Communications in Applied and Industrial Mathematics, DOI: 10.1685/journal.caim.

482
ISSN 2038-0909, e-482 Research Paper

Fractional heat conduction in a semi-infinite

composite body
Yuriy Povstenko1,2
1
Institute of Mathematics and Computer Science,
Jan Dlugosz University in Czȩstochowa, Poland
j.povstenko@ajd.czest.pl

2
Department of Informatics,
European University of Information Technology and Economics, Warszaw, Poland

Dedicated to Professor Francesco Mainardi

on the occasion of his retirement

Communicated by Gianni Pagnini and Enrico Scalas

Abstract
A medium consisting of a region 0 < x < L and a region L < x < ∞ is con-
sidered. Heat conduction in one region is described by the equation with the Caputo
time-fractional derivative of order α, whereas heat conduction in another region is de-
scribed by the equation with the time derivative of the order β. The problem is solved
under conditions of perfect contact, i.e. when the temperatures at the contact point and
the heat fluxes through the contact point are the same for both regions. The solution
valid for small values of time is expressed in terms of the Mittag-Leffler function and the
Mainardi function. Several particular cases are considered and illustrated graphically.

Keywords: non-Fourier heat conduction, fractional calculus, Mainardi

function, Mittag-Leffler function.

AMS subject classification: 26A33, 35K05, 45K05.

1. Introduction.
The classical theory of heat conduction is based on the Fourier law

(1) q = −k grad T,

where q is the heat flux vector, T denotes temperature, and k is the thermal
conductivity. In combination with the law of conservation of energy, the
standard Fourier law results in the parabolic heat conduction equation.

Received on 01 30, 2014. Accepted on 06 26, 2014. Published on 10 30, 2014.

Licensed under the Creative Commons Attribution NonCommercial NoDerivs 3.0 License.
 Y. Povstenko

The time-nonlocal dependence between the heat flux vector and the
temperature gradient with the “long-tail” power kernel [1]–[3] can be inter-
preted in terms of fractional calculus
1−α
(2) q(t) = −kDRL grad T (t), 0 < α ≤ 1,

(3) q(t) = −kI α−1 grad T (t), 1 < α ≤ 2.

Here I α f (t) and DRL

α f (t) are the Riemann–Liouville fractional integral and

derivative of the order α, respectively, [4]–[7]:

Z t
α 1
(4) I f (t) = (t − τ )α−1 f (τ ) dτ, α > 0,
Γ(α) 0

dm
 Z t 
α 1 m−α−1
(5) DRL f (t) = m (t − τ ) f (τ ) dτ ,
dt Γ(m − α) 0
where m − 1 < α < m.
It should be noted that in fractional calculus, where integrals and deriva-
tives of arbitrary (not integer) order are considered, there is no sharp
boundary between integration and differentiation. For this reason, some
authors [6], [8] do not use a separate notation for the fractional integral
I α f (t). The fractional integral integral I α f (t) of the order α > 0 is denoted
−α
as DRL f (t). Using this notation, Equations (2) and (3) can be rewritten as
one dependence
1−α
(6) q(t) = −kDRL grad T (t), 0 < α ≤ 2.

In combination with the law of conservation of energy, the constitutive

equation (6) leads to the time fractional heat conduction equation
∂αT
(7) = a ∆T, 0 < α ≤ 2,
∂tα
with the Caputo fractional derivative
Z t m
∂ α f (t) 1 m−α−1 d f (τ )
(8) = (t − τ ) dτ, m − 1 < α < m.
∂tα Γ(m − α) 0 dτ m

Starting from the pioneering papers [9]–[13], considerable interest has

been shown in solutions to Equation (7). Different kinds of boundary condi-
tions for time-fractional heat conduction equation were analized in [14], [15].

2
 DOI: 10.1685/journal.caim.482

If the surfaces of two solids are in perfect thermal contact, the temperatures
on the contact surface and the heat fluxes through the contact surface are
the same for both solids, and we obtain the boundary conditions of the
fourth kind:
 
(9) T1  = T2  ,
 
S S

 
1−α ∂T1  1−β ∂T2 
 
(10) k1 DRL = k2 DRL  , 0 < α ≤ 2, 0 < β ≤ 2,
∂n ∂n 


S S

where subscripts 1 and 2 refer to solids 1 and 2, respectively, and n is the

common normal at the contact surface.
Classical heat conduction in composite medium was considered by many
authors (see, for example, [16]–[20]). In the previous papers [15], [21], the
problem of fractional heat conduction in two semi-infinite regions, x > 0 and
x < 0, was investigated. In [22], the central-symmetric problem was studied
for a composite medium consisting of a spherical inclusion 0 < r < R
and a matrix R < r < ∞. In the present paper, we consider a composite
medium consisting of a region 0 < x < L and a region L < x < ∞. Heat
conduction in one region is described by the equation with the Caputo
time-fractional derivative of order α, whereas heat conduction in another
region is described by the equation with the time derivative of the order β.
The problem for uniform initial temperature in a layer 0 < x < L and zero
initial temperature in a region L < x < ∞ is solved under the conditions
of perfect contact at x = L and the insulation condition at the boundary
surface x = 0. The solution is expressed in terms of the Mittag-Leffler
function and the Mainardi function.

