Mixed convection within a porous heat generating

horizontal annulus
Khalil Khanafer
a,
*
, Ali J. Chamkha
b
a
Mechanical Engineering Department, University of California, Riverside, CA 92510, USA
b
Mechanical Engineering Department, Kuwait University, Safat 13060, Kuwait
Received 22 August 2002; received in revised form 16 November 2002
Abstract
A numerical investigation of mixed convection in a horizontal annulus lled with a uniform uid-saturated porous
medium in the presence of internal heat generation is carried out. The inner cylinder is heated while the outer cylinder is
cooled. The forced ow is induced by the cold outer cylinder rotating at a constant angular velocity. The ow eld is
modeled using a generalized form of the momentum equation that accounts for the presence of porous medium viscous,
Darcian and inertial eects. Discretization of the governing equations is achieved using a nite element scheme based on
the Galerkin method of weighted residuals. Comparisons with previous works are performed and the results show
excellent agreement. The eects of pertinent parameters such as the internal Rayleigh number, the Darcy number, the
annulus gap, and the Richardson number on the ow and heat transfer characteristics are considered in the present
study. The obtained results depict that the Richardson number plays a signicant role on the heat transfer charac-
terization within the annulus. The present results show that an increase in Reynolds number has a signicant eect on
the ow patterns within the annulus with respect to two-eddy, one-eddy and no-eddy ows. Categorization of the ow
regimes according to the number of eddies is established on the RaRe plane for various Rayleigh numbers.
2003 Elsevier Science Ltd. All rights reserved.
Keywords: Heat generation; Horizontal annulus; Mixed convection and porous medium
1. Introduction
Mixed convection heat transfer in an annulus be-
tween heated rotating horizontal concentric cylinders
has received attention by various investigators in recent
years. The motivation for these studies is derived from
their technological applications such as cooling of tur-
bine rotors, cooling of high speed gas bearings, the ow
condensers for sea water distillation [1], and in the
chemical vapor deposition processes for the semicon-
ductor device fabrication. The uid ow and heat
transfer processes within the annulus are controlled by
the combined eects of inertia, buoyancy and centrifugal
forces. The resultant eect of these determines the ow
pattern and the heat transfer mechanism through the
Grashof number and the rotational Reynolds number,
respectively. Of particular relevance for this problem is
the ratio of the Grashof number to the Reynolds num-
ber, which is termed as the Richardson number. This
ratio indicates the importance of the buoyancy and the
rotational eects on the uid ow and heat transfer
phenomenon within the annulus.
DiPrima and Swinney [2] conducted a comprehensive
review of the analytical and experimental studies for
ow between concentric rotating cylinders. The inter-
action of centrifugal and buoyancy forces in a three
dimensional horizontal annulus with a heated rotating
inner cylinder was studied by Yang and Farouk [1].
In this study, the authors showed that the resulting
ows were fully three dimensional for cases where the
*
Corresponding author. Tel.: +1-909-787-6428; fax: +1-909-
787-2899.
E-mail address: [email protected] (K. Khanafer).
0017-9310/03/$ - see front matter 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0017-9310(02)00483-0
International Journal of Heat and Mass Transfer 46 (2003) 17251735
www.elsevier.com/locate/ijhmt
rotational instability triggered the formation of Taylor
cells. Fusegi et al. [3] performed a numerical study for a
two-dimensional mixed convection within a horizontal
concentric annulus with a heated rotating inner cylinder.
The three-dimensional linear stability of the steady
mixed convection ow in a moderately wide gap annulus
between concentric horizontal cylinders with a heated
rotating inner cylinder was studied numerically by Choi
and Kim [4]. The results of this study illustrated that the
heating of the inner cylinder delayed the formation of
Taylor vortices when the eect of rotation dominated
the buoyancy eect. Moreover, the results showed that
when the eect of buoyant forces predominated over the
eect of the centrifugal forces, the rotation of the inner
cylinder stabilized the natural convection ow within the
horizontal annulus. Lee [5] carried a numerical investi-
gation to determine the temperature and ow patterns of
a uid bounded by two horizontal isothermal concentric
and eccentric cylinders. The author showed in this in-
vestigation that the average Nusselt number increased
with Rayleigh number for both concentric and eccentric
cylinders. Moreover, for a xed Rayleigh number, the
average Nusselt number decreased throughout the ow
for an inner rotating cylinder. Yoo [6] investigated nu-
merically the eect of the cooled rotating outer cylinder
on the mixed convection of air between two horizontal
concentric cylinders. The results demonstrated that the
ow patterns could be categorized into three basic types
according to the number of eddies: two-eddy, one-eddy,
and no-eddy ows. In addition, their results illustrated
that the overall heat transfer at the wall decreased as the
Reynolds number approached the transitional Reynolds
number between the two- and one-eddy ows.
Buoyancy-driven ow and heat transfer between
horizontal concentric cylinders lled with a porous me-
dium has been the subject of many investigations in re-
cent years. This is due to the signicance of such
geometry in a number of engineering applications such
as thermal insulation, thermal storage systems, nuclear
reactors, underground electrical transmission lines, etc.
[7]. Caltagirone [8,9] conducted an extensive analysis of
steady-state natural convection in an annulus lled with
Nomenclature
e
r
; e
/
unit vectors in the radial and angular di-
rections, respectively.
Da Darcy number, j=r
o
r
i

