Heat Mass Transfer (2007) 43:1241–1247

DOI 10.1007/s00231-006-0182-y

ORIGINAL

Variable permeability effect on buoyancy-induced inclined

boundary layer flow in a saturated porous medium with variable
wall temperature
I. A. Hassanien Æ A. M. Elaiw

Received: 18 June 2005 / Accepted: 14 August 2006 / Published online: 17 November 2006

Springer-Verlag 2006

Abstract The analysis is carried out for buoyancy- Nux local Nusselt number
induced boundary layer flow adjacent to an inclined Rax local Rayliegh number
heated surface in a saturated porous medium incor- P Pressure
porating the variation of permeability and thermal u streamwise velocity component
conductivity due to paking particles with non-uniform v normal velocity component
temperature. The surface temperature is assumed to x axial coordinate
vary as a power function of the axial coordinate mea- y normal coordinate
sured from the leading edge of the surface. Both the a (y) thermal diffusivity
streamwise and normal component of the buoyancy b volumetric coefficient of thermal expansion
force are retained in the momentum equations. e (y) porosity of the saturated porous medium
Numerical solutions are obtained in the cases of uni- g pseudo-similarity variable
form and nonuniform permeability and various values h dimensionless temperature
of the inclination parameter n (x) = (Rax cos u)1/3 kf thermal conductivity of the fluid
tan u by using finite difference method. The problem is ks thermal conductivity of the solid
solved using nonsimilarity solutions for the case of km effective thermal conductivity of the saturated
variable wall temperature. Results for the details of the porous medium
velocity and temperature fields as well as local Nusselt l dynamic viscosity of the fluid
number have been presented. n inclination parameter
q fluid density
List of symbols r ratio of thermal conductivity of the solid to the
A, m real constants in Eq. 8 fluid
d, d* constants defined in Eqs. 10 and 11 F dimensionless temperature gradient with (¶h/
f dimensionless stream function ¶f)
G dimensionless normal velocity ¶f/¶f w stream function
g gravitational acceleration w conditions at the wall
K(y) permeability of porous medium ¥ conditions at the free stream

I. A. Hassanien (&)
Department of Mathematics, Faculty of Science,
Assiut University, Assiut, Egypt 1 Introduction
e-mail: [email protected]
Heat transfer in saturated porous media has received
A. M. Elaiw considerable attention in recent years because of its
Department of Mathematics, Faculty of Science,
Al-Azhar University, Assiut, Egypt important applications in geophysics and energy related
e-mail: [email protected] engineering problems. These include the utilization of

123
1242 Heat Mass Transfer (2007) 43:1241–1247

geothermal energy, the control of pollutant spread in component of the buoyancy force are retained in the
groundwater, as well as the design of nuclear reactors, momentum equations. The transformed boundary
compact heat exchangers, solar power collectors, and layer equations which are given by using the nonsimilar
high performance insulation for buildings and many solution approach, are solved by means of a finite
more. A comprehensive literature survey on this subject difference method. The numerical results are obtained
can be found in the recent books by Nield and Bejan [1], for the velocity and temperature fields for both uni-
and Ingham and Pop [2]. form permeability (UP) and VP.
The buoyancy-induced motion of fluid through
permeable material is an important mechanism of
transport. Cheng and Chang [3] developed the simi- 2 Analysis
larity solutions for buoyancy-induced flow in a satu-
rated porous medium adjacent to impermeable Consider an inclined impermeable surface embedded
horizontal surfaces. Cheng and Minkowycz [4] pre- in a porous medium as shown in Fig. 1, where x rep-
sented a similarity analysis for a vertical flat plate resents the distance along the plate from its leading
embedded in a saturated porous medium. For inclined edge, and y represents the distance normal to the sur-
surface, Jang and Chang [5] studied the buoyancy-in- face. The wall temperature is assumed to be a power
duced boundary layer flow in a saturated porous function of x, i.e. Tw = T¥ + Axm, where A is a con-
medium. In some applications, such as fixed-bed cata- stant and T¥ is the free stream temperature. The angle
lytic reactors, packed bed heat exchangers and drying, of inclination, u, is measured from the horizontal. For
the porosity is not uniform but has a maximum value at the mathematical analysis of the problem we assume
the wall and a minimum value away from the wall. This that (1) local thermal equilibrium exists between the
wall-channeling phenomenon has been reported by a fluid and the solid phase, (2) the flow velocity and the
number of investigators such as Vafai [6], Chandr- pores of porous medium are assumed to be small for
asekhara et al. [7], Chandrasekhara [8] and Hong et al. Darcy’s model to be valid, (3) permeability and ther-
[9] for forced, natural and mixed convection boundary mal resistance are functions of the vertical coordinate
layer flows adjacent to horizontal and vertical surfaces. y, (4) the Boussinesq approximation is valid. With
It is shown that the variable porosity effect increases these assumptions, the governing equations are given
the temperature gradient adjacent to the wall resulting by
in the enhancement of the surface heat flux. On the
other hand Chandrasekhara and Namboodiri [10] ob- @u @v
þ ¼ 0; ð1Þ
tained the similarity solutions for combined free and @x @y
forced convection in the presence of inclined surface in