2. Mathematical preliminaries.
Recall the Laplace transform rules for fractional integrals and deriva-
tives [5]–[7]:
1 ∗
(11) L {I α f (t)} = f (s),
sα

L {DαRL f (t)} = sα f ∗ (s)

(12) m−1
X
− Dk I m−α f (0+ )sm−1−k , m − 1 < α < m,
k=0

3
 Y. Povstenko

m−1
dα f (t)
  X
α ∗
(13) L = s f (s) − f (k) (0+ )sα−1−k , m − 1 < α < m,
dtα
k=0

where s is the Laplace transform variable, the asterisk denotes the trans-
form.
The Mittag-Leffler function in one parameter α [5]–[7]
∞
X zk
(14) Eα (z) = , α > 0, z ∈ C,
Γ(αk + 1)
k=0

provides a generalization of the exponential function.

The Mittag-Leffler type function in two parameters α and β [5]–[7] is
described by the following series representation:
∞
X zk
(15) Eα,β (z) = , α > 0, β > 0, z ∈ C.
Γ(αk + β)
k=0

The essential role of the Mittag-Leffler functions in fractional calculus

results from the formula for the inverse Laplace transform [6]
 α−β 
−1 s
(16) L = tβ−1 Eα,β (−btα ).
sα + b

The Wright function is defined as [6], [7], [12], [13]

∞
X zk
(17) W (α, β; z) = , α > −1, z ∈ C.
k! Γ(αk + β)
k=0

The Mainardi function M (α; z) [6], [12], [13] is the particular case of
the Wright function

M (α; z) = W (−α, 1 − α; −z)

(18) ∞
X (−1)k z k
= , 0 < α < 1, z ∈ C.
k! Γ[−αk + (1 − α)]
k=0

The Mainardi and Wright functions appear in the formulae for the in-
verse Laplace transform (see [12], [13], [23]–[26])

αλ
L−1 {exp (−λsα )} = M α; λt−α ,


(19) α+1
0 < α < 1, λ > 0,
t

4
 DOI: 10.1685/journal.caim.482

L−1 sα−1 exp (−λsα ) = t−α M α; λt−α ,



(20) 0 < α < 1, λ > 0,

n o
−1 −β α
L s exp (−λs )
(21)
= tβ−1 W −α, β; −λt−α ,


0 < α < 1, λ > 0.

3. Statement of the problem.

Consider the time-fractional heat conduction equations with the Caputo
derivative in a two-layer medium composed of a region 0 < x < L and
a region L < x < ∞:
∂ α T1 ∂ 2 T1
(22) = a1 + Φ1 (x, t), 0 < x < L, 0 < α ≤ 2,
∂tα ∂x2

∂ β T2 ∂ 2 T2
(23) = a 2 + Φ2 (x, t), L < x < ∞, 0 < β ≤ 2,
∂tβ ∂x2
under the initial conditions
(24) t=0: T1 = f1 (x), 0 < x < L, 0 < α ≤ 2,

∂T1
(25) t=0: = F1 (x), 0 < x < L, 1 < α ≤ 2,
∂t
(26) t=0: T2 = f2 (x), 0 < x < ∞, 0 < β ≤ 2,

∂T2
(27) t=0: = F2 (x), 0 < x < ∞, 1 < β ≤ 2,
∂t
and the boundary conditions of perfect thermal contact
(28) x = L : T1 (x, t) = T2 (x, t),

1−α ∂T1 (x, t) 1−β ∂T2 (x, t)

(29) x = L : k1 DRL = k1 DRL ,
∂x ∂x
0 < α ≤ 2, 0 < β ≤ 2.
The boundary surface x = 0 is kept insulated:
∂T (x, t)
(30) x=0: = 0.
∂x
In addition, the boundedness condition at infinity is assumed
(31) lim T (x, t) = 0.
x→∞