2
F geometric function
~gg gravitational acceleration vector, ms
2
k thermal conductivity, Wm
1
K
Nu average Nusselt number
P dimensionless pressure, Nm
2
Pr Prandtl number, m=a
q
000
volumetric heat generation, Wm
3
r
i
inner cylinder radius, m
r
o
outer cylinder radius, m
Ra Rayleigh number, gbDTr
o
r
i

3
=ma
Ra
I
internal Rayleigh number, gbq
000
r
o
r
i

5
=
mak
Re Reynolds number, r
o
xr
o
r
i
=m
R
I
Richardson number, Ra=PrRe
2
t time, s
T temperature, C
T
r
nondimensional torque acting on the outer
cylinder
T
Cuoette
torque obtained with the Couette velocity
distribution, Nm
u; v nondimensional velocity components in the
radial and angular directions, respectively,
ms
1
u velocity vector, ms
1
x; y Cartesian coordinates, m
X; Y dimensionless coordinates
Greek symbols
a thermal diusivity, m
2
s
1
b thermal expansion coecient m K
1
q
o
density at reference temperature, kg m
3
e porosity
/ angular coordinate
C net circulation of the uid, jW
o
W
i
j
j permeability, m
2
w stream function, m
2
s
1
W nondimensional stream function, w=r
o

r
i
xr
o
W
i
value of the stream function at the inner
cylinder
W
o
value of the stream function at the outer
cylinder
x angular velocity, s
1
X nondimensional angular velocity, xr
o
r
i
=
xr
o
r ratio of the inner cylinder diameter to the
gap width, 2r
i
=r
o
r
i

