW VHF

Figure 8: Time variation of the temperature at r/R1 = 1.025 in the hollow cylinder
with homogeneous material properties



7HPSHUDWXUH7

W VHF

W VHF
W VHF

W VHF



     
U5
Figure 9: Temperature distribution along radial direction at different time levels for
a hollow cylinder with homogeneous material properties
A finite volume method 17




7HPSHUDWXUH7


γ=0.0

γ=0.2

γ=0.4



      
7LPHW VHF

Figure 10: Time history of temperature at r/R1 = 1.175 in a functionally graded

hollow cylinder


7HPSHUDWXUH7


γ=0.0
 γ=0.2
γ=0.4

γ=0.6

     
U5

Figure 11: Steady state temperature distribution along radial direction for a func-
tionally graded hollow cylinder

Moreover, the simple and effective calculation technique for FGMs is also neces-
sary. In this method, an exponential variation of material properties was used for
modeling the functionally graded material. However, there is no restriction in the
function variation of material properties, as seen in the discretization step. The gov-
erning equation for the nodal temperature values in the time-domain formulation
are given by a system of first-order ordinary differential equations, which can be
solve by a time-stepping technique such as Crank-Nicolson method. The standard
finite volume technique is used to approximate the system of differential equations
with the system of algebraic equations. Good agreements of the numerical results
by using the FVM with the exact results can be achieved in all considered test ex-
amples. In the present work, we have developed two separate codes corresponding
to the Cartesian and Polar meshes. Our primary aim is to introduce an alternative
18 Copyright © 2009 ICCES Proceedings of ICCES’09, pp.4-19

numerical formulation which is easy to implement and to investigate the effects

of non-homogeneous material constants on the characteristics of the FGMs. An
extension of this work includes testing the performance of the unstructured finite
volume formulation for analyzing the thermal heat conduction in FGMs and other
engineering applications of FGMs. This investigation is currently being pursued by
the authors.

References
1. S. Suresh, A. Mortensen, Fundamentals of Functionally Graded Materials,
Maney Publishing, London, 1998.
2. Y. Miyamoto, W.A. Kaysser, B.H. Rabin, A. Kawasaki, R.G. Ford, Function-
ally Graded Materials: Design, Processing and Applications, Kluwer Aca-
demic Publishers, Boston, MA, 1999.
3. Z.H. Jin, N. Noda, An internal crack parallel to the boundary of a nonho-
mogeneous half plane under thermal loading, Int. J. Eng. Sci. 31 (1993)
793-806.
4. N. Noda, Z.H. Jin, Thermal stress intensity factors for a crack in a strip of a
functionally gradient material, Int. J. Solids Struct. 30 (1993) 1039-1056.
5. F. Erdogan, B.H. Wu, Crack problems in FGM layers under thermal stresses,
J. Therm. Stresses 19 (1996) 237-265.
6. Z.H. Jin, R.C. Batra, Stress intensity relaxation at the tip of an edge crack in a
functionally graded material subjected to a thermal shock, J. Therm. Stresses
19 (1996) 317-339.
7. Z.H. Jin, G.H. Paulino, Transient thermal stress analysis of an edge crack in
a functionally graded material, Int. J. Fract. 107 (2001) 73-98.
8. B.L. Wang, Y.W. Mai, X.H. Zhang, Thermal shock resistance of functionally
graded materials, Acta Materialia 52 (2004) 4961-4972.
9. B.L. Wang, Z.H. Tian, Application of finite element-finite difference method
to the determination of transient temperature field in functionally graded ma-
terials, Finite Elements in Analysis and Design 41 (2005) 335-349.
10. M.H. Santare, J. Lambros, Use of graded finite elements to model the be-
havior of nonhomogeneous materials, ASME J. Appl. Mech. 67 (2000)
819-822.
11. J.H., Kim, G.H. Paulino, Isoparametric graded finite elements for nonho-
mogeneous isotropic and orthotropic materials, ASME J. Appl. Mech. 69
(2002) 502-514.
 A finite volume method 19

12. L.J. Gray, T. Kaplan, J.D. Richardson, G.H. Paulino, Green’s functions and
boundary integral analysis for exponentially graded materials: heat conduc-
tion, ASME J. Appl. Mech. 70 (2003) 543-549.
13. A. Sutradhar, G.H. Paulino, L.J. Gray, Transient heat conduction in homo-
geneous and non-homogeneous materials by the Laplace transform Galerkin
boundary element method, Eng. Anal. Bound. Elem. 26 (2002) 119-132.
14. J. Sladek, V. Sladek, Ch. Zhang, Transient heat conduction analysis in func-
tionally graded materials by meshless local boundary integral equation method,
Comput. Mater. Sci. 28 (2003a) 494-504.
15. J. Sladek, V. Sladek, J. Krivacek, Ch. Zhang, Local BIEM for transient heat
conduction analysis in 3-D axisymmetric functionally graded solids, Com-
put. Mech. 32 (2003b) 169-176
16. J. Sladek, V. Sladek, Ch. Zhang, A local BIEM for analysis of transient heat
conduction with nonlinear source terms in FGMs, Eng. Anal. Bound. Elem.
28 (2004) 1-11.
17. A. Sutradhar, G.H. Paulino, The simple boundary element method for tran-
sient heat conduction in functionally graded materials, Comput. Methods
Appl. Mech. Engrg. 193 (2004) 4511-4539.
18. H.K. Versteeg, W. Malalasekera, An introduction to computational fluid dy-
namics The finite volume method, John Wiley & Sons, New York, 1995.
19. H.S. Carslaw, J.C. Jaeger, Conduction of Heat in Solids, Clarendon, Oxford,
1959.

View publication stats