KðyÞ @P
saturated porous media with variable permeability u¼
þ qg sin / ; ð2Þ
(VP) in two special cases, namly (1) uniform wall l @x
temperature, (2) linear variation of temperature with

KðyÞ @P
distance from the origin. Also the normal component v¼
þ qg cos / ; ð3Þ
l @y
of the buoyancy force is neglected. Ibrahim and
Hassanien [11] reported nonsimilarity solutions for the  
@T @T @ @T
VP on combined convection along a non-isothermal u þv ¼ aðyÞ ; ð4Þ
@x @y @y @y
wedge in a saturated porous medium.
The purposed of this paper is to study the effect of q ¼ q1 ½1
bðT
T1 Þ; ð5Þ
VP on free convection boundary layer flow over an
inclined, upward-facing heated plate in a saturated
porous medium. For an inclined surface, the buoyancy
g
force causing motion having a component in both the T = T∞
tangential and normal directions. This causes a pres- y
sure gradient across the boundary layer and leads to a x
theoretical analysis more complicated than that for a δm δt
vertical or horizontal surface. The normal component T = Tw
of the buoyancy force can not be neglected when the
φ
angles of inclination from the horizontal are small. The
wall temperature is a power function of the distance
from the origin. Both the streamwise and normal Fig. 1 Coordinate system

123
Heat Mass Transfer (2007) 43:1241–1247 1243

where u and v are the Darcian velocities in the x and y where, km(y) can be computed according to the
directions; T is the temperature; q is the density; P is following semi-analytical expression given by
the pressure, l is the dynamic viscosity; K(y) is the Nayagam et al. [12]
permeability of the porous medium; b is the thermal
expansion coefficient of the fluid and a (y) = km (y)/ km ðyÞ ¼ ekf þ ð1
eÞks ; ð12Þ
(q¥ cp)f is the effective thermal diffusivity of the porous
medium. km (y) denotes the effective thermal conduc- where kf and ks are the thermal conductivities of the
tivity of the saturated porous medium and (q¥ cp)f fluid and solid, respectively.
denotes the product of density and specific heat of the Hence the expression for the thermal diffusivity has
fluid. the form
The pressure terms appearing in Eqs. 2 and 3 can be h n oi
eliminated through cross-differentiation. The boundary aðyÞ ¼ a1 e1 ð1 þ d e
y=c Þ þ r 1
e1 ð1 þ d e
y=c Þ ;
layer assumption yields ¶/¶x < < ¶/¶y and v < < u.
With w being a stream function such that u ¼ @w ð13Þ
@y ;
@w
v ¼
@x ; Eqs. 1, 2, 3, 4 and 5 become where a¥ = kf/(q¥ cp)f and r = ks/kf, we choose c = x/

 (Rax cos u)1/3 such that K(y) and a(y) are purely
@2w d 1 @w KðyÞq1 gb functions of g only. Equations 6 and 7 become
¼
KðyÞ

@y2 dy KðyÞ @y l


@T @T d e
g 0
 cos /
sin / ; ð6Þ f 00 þ f þ ð1 þ d e
g Þ
@x @y 1 þ d e
g

    
m þ 1 @h m
2
@w @T @w @T @ 2 T daðyÞ @T  mh þ n þ gh0
nh0 ¼ 0; ð14Þ

¼ aðyÞ 2 þ : ð7Þ 3 @n 3
@y @x @x @y @y dy @y
 
aðgÞ 00 d aðgÞ 0 m þ 1 0
The boundary conditions of this problem are: h þ h þ fh
a1 dg a1 3