h dimensionless temperature, T T
o
=T
i

T
o

s dimensionless time, txr

m kinematic viscosity, m
2
s
1
Subscripts
i inner cylinder
o outer cylinder
1726 K. Khanafer, A.J. Chamkha / International Journal of Heat and Mass Transfer 46 (2003) 17251735
a porous medium. Both a perturbation method and a
nite dierence technique were used to solve the two-
dimensional governing equations. A uctuating three-
dimensional regime in the upper part of the porous layer
was observed even though the lower part remained
strictly two-dimensional. In a related work, Burns and
Tien [10] investigated natural convection in concentric
spheres and horizontal cylinders lled with a porous
medium. Rao et al. [11,12] solved the Boussinesq equa-
tions in both two and three dimensions using the
Galerkin method. The authors in these studies obtained
three possible numerical solutions depending on the
initial conditions for a radius ratio of 2 and for Rayleigh
numbers above 65. Pop et al. [13] analyzed analytically
transient natural convection in a horizontal concentric
annulus lled with a porous medium. The inner and
outer cylinders were maintained at uniform tempera-
tures. Asymptotic solutions for the inner layer, the outer
layer, and the core were obtained for small times. The
authors found that, for relatively high Rayleigh numbers,
the core was stratied and not isothermal. Stewart and
Burns [14] investigated numerically the eects of a per-
meable inner boundary on the maximum temperature
and the convective ows for a two-dimensional horizon-
tal annulus containing a uniformly heat generating po-
rous media. The results of that study illustrated that
multi-cellular ows occurred at the highest Rayliegh
numbers investigated. In addition, inverted symmetry in
ow patterns and temperature distributions happened
when the heated isothermal wall condition changed from
one cylinder wall to another. Recently, Charrier-Mojtabi
[7] conducted a numerical investigation of two- and three-
dimensional free convection ows in a saturated porous
horizontal annulus heated from the inner surface using a
Fourier-Galerkin approximation for the periodic azi-
muthal and axial directions and a collocation-Chebyshev
approximation in the conned radial direction. Bifurca-
tion points between two-dimensional uni-cellular ows
and either two-dimensional multi-cellular or three-
dimensional ows were determined numerically.
To the best knowledge of the authors, no attention
has been given to the problem of mixed convection ow
within a horizontal concentric annulus lled with a po-
rous medium for a cooled rotating outer cylinder with
internal heat generation. The present study is focused on
the analysis of the uid ow and heat transfer within the
annulus using a generalized form of the momentum
equation that accounts for the porous medium viscous,
Fig. 1. Schematic of the physical model and coordinate system.
Fig. 2. Comparison of the streamlines and isotherms between
the present solution and that of Yoo [6]: (a) Re 100 and
Ra 10
4
and (b) Re 200 and Ra 10
4
.
K. Khanafer, A.J. Chamkha / International Journal of Heat and Mass Transfer 46 (2003) 17251735 1727
Darcian and inertial eects. In addition, the eect of the
of the forced ow induced by the cooled rotating outer
cylinder on the characteristics of buoyancy-driven ow
and heat transfer in a porous layer is considered in the
present study. The eects of pertinent parameters such
as the internal Rayleigh number, Reynolds number,
Richardson number, Darcy number, and the ratio of the
diameter of the inner cylinder to the annulus width on
the ow and heat transfer characteristics are investigated
in the present study.
2. Problem formulation
The problem under investigation is a laminar two-
dimensional mixed-convective ow and heat transfer in
an innitely long horizontal annulus between two con-
centric cylinders lled with a uniform uid-saturated
porous medium in the presence of internal heat gener-
ation. The geometry of the problem and the coordinate
system are shown in Fig. 1. The inner cylinder of radius
r
i
and the outer cylinder of radius r
o
are kept at uniform
and constant temperatures T
i
and T
o
, respectively, with
T
i
> T
o
. The inner cylinder is xed, while the outer
cooled cylinder is rotating in the counter-clockwise di-
rection with a constant angular velocity x. The porous
medium of porosity e and permeability j is saturated
with an incompressible Newtonian uid. Viscous heat
dissipation in the uid is assumed to be negligible in
comparison to conduction and convection heat transfer
eects. It is assumed in the analysis that the thermo-
physical properties of the uid are independent of tem-
Fig. 3. Comparison of the average Nusselt number for various Reynolds numbers r 2.
Fig. 4. Comparison of the streamlines and isotherms between the present solution and that of Barbosa and Saatdjian [18] in a porous
horizontal cylindrical annulus Ra 100; R R
o
=R
i
2.
1728 K. Khanafer, A.J. Chamkha / International Journal of Heat and Mass Transfer 46 (2003) 17251735
perature except for the density in the buoyancy term,
that is, the Boussinesq approximation is invoked. This
density variation can be described by the following
equation:
q q
o
1 bT T
o
1
where b is the coecient for thermal expansion given
by
b
1
q
o
oq
oT
_ _
P
2
Taking into consideration the above assumptions and
using the following dimensionless parameters
R
i