5
 Y. Povstenko

4. Uniform initial temperature in the layer.

In this case we consider the time-fractional heat conduction equations
∂ α T1 ∂ 2 T1
(32) = a1 , 0 < x < L, 0 < α ≤ 2,
∂tα ∂x2

∂ β T2 ∂ 2 T2
(33) = a 2 , L < x < ∞, 0 < β ≤ 2,
∂tβ ∂x2
under the initial conditions
(34) t = 0 : T1 = T0 , 0 < x < L, 0 < α ≤ 2,

∂T1
(35) t=0: = 0, 0 < x < L, 1 < α ≤ 2,
∂t
(36) t=0: T2 = 0, 0 < x < ∞, 0 < β ≤ 2,

∂T2
(37) t=0: = 0, 0 < x < ∞, 1 < β ≤ 2,
∂t
and the boundary conditions (28)–(30).
The Laplace transform with respect to time gives two ordinary differ-
ential equations
d2 T1∗
(38) sα T1∗ − sα−1 T0 = a1 , 0 < x < L,
dx2

d2 T2∗
(39) sβ T2∗ = a2 , L < x < ∞,
dx2
and the boundary conditions
dT1∗
(40) x=0: = 0,
dx
(41) x = L : T1∗ = T2∗ ,

dT1∗ dT ∗
(42) x = L : k1 s1−α = k2 s1−β 2 ,
dx dx
(43) x → ∞ : T2∗ = 0.
The solutions of equations (38) and (39) have the following form:
s 
r α  α/2
T 0 s s
(44) T1∗ = + A sinh x + B cosh  x , 0 < x < L,
s a1 a1

6
 DOI: 10.1685/journal.caim.482

 s  s 
sβ sβ
(45) T2∗ = C exp − x + D exp  x , L < x < ∞.
a2 a2

From conditions (40) and (43) it follows that

(46) B = 0, D = 0,

wheras the conditions of the perfect thermal contact (42) and (43) give
T0 1
(47) A=− q  q ,
s cosh sα
L + γsβ/2−α/2 sinh sα
L
a1 a1

s 
T0 sβ
C= exp  L
s a2

(48) q  q 
sα sβ
cosh a1 L
exp a2 L
T0
− q  q ,
s cosh sα
L + γsβ/2−α/2 sinh sα
L
a1 a1

where √
k1 a2
γ= √ .
k2 a1
Hence,
q 
sα
T0 T0 cosh a1 x
(49) T1∗ = − q  q 
s s cosh sα
L + γs β/2−α/2 sinh sα
L
a1 a1

 s 
T0 sβ
T2∗ = exp − (x − L)
s a2

(50) q  q  
sα sβ
cosh a1 L
exp − a2 (x − L)
T0
− q  q .
s cosh sα
L + γsβ/2−α/2 sinh sα
L
a1 a1

Now we will investigate the approximate solution of the considered prob-

lem for small values of time. In the case of classical heat conduction equation

7
 Y. Povstenko

this method was described in [16], [18]. Based on the Tauberian theorems
for the Laplace transform, for small valies of time t (the large values of the
transform variable s) we can neglect the exponential term in comparison
with 1:
 r α 
s
(51) 1 ± exp −2 L ' 1,
a1

thus obtaining
 h q i
sα
T0  exp − a1 (L − x) 
(52) T1∗ ' 1− ,
s  1 + γsβ/2−α/2 

 s 
T0 γsβ/2−α/2 sβ
(53) T2∗ ' exp − (x − L) .
s 1 + γsβ/2−α/2 a2

Inverting the Laplace transform, we get:

a) β > α
t
(t − τ )β/2−1
 
T0 α L−x
Z
T1 (x, t) ' T0 − M ;√
γ 0 τ α/2 2 a1 τ α/2
(54)  
1 β/2−α/2
×Eβ/2−α/2 , β/2 − (t − τ ) dτ, 0 ≤ x ≤ L,
γ

t
(t − τ )α/2−1
 
β x−L
Z
T2 (x, t) ' T0 M ;√
0 τ β/2 2 a2 τ β/2
(55)  
1 β/2−α/2
×Eβ/2−α/2 , α/2 − (t − τ ) dτ, L ≤ x < ∞.
γ

b) α > β

Zt
(t − τ )α/2−1
 
α L−x
T1 (x, t) ' T0 − T0 M ;√
(56) τ α/2 2 a1 τ α/2
0
h i
×Eα/2−β/2 , α/2 −γ (t − τ )α/2−β/2 dτ, 0 ≤ x ≤ L,

8
 DOI: 10.1685/journal.caim.482

1.0
✁✕
✁
✁
✁
0.8 α=2
❩
❩
⑦
❩

0.6 α=β=1
✁
T̄ ✁
✁
0.4 ✁
✁☛ β=1
✁
✁
0.2 ✁
✁
☛✁

0.0
0.0 0.5 1.0 1.5 2.0
x̄

Figure 1. Dependence of the solution in a semi-infinite composite body on distance.