@w 0 m þ 1 0 @h 0 @f
y¼0 v¼
¼ 0; Tw ¼ T1 þ Axm ;
mf h ¼ n f
h ð15Þ
@x 3 @n @n
ð8Þ
@w with boundary conditions
y!1 u¼ ! 0; T ! T1 ;
@y
9
@f ðn; 0Þ
The following dimensionless variables are intro- f ðn; 0Þ þ n ¼ 0; hðn; 0Þ ¼ 1 =
duced: @n : ð16Þ
0 ;
f ðn; 1Þ ¼ 0; hðn; 1Þ ¼ 0
nðxÞ ¼ ðRax cos /Þ1=3 tan /; gðx;yÞ ¼ yðRax cos /Þ1=3 =x
In the above equations, the primes denote partial
wðx;yÞ T
T1
f ðn;gÞ ¼ ; hðn; gÞ ¼ ; ð9Þ differentiation with respect to g. Equations 14, 15 and
a1 ðRax cos /Þ1=3 Tw
T1
16 are valid for all angels except u = p/2 because n fi
w
T1 Þx
¥ as u fi p/2. They reduce to those equations for the
whereRax ¼ K1 q1 gbðT
la1 is the modified local flow over a horizontal flat plate by Chandrasekhara
Rayleigh number. et al. [7] when n = 0.
Here we consider that the porosity e(y) and per- We are now looking for an approximate solution of
meability K(y) vary exponentially from the wall Eqs. 14 and 15 based on the local similarity and non-
similarity methods (see [13]). The local non-similarity
eðyÞ ¼ e1 ð1 þ de
y=c Þ; ð10Þ
method is applied by differentiating partially with n the
governing Eqs. 14 and 15 and their boundary condi-
KðyÞ ¼ K1 ð1 þ d e
y=c Þ; ð11Þ
tions (16) to obtain a system of equations for ¶f/¶n,
where e¥ and K¥ are the porosity and permeability at ¶h/¶n and their boundary conditions. Terms involving
the edge of the boundary layer; d and d* are constants ¶2f/¶n2 and ¶2h/¶n2 are then neglected in the later
whose values are taken as 1.5 and 3, respectively (see set of equations. For the first level of truncation the
Chandrasekhara [8]). Further, a (y) = km (y)/(q¥ cp)f n derivatives in Eqs. 14 and 15 can be neglected.
also varies since it is related to the effective thermal Thus, the governing equations for the first level of the
conductivity of the saturated porous medium km(y), truncation are

123
1244 Heat Mass Transfer (2007) 43:1241–1247

)
d e
g 0 Gðn; 0Þ ¼ 0; Uðn; 0Þ ¼ 0
f 00 þ f þ ð1 þ d e
g Þ : ð25Þ
1 þ d e
g G0 ðn; 1Þ ¼ 0; Uðn; 1Þ ¼ 0

  
m
2
 mh þ gh0
nh0 ¼ 0; ð17Þ
3 In terms of new variables, it can be shown that the
  velocity components and the local Nusselt number are
aðgÞ 00 d aðgÞ 0 m þ 1 0
h þ h þ f h
mf 0 h ¼ 0 ð18Þ given by
a1 dg a1 3
subjected to the boundary conditions a1 ðRax cos /Þ2=3 0
uðx; yÞ ¼ f ðn; gÞ; ð26Þ
x
)
f ðn; 0Þ ¼ 0; hðn; 0Þ ¼ 1 a1 ðRax cos /Þ1=3
: ð19Þ vðx;yÞ ¼

f 0 ðn; 1Þ ¼ 0; hðn; 1Þ ¼ 0 3x


0 @f
 ðm þ 1Þf þ ðm
2Þgf þ ðm þ 1Þn ; ð27Þ
For the second level of truncation, we introduce G @n
= ¶f/¶n and F = ¶h/¶n and restore all of neglected
terms in the first level of truncation. Thus, the gov- Nux =ðRax cos /Þ1=3 ¼
h0 ðn; 0Þ: ð28Þ
erning equations are
The system of Eqs. 17, 18, 19, 20, 21, 22, 23, 24 and
d e
g 0 25 can be solved using finite difference method as de-
f 00 þ f þ ð1 þ d e
g Þ
1 þ d e
g scribed in the next section.