r
i
r
o
r
i
; r
o

r
o
r
o
r
i
; u
V
xr
o
h
T T
o
T
i
T
o
; P
p
q
o
xr
o

2
and s
xr
o
t
r
o
r
i
; 3
The governing equations for this investigation can be
written in dimensionless form as
r u 0 4
Fig. 6. Eect of the Reynolds number on the streamlines Da 10
1
; Ra 10
3
; Ra
I
0 and r 2.
Fig. 5. Comparison of the streamlines and isotherms between the present solution and that of Mojtabi [7] in a porous horizontal
cylindrical annulus Ra 200; R R
o
=R
i
2.
K. Khanafer, A.J. Chamkha / International Journal of Heat and Mass Transfer 46 (2003) 17251735 1729
1
e
ou
os
rP
r
2
u
Re

u
ReDa

F

Da
p juju
R
I
h cos /e
r
h sin /e
/
5
oh
os
u rh
r
2
h
PrRe

Ra
I
Ra
1
PrRe
6
where u is the velocity vector (u; v), F is a function that
depends on the Reynolds number and the microstruc-
ture of the porous medium [15] and e is the porosity. The
nondimensional parameters in the above equations are
the Darcy number Da, the external Rayleigh number Ra,
the internal Rayleigh number Ra
I
, the Richardson
number R
I
, the Reynolds number Re, and the Prandtl
number Pr. These parameters are dened by
Da
j
r
o
r
i

2
; Ra
gbT
i
T
o
r
o
r
i

3
ma
;
Ra
I

gbq
000
r
o
r
i

5
mak
; Re
xr
o
r
o
r
i

m
;
R
I

Ra
PrRe
2
; Pr
m
a
;
_

_
7
where m and a are the uid kinematic viscosity and the
eective thermal diusivity of the porous medium, re-
spectively. It should be noted that the convective terms
in the generalized momentum equation, Eq. (5), are
dropped based on the analysis given by Vafai and Tien
[15].
The dimensionless form of the initial conditions for
the present investigation is given by:
u v h 0 at s 0 8
The dimensionless form of the boundary conditions for
the problem is expressed as:
u v 0; h 1 at R R
i
9
u 0; v 1; h 0 at R R
o
10
The local Nusselt numbers along the inner and outer
cylinders are calculated as the actual heat transfer di-
vided by the heat transfer in the absence of any uid
motion
Nu
i
/ R
oh
oR
_ __
Nu
cond
ln
R
o
R
i
R
oh
oR
_ _
RR
i
11
Nu
o
/ R
oh
oR
_ __
Nu
cond
ln
R
o
R
i
R
oh
oR
_ _
RR
o
12
The average Nusselt numbers at the inner and outer
cylinders are given, respectively, by
Nu
i

1
2p
_
2p
0
Nu
i
/d/ and
Nu
o

1
2p
_
2p
0
Nu
o
/ d/ 13
For steady state solutions and in the absence of heat
generation, both expressions in Eq. (13) predict the same
result. The nondimensional torque acting on the outer
cylinder is given by
T
r