Zt
(t − τ )α/2−1
 
β x−L
T2 (x, t) ' T0 γ M ;√
(57) τ β/2 2 a2 τ β/2
0
h i
×Eα/2−β/2 , α/2 −γ (t − τ )α/2−β/2 dτ, L ≤ x < ∞.

For example, for α = 2, β = 1

 √
T0 ,


0 ≤ x < L − a1 t,

 ( s !)
   
 L − x L − x
(58) T1 ' T0 1 − exp γ 2 t − √ erfc γ t − √ ,

 a1 a1


√


L − a1 t < x < L ;


    
x−L x−L 2
T2 ' T0 erfc √ − exp γ √ +γ t
2 a2 t a2
(59)
√
 
x−L
×erfc √ +γ t , L < x < ∞.
2 a2 t

9
 Y. Povstenko

1.0
✚❃
✚
✚
✚
0.8 ✚
α=1
❩❩
⑥
❩
❩
0.6 β=2

T̄ ❃
✚
✚
✚
0.4 ✚
✚
α=β=1
β=2
0.2
✂
✂
✂
0.0 ✌✂✂
0.0 0.5 1.0 1.5 2.0
x̄

Figure 2. Dependence of the solution in a semi-infinite composite body on distance.

For α = 1, β = 2
    
L−x L−x t
T1 ' T0 erf √ + exp √ + 2
2 a1 t γ a1 γ
(60)  √ 
L−x t
× erfc √ + , 0 ≤ x < L,
2 a1 t γ

    s !
 1 x − L 1 x − L

T0 exp 2 t − √ erfc t− √ ,
γ a2 γ a2



(61) T2 ' √

 L < x < L + a2 t ,

√


0, L + a1 t < x < ∞ .


In particular, if α = β, then
 
T0 α L−x
(62) T1 ' T0 − W − , 1; − √ α/2 , 0 ≤ x ≤ L,
1+γ 2 a1 t

 
T0 γ α x−L
(63) T2 ' W − , 1; − √ α/2 , L ≤ x < ∞.
1+γ 2 a2 t

10
 DOI: 10.1685/journal.caim.482

1.0
✣
✡ α = 0.1
✡ ✟
✒ ✟
α=2 ✙✟
✟ α=2
0.8
α=1
✑✑
✸
✠
✑ α = 1.75
✑
0.6 ✂✍✂ ✑
α = 1.75 ✂ ✑
✰
✑
✂ α=1
T
α=2 ✑
✑
✰
✑
0.4
α = 0.1
✡
✡
0.2 ✡
✢

0.0
0.0 0.5 1.0 1.5 2.0
x̄

Figure 3. Dependence of the solution in a semi-infinite composite body on distance.

Several results of numerical calculations are presented in Figures 1–3

with the following nondimensional quantities:
√ √
T x a1 tα/2 a1
T = , x̄ = , κ= , γ̄ = γtα/2−β/2 ,  = √ tα/2−β/2 .
T0 L L a2

In calculations we have taken κ = 0.2, γ̄ = 2, and  = 0.6. Such values of

nondimensional parameters show the typical features of the solution.

5. Conclusions.
We investigated the solution to the time-fractional heat conduction
equations with different orders of the Caputo time derivatives in a com-
posite medium consisting of two regions being in perfect thermal contact.
In the case 0 < α < 1, the time-fractional heat conduction equation in-
terpolates the elliptic Helmholtz equation (α → 0) and the parabolic heat
conduction equation (α = 1). When 1 < α < 2, the time-fractional heat
conduction equation interpolates the standard heat conduction equation
(α = 1) and the hyperbolic wave equation (α = 2). The Laplace trans-
form with respect to time t reduces the problem to two ordinaty differen-
tial equations which solutions have been analyzed for large values of the
Laplace transform variable. Based on the Tauberian theorems for Laplace
transorm, the approximate solution valid for small values of time was ob-

11
 Y. Povstenko

tained. The numerical results for several values of the order of fractional
derivatives show the dependence of the solution on distance.