    
mþ1 m
2 0 0
 mh þ nU þ gh
nh ¼ 0; ð20Þ
3 3
3 Numerical scheme
 
aðgÞ 00 d aðgÞ 0 m þ 1 0
h þ h þ fh The numerical scheme to solve the coupled nonlinear
a1 dg a1 3
equations 17, 18, 23 and 24 with boundary conditions
mþ1 0 19 and 25 is based on the following concepts:

mf 0 h ¼ n ½f U
h0 G ð21Þ
3
1. The boundary conditions for g fi ¥ are replaced
with boundary conditions by f¢(n, gmax) = 0, h (n, gmax) = 0 and G¢(n, gmax) =
) 0, F (n, gmax) = 0, where gmax is sufficiently big
f ðn; 0Þ þ nGðn; 0Þ ¼ 0; hðn; 0Þ ¼ 1 value of g when the boundary conditions for
: ð22Þ velocity field is satisfied.
f 0 ðn; 1Þ ¼ 0; hðn; 1Þ ¼ 0
2. The two-dimensional domain of interest, (n, g) is
discretized with an equi-spaced mesh in the n
The introduction of the two new dependent vari-
direction and another equi-spaced mesh in the g
ables G and F in the problem requires two additional
direction.
equations with appropriate boundary conditions. This
3. The partial derivatives with respect to n and g are
can be obtained by differentiating Eqs. 20, 21 and 22
all evaluated by the central difference approxima-
with respect to n and neglecting the terms ¶G/¶n and
tions.
¶F/¶n which lead to
4. Two iteration loops based on the successive sub-

g
d e stitution are used because of the nonlinearity of the
G00 þ G0 þ ð1 þ d e
g Þ equations.
1 þ d e
g

     5. In the inner iteration loop, the value of n is fixed
4m þ 1 m
2 0 0 0 while each of the equations is solved on the g do-
 Uþ gU
h
nU ¼ 0; ð23Þ
3 3 main as second-order boundary value problems for
  linear ordinary differential equations. The inner
aðgÞ 00 d aðgÞ 0 mþ1
U þ Uþ ðf U0 þh0 GÞ
mðG0 hþf 0 UÞ iteration is continued until the nonlinear solution
a1 dg a1 3 converges for the fixed value of n .
6. In the outer iteration loop, the value of n is ad-
mþ1 0
¼ ½ðf U
h0 GÞþnðG0 U
U0 GÞ ð24Þ vanced from 0 to 1.0 and the derivatives with re-
3 spect to n are updated after every outer iteration
with boundary conditions step.

123
Heat Mass Transfer (2007) 43:1241–1247 1245

More details on the numerical scheme are ex- when m = 0 (constant wall temperature). It can be seen
plained in [14]. It is worth noting that the step size of D that, the dimensionless tangential velocity increases
g = 0.02, D n = 0.05, and g¥ of 8–18 were found to be with increasing n; that is, the tangential velocity in-
satisfactory for a convergence of 10– 6 in nearly all creases with increasing value of Rax, for a given angle
cases. u or increases with increasing inclination angle u for a
given Rayleigh number Rax.

4 Results and discussion 7

Numerical results are obtained for different values of m

6
and n, for both UP, i.e. d = d* = 0 and VP i.e. d,d* „ 0 UP
cases. For the purpose of numerical integration we have ξ =1.0 VP
assumed d = 1.5, d* = 3, r = 2 and e¥ = 0.4 (see 5
[7]). To assess the accuracy of our results, we have
shown a comparison of our results with those from [7] in 4

f'(ξ,η)
Table 1 for the buoyancy induced flow over horizontal
surface. A comparison of our results with those from the m=0.0, 0.5, 1.0
literature indicates that the agreement between the two 3
calculations is good.
Figure 2 shows the effect of the inclination param- 2
eter n on the dimensionless tangential velocity profile f¢
across the boundary layer for both UP and VP cases
1

Table 1 Comparison values of Nux/(Rax cos u)1/3 = –h¢(n, 0) at

n = 0 for the variable permeability (VP) cases 0
0 1 2 3 4
m Present results Chandrasekhara et al. [7] η

0.5 1.4012 1.4013 Fig. 3 The variations of the tangential velocity component
1.0 1.8904 1.890 across the boundary layer for various m
1.5 2.3073 2.307
2 2.6824 2.682

40 1

35 UP
m=0 UP
m=0 VP 0.8
VP
30

25 0.6
θ(ξ,η)
f'(ξ,η)