_
2p
0
sR
2
d/; s R
o
oR
v
R
_ _
14
Fig. 7. Eect of the Rayleigh number on the streamlines and
isotherms Da 10
1
; Ra
I
0; Re 10 and r 2.
1730 K. Khanafer, A.J. Chamkha / International Journal of Heat and Mass Transfer 46 (2003) 17251735
3. Numerical method
A nite element formulation based on the Galerkin
method is employed to solve the governing equations
subject to the initial and boundary conditions for the
present study. The application of this technique is well
described by Taylor and Hood [16] and Gresho et al.
[17]. The highly coupled and non-linear algebraic
equations resulting from the discretization of the gov-
erning equations are solved using an iterative solution
scheme using the segregated solution algorithm. The
advantage of using this method is that the global system
matrix is decomposed into smaller submatrices and then
solved in a sequential manner. This technique will result
in considerably fewer storage requirements. The conju-
gate residual scheme is used to solve the symmetric
pressure-type equation systems, while the conjugate
gradient squared is used for the non-symmetric advec-
tion-diusion-type equations. A variable grid-size sys-
tem was implemented in the present investigation
especially near the walls to capture the rapid changes in
the dependent variables. Extensive numerical experi-
mentation was also performed to attain grid-indepen-
dent results for all the eld variables. When the relative
change in variables between consecutive iterations was
less than 10
5
, convergence was assumed to have been
achieved.
The validation of the present numerical code is per-
formed against the work of Yoo [6] in the absence of the
porous medium and is shown in Fig. 2. It can be seen
from this gure that the solution of the present numer-
ical code is in excellent agreement with the numerical
results of Yoo [6] for various Reynolds and Rayleigh
numbers. In addition, comparison of the average Nus-
selt number between the present results and that of Yoo
[6] is made as shown in Fig. 3. Both results are found to
be in excellent agreement for various Rayleigh and
Reynolds numbers. Moreover, the present numerical
approach was validated against the works of Barbosa
and Saatdjian [18] and Mojtabi [7] in a porous hori-
zontal cylindrical annulus for a radius ratio of 2 and
Rayleigh numbers of 100 and 200 as shown in Figs. 4
and 5. These comparisons show excellent agreements
between the present results and other works available in
the literature.
4. Results and discussion
A wide range of pertinent parameters such as
Reynolds number, Rayleigh number, Darcy number,
and the annulus gap width are analyzed in this study.
The range of the Reynolds number used in this study
was varied between 5 6Re 6200, Rayleigh number
between 10
3
6Ra 610
5
, internal Rayleigh number be-
tween 0 6Ra
I
610
6
, Darcy number between 10
6
6
Da 610
1
, and the ratio of the inner cylinder diameter
to the annulus gap width was taken between 0:5 6
r65. Based on the measurements reported by Hunt
and Tien [19] for a foam material made of carbon, the
porosity of the porous medium was assumed constant
e 0:97 and the inertia coecient was selected to be
F 0:1.
The eect of the Reynolds number on the streamlines
is shown in Fig. 6 for a stagnant outer cylinder and
relatively small Rayleigh numbers. For small Reynolds
Fig. 8. Categorization of the ow regimes on the RaRe plane for various Rayleigh numbers (Da 0:1; Ra
I
0; r 2).
K. Khanafer, A.J. Chamkha / International Journal of Heat and Mass Transfer 46 (2003) 17251735 1731
numbers the ow eld within the annulus is character-
ized by two symmetric kidney-shaped eddies with re-
spect to the vertical axis as depicted by Kuehn and
Goldstein [20]. This pattern in the uid ow is mainly
due to the inuence of the buoyancy force. As the
Reynolds number increases due to the rotation of the
outer cylinder, Fig. 6 illustrates three dierent ow