REFERENCES
1. Y. Povstenko, Fractional heat conduction equation and associated ther-
mal stresses, Journal of Thermal Stresses, vol. 28, pp. 83–102, 2005.
2. Y. Povstenko, Thermoelasticity which uses fractional heat conduction
equation, Mathematical Methods and Physico-Mechanical Fields, vol. 51,
239–246, 2008. (See also Journal of Mathematical Science, vol. 162,
pp. 296–305, 2009).
3. Y. Povstenko, Theory of thermoelasticity based on the space-time-
fractional heat conduction equation, Physica Scripta T, vol. 136, 014017,
2009.
4. S. G. Samko, A. A. Kilbas, and O. I. Marichev, Fractional Integrals and
Derivatives, Theory and Applications. Gordon and Breach, Amsterdam,
1993.
5. R. Gorenflo and F. Mainardi, Fractional calculus: integral and differen-
tial equations of fractional order, in: Fractals and Fractional Calculus in
Continuum Mechanics (A. Carpinteri and F. Mainardi, eds.), Springer-
Verlag, New York, pp. 223–276, 1997.
6. I. Podlubny, Fractional Differential Equations. Academic Press, San
Diego, 1999.
7. A. A. Kilbas, H. M. Srivastava, and J. J. Trujillo, Theory and Applica-
tions of Fractional Differential Equations. Elsevier, Amsterdam, 2006.
8. R. L. Magin. Fractional Calculus in Bioengineering. Begell House Pub-
lishers, Inc, Connecticut, 2006.
9. W. Wyss, The fractional diffusion equation, Journal of Mathematical
Physics, vol. 27, pp. 2782–2785, 1986.
10. W. R. Schneider and W. Wyss, Fractional diffusion and wave equations.
Journal of Mathematical Physics, vol. 30, pp. 134–144, 1989.
11. Y. Fujita, Integrodifferential equation which interpolates the heat equa-
tion and the wave equation, Osaka Journal of Mathematics, vol. 27,
pp. 309–321, 1990.
12. F. Mainardi, The fundamental solutions for the fractional diffusion-wave
equation, Applied Mathematics Letters, vol. 9, pp. 23–28, 1996.
13. F. Mainardi, Fractional relaxation-oscillation and fractional diffusion-
wave phenomena, Chaos, Solitons and Fractals, vol. 7, pp. 1461–1477,
1996.
14. Y. Povstenko, Different kinds of boundary condition for time-fractional
heat conduction equation, in 13th International Carpathian Con-

12
 DOI: 10.1685/journal.caim.482

trol Conference, Podbanské, Slovak Republic (I. Petraš, I. Podlubny,

K. Kostúr, J. Kačúr, A. Mojžišová, eds.), IEEE, pp. 588–591, 2012.
15. Y. Povstenko, Fractional heat conduction in infinite one-dimensional
composite medium, Journal of Thermal Stresses, vol. 36, pp. 351–363,
2013.
16. A. V. Luikov, Analytical Heat Diffusion Theory. Academic Press, New
York, 1968.
17. I. A. Strakhov, Heat transfer in the case of a semiinfinite body heated by
thin parallel plates, Journal of Engineering Physics and Thermophysics,
vol. 22, pp. 232–237, 1972.
18. M. N. Özişik, Heat Conduction. John Wiley, New York, 1980.
19. F. de Monte, An analytic approach to the unsteady heat conduction
processes in one-dimensional composite media, International Journal of
Heat and Mass Transfer, vol. 45, pp. 1333–1343, 2002.
20. R. Kulchytsky-Zhyhailo and S. J. Matysiak, On heat conduction prob-
lem in a semi-infinite periodically laminated layer, International Com-
munications in Heat and Mass Transfer, vol. 32, pp. 123–132, 2005.
21. Y. Povstenko, Fundamental solutions to time-fractional heat conduction
equations in two joint half-lines, Central European Journal of Physics,
vol. 11, pp. 1284–1294, 2013.
22. Y. Povstenko, Fractional heat conduction in an infinite medium with a
spherical inclusion, Entropy, vol. 15, pp. 4122–4133, 2013.
α
23. J. Mikusiński, On the function whose Laplace transform Is e−s , Studia
Mathematica, vol. 18, pp. 191–198, 1959.
24. B. Stanković, On the function of E. M. Wright, Publications de l’Institut
Mathématique, vol. 10 (24), pp. 113–124, 1970.
25. Lj. Gajić and B. Stanković, Some properties of Wright’s function, Pub-
lications de l’Institut Mathématique, vol. 20 (34), pp. 91–98, 1976.
26. R. Gorenflo, Y. Luchko, and F. Mainardi, Analytical properties and ap-
plications of the Wright function, Fractional Calculus & Applied Anal-
ysis, vol. 2, pp. 383–413, 1999.

13