20
ξ=0, 3, 7, 10 ξ=0.0, 5, 10
0.4
15

10
0.2
5

0 0
0 0.5 1 1.5 2 2.5 0 1 2 3 4
η η

Fig. 2 The variations of the tangential velocity component Fig. 4 The temperature profiles across the boundary layer for
across the boundary layer for various n various n

123
1246 Heat Mass Transfer (2007) 43:1241–1247

Results are also presented for the cases of power- Figures 4 and 5 show that the effects of the param-
law variation of wall temperature. The effects of the eters n and m on the dimensionless temperature pro-
exponent m on the dimensionless tangential velocity files across the boundary layer. We can see that, as n or
profile f¢ is shown in Fig. 3. It seen that, the velocity m increases, the temperature boundary layer thickness
gradient at the wall increases and the momentum decreases and the temperature gradient at the wall
boundary layer thickness decreases as m increases. increases. This means a higher value of the heat
Furthermore, from these figures, for fixed values of n transfer rate is associated with higher values of n and
and m, the velocity is higher for the case of VP. m. Further, from theses figures VP effect reduce the
thermal boundary layer leading to an enhancement of
heat transfer rate. Numerical solutions of the local
1 Nusselt number for selected values of m are shown in
Fig. 6 for various values of inclination parameter n and
UP for UP and VP cases. As expected, the local Nusselt
0.8 ξ=1 VP number increases as n increases. At a given value of n
as m increases the local Nusselt number increases. This
increment is much higher for VP case than UP one.
0.6
θ(ξ,η)

m=0, 1, 1.5
0.4 5 Conclusions

The Buoyancy-induced boundary layer flow in a satu-

0.2
rated porous medium adjacent to an inclined heated
surface incorporating the variation of permeability and
thermal conductivity is studied. The permeability of
the medium is assumed to vary exponentially with
0
0 1 2 3 4 distance from the wall. With power-law variation in the
η wall temperature, the problem is solved using non-
similarity solutions. The transformed nonlinear sys-
Fig. 5 The temperature profiles across the boundary layer for
various m tems of the governing equations were solved
numerically by means of a finite difference scheme.
Velocity and temperature profiles as well as local
7 Nusselt number are presented for the uniform and VP
UP cases. The results clearly indicate that the local Nusselt
6 VP number increase with increasing value of the exponent
m for a given n for both uniform and VP cases. In
comparison to UP, the VP is found to have higher local
5
m=0.0, 0.5, 1 heat transfer rate. This fact may be helpful in choosing
a proper fluid for a practical application.
4
-θ'(ξ,0)

Acknowledgments The authors are grateful to Professor Dieter

Mewes Editor-in-Chief of Heat and Mass Transfer and the un-
3 known referees for many constructive suggestions, which helped
to improve the presentation of the paper.

2
References
1
1. Nield DA, Bejan A (1992) Convection on porous media.
Spriger, Berlin Heidelberg New York
0 2. Ingham DB, Pop I (1997) Transport phenomena in porous
0 2 4 6 8 10 media. Elsevier, London
ξ 3. Cheng P, Chang ID (1976) Int J Heat Mass Transf 19:1267
4. Cheng P, Minkowycz WJ (1977) J Geophys Res 82:2040
Fig. 6 The local Nusselt number versus n for selected values 5. Jang, Chang (1988) Comput Methods Appl Mech Eng 68:333
of m 6. Vafai K (1984) J Fluid Mech 147:233

123
Heat Mass Transfer (2007) 43:1241–1247 1247

7. Chandrasekhara BC, Namboodiri PMS, Hanumanthappa 12. Nayagam M, Jain P, Fairweather (1987) Int Commun Heat
AR (1984) Warm und Stoffubetragung 18:17 Mass transf 14:495
8. Chandrasekhara BC (1985) Warm und Stoffubetragung 13. Sparrow EM, Quack H, Boerner J (1970) AIAA 8:1936
19:195 14. Pereyra V (1978) PASVA3: an adaptive finite-difference
9. Hong JT, Yamada Y, Tien CL (1987) J Heat Transf 109:356 FORTAN program for first order nonlinear boundary value
10. Chandrasekhara BC, Namboodiri PMS (1985) Int J Heat problems, in Lecture Notes in Computer Science, vol 76.
Mass Transf 28:199 Springer, Berlin Heidelberg New York
11. Ibrahim FS, Hassanien IA (2000) Transport Porous Media
39:57

123