patterns characterized by two eddies, one eddy, and no
eddy depending on the speed of the outer cylinder. Ini-
tially as Reynolds number is increased slightly Re 5,
the symmetry in the ow patterns breaks down and the
speed of the eddy in the region of p 6/62p is reduced.
An additional increase in the Reynolds number
Re 25 causes the eddy in that region to disappear as
a result of strong induced ow by the rotating outer
cylinder. This eect becomes more pronounced at higher
Reynolds numbers Re 150 where the remaining eddy
in the region 0 6/6p disappears. The isotherms are not
shown in Fig. 6 since the velocities are too small to aect
the temperature distribution at relatively small Reynolds
and Rayleigh numbers. Moreover, for higher Reynolds
numbers, the isotherms also remain unchanged since the
buoyancy force is overwhelmed by the strong inuence
of the induced ow, which indicates a pure heat transfer
by conduction mechanism.
The eect of the Rayleigh number on the streamlines
and the isotherms for a Reynolds number of Re 10
and a Darcy number of Da 10
1
is shown in Fig. 7. It
Fig. 9. Eect of the Darcy number on the streamlines and
isotherms Ra 10
4
; Ra
I
0; Re 10 and r 2.
Fig. 10. Eect of the ratio of the inner cylinder diameter to the
gap width on the streamlines and isotherms Da 10
1
; Ra
10
3
; Ra
I
0 and Re 10.
1732 K. Khanafer, A.J. Chamkha / International Journal of Heat and Mass Transfer 46 (2003) 17251735
can be seen from this gure that as the Rayleigh number
increases, the temperature distribution becomes dis-
torted resulting in an increase in the overall heat trans-
fer. This result can be attributed to the dominance of the
buoyancy force as compared to the centrifugal force. It
is worth noting in Fig. 7 that as the Rayleigh number
increases, the thickness of the thermal plume above the
inner cylinder increases which indicates a steep temper-
ature gradient and hence an increase in the overall heat
transfer within the annulus.
The ow patterns in Figs. 6 and 7 can be categorized
into three dierent patterns namely two eddies, one eddy
and no eddy. The classication of the ow regimes ac-
cording to the number of eddies on the RaRe plane
when r 2 and Da 0:1 is shown in Fig. 8. This gure
gives the transitional Reynolds numbers between dif-
ferent regimes.
The eect of the Darcy number on the streamlines
and the isotherms is depicted in Fig. 9. It can be seen
from this gure that as the Darcy number decreases, the
isotherm pattern change signicantly resulting in a pure
conduction regime. In this case, isotherms resemble
concentric circles for small Darcy numbers. This is due
to the bulk damping caused by the presence of the po-
rous matrix. The eect of the Darcy number on the
speed of the eddies within the annulus is also depicted in
Fig. 9. It is clearly seen in this gure that the velocities
within the annulus are substantially decreased except at
a very thin zone next to the outer cylinder where the
mechanical eect of the rotating outer cylinder is sig-
nicant. For small values of the Darcy number
Da 10
5
, the uid experiences a pronounced large
resistance as it ows through the porous matrix causing
the ow to cease in the bulk of the cavity. In this situ-
ation, the convective heat transfer mechanism is com-
pletely suppressed indicating a pure conduction regime
in the bulk of the annulus.
Fig. 10 illustrates the eect of the ratio of the inner
cylinder diameter to the gap width on the streamlines
and the isotherms for Da 10
1
, Ra 10
3
and Rey-
nolds numbers of Re 10, respectively. It can be shown
in this gure that the geometric parameter of the an-
nulus r signicantly alters the uid ow and the tem-
perature patterns within the annulus. As the ratio of the
inner cylinder diameter to the gap width increases,
r 5:0, the uid has a less space to move within the
annulus resulting in a signicant decrease in the heat
transfer rate. In this case, the isotherms for this case
resemble those for the case of pure conduction mecha-
nism within an annulus.
The net circulation of the uid (C jW
o
W
i
j) in the
annulus and the torque normalized with respect to the
torque obtained from Couette ow are shown in Fig. 11
for Re 25, Da 0:1, r 2 and various Rayleigh
numbers. This gure shows that as the Rayleigh number
increases the net circulation of the uid in the annulus
decreases while the torque acting on the outer cylinder
increases due to a strong eect of the natural convection
of uid, which tends to obstruct the circulation of uid.
The eect of the internal heat generation on the
streamlines and the isotherms for Da 0:1, Ra 10
3
,
and a Reynolds number of Re 10 is shown in Fig. 12.
It is seen in this gure that as the internal Rayleigh
number increases, the thermal boundary-layer thickness
along the outer and inner cylinders decreases thus in-
creasing heat transfer. Moreover, the isotherm plot in-
dicates a localized region of high temperature (relative to
the hot wall temperature) between the inner (hot) and
the outer (cold) cylinders. The eect of the internal heat
generation on the streamlines is also shown in Fig. 12.
Fig. 11. The eect of Rayleigh number on the net circulation of uid in the annulus and the normalized torque with respect to Couette
ow acting on the outer cylinder Da 0:1; Ra
I
0; Re 25; r 2.
K. Khanafer, A.J. Chamkha / International Journal of Heat and Mass Transfer 46 (2003) 17251735 1733
This gure illustrates that the streamlines change sig-
nicantly at higher internal Rayleigh numbers.
Finally, the inuence of the internal Rayleigh number
on the average Nusselt numbers along the inner and
outer cylinders is shown in Fig. 13. As expected, as the
internal Rayleigh number increases, both inner and
outer Nusselt numbers increase. It is worth noting that
the outer average Nusselt number is greater than the
inner average Nusselt number for all internal Rayleigh
numbers.
5. Heat transfer correlations
The average Nusselt numbers along the inner cylin-
der and outer cylinder are correlated in terms of Rey-
nolds number 5 6Re 6200, ratio of the inner cylinder
diameter to the gap width 0:25 6r 65, Rayleigh
number 10
3
6Ra 610
5
, Darcy number 10
6
6Da 6
10
1
and the internal Rayleigh number 0 6Ra
I
610
6
.
These correlations can be expressed as follows
Nu
i
1:16985 0:13859r 3:85232Da
0:00803Re 2:301 10
5
Ra
5:273 10
5
Ra
I
Ra
_ _
2
3:30481 10
8
Ra
I
Ra
_ _
3
8:8 10
6
rRe
2
15
Nu
o
1:02052 0:09598r 3:73308Da 0:00186Re
3:692 10
5
Ra
I
2:363 10
5
Ra 16
6. Conclusions
Mixed convection heat transfer in a two-dimensional
horizontal annulus lled with a uniform uid-saturated
porous medium is investigated numerically for a wide
range of pertinent parameters. The inner cylinder is
heated while the outer cylinder is cooled and rotating at
a constant angular velocity. The ow eld is modeled
using the generalized form of the momentum equation
accounting for the viscous, Darcian and the porous
medium inertial eects. The present results showed that
the Reynolds number has a signicant eect on the ow
patterns within the annulus with respect to two-eddy,
one-eddy and no-eddy ows. Moreover, the obtained
results depict that the ratio of the inner cylinder diam-
eter to the gap width plays a signicant role on the
characterization of the heat transfer mechanism within
the porous medium-lled annulus. Categorization of the
ow regimes according to the number of eddies is es-
tablished on the RaRe plane for various Rayleigh
numbers. The eect of the Darcy number on the
streamlines, isotherms and the heat transfer rates in
terms of the average Nusselt number is presented and
discussed in this study. The presence of the internal heat
Fig. 12. Eect of the internal Rayleigh number on the
streamlines and isotherms Da 10
1
; Ra 10
3
; Re 25; e
0:97; r 2.
1734 K. Khanafer, A.J. Chamkha / International Journal of Heat and Mass Transfer 46 (2003) 17251735
generation within the annulus is found to play a signif-
icant role on the ow patterns and isotherms.
References
[1] L. Yang, B. Farouk, Three-dimensional mixed convection
ows in a horizontal annulus with a heated rotating inner
circular cylinder, Int. J. Heat Mass Transfer 35 (1992)
19471956.
[2] R.C. DePrima, H.L. Swinney, Instabilities and transition
in ow between concentric rotating cylinders, in: H.L.
Swinney, J.P. Gollub (Eds.), Hydrodynamic Instabilities
and the Transition to Turbulence, Springer, New York,
1985, pp. 139186.
[3] T. Fusegi, B. Farouk, K. Ball, Mixed-convection ows
within a horizontal concentric annulus with a heated rotating
inner cylinder, Numer. Heat Transfer 9 (1986) 591604.
[4] J.Y. Choi, M. Kim, Three-dimensional linear stability of
mixed-convection ow between rotating horizontal con-
centric cylinders, Int. J. Heat Mass Transfer 38 (1995) 275
285.
[5] T.S. Lee, Numerical experiments with laminar uid con-
vection between concentric and eccentric heated rotating
cylinders, Numer. Heat Transfer 7 (1984) 7787.
[6] J.S. Yoo, Mixed convection of air between two horizontal
concentric cylinders with a cooled rotating outer cylinder,
Int. J. Heat Mass Transfer 41 (1998) 293302.
[7] M.C. Charrier-Mojtabi, Numerical simulation of two- and
three-dimensional free convection ows in a horizontal
porous annulus using a pressure and temperature formu-
lation, Int. J. Heat Mass Transfer 40 (1997) 15211533.
[8] J.P. Caltagirone, Instabilities thermoconvectives en milieu
poreux, These detat, Univ. Pierre et Marie Curie, Paris,
VI, France, 1976.
[9] J.P. Caltagirone, Thermoconvective instabilities in a po-
rous medium bounded by two concentric horizontal
cylinders, J. Fluid Mech. 65 (1976) 337362.
[10] J.P. Burns, C.L. Tien, Natural convection in porous media
bounded by concentric spheres and horizontal cylinders,
Int. J. Heat Mass Transfer 22 (1979) 929939.
[11] Y.F. Rao, K. Fukuda, S. Hasegawa, Steady and transient
analyses of natural convection in a horizontal porous
annulus with the Galerkin method, ASME J. Heat Trans-
fer 109 (1987) 919927.
[12] Y.F. Rao, K. Fukuda, S. Hasegawa, A numerical study of
three-dimensional natural convection in a horizontal
annulus with a Galerkin method, Int. J. Heat Mass
Transfer 31 (1988) 695707.
[13] I. Pop, D.B. Ingham, P. Cheng, Transient natural con-
vection in a horizontal concentric annulus lled with a
porous medium, ASME J. Heat Transfer 114 (1992) 990
997.
[14] W.E. Stewart Jr., A.S. Burns, Convection in a concentric
annulus with heat generating porous media and a perme-
able inner boundary, Int. Comm. Heat Mass Transfer 19
(1992) 859868.
[15] K. Vafai, C.L. Tien, Boundary and inertia eects on ow
and heat transfer in porous media, Int. J. Heat Mass
Transfer 24 (1981) 195203.
[16] C. Taylor, P. Hood, A numerical solution of the Navier
Stokes equations using nite-element technique, Comput.
Fluids 1 (1973) 7389.
[17] P.M. Gresho, R.L. Lee, R.L. Sani, On the time-dependent
solution of the incompressible NavierStokes equations in
two and three dimensions, in: Recent Advances in Numer-
ical Methods in Fluids, Pineridge, Swansea, UK, 1980.
[18] J.P. Barbosa Mota, E. Saatdjian, Natural convection in a
porous horizontal cylindrical annulus, ASME J. Heat
Transfer 116 (1994) 621626.
[19] M.L. Hunt, C.L. Tien, Eects of thermal dispersion on
forced convection in brous media, Int. J. Heat Mass
Transfer 31 (1988) 301310.
[20] T.H. Kuehn, R.J. Goldstein, An experimental and theo-
retical study of natural convection in the annulus between
horizontal concentric cylinders, J. Fluid Mech. 74 (1976)
695719.
Fig. 13. Eect of the internal Rayleigh number on the average Nusselt number Da 10
1
; Ra 10
3
; Re 25; e 0:97; r 2.
K. Khanafer, A.J. Chamkha / International Journal of Heat and Mass Transfer 46 (2003) 17251735 1735