International Journal of Heat and Mass Transfer 66 (2013) 200–209

Contents lists available at SciVerse ScienceDirect

International Journal of Heat and Mass Transfer

journal homepage: www.elsevier.com/locate/ijhmt

Laminar mixed convection flow and heat transfer characteristics in a lid

driven cavity with a circular cylinder
Khalil Khanafer a,b,⇑, S.M. Aithal c
a
Frankel Vascular Mechanics Laboratory, Biomedical Engineering Department, University of Michigan, Ann Arbor, MI 48109, USA
b
Section of Vascular Surgery, Samuel and Jean Frankel Cardiovascular Center, University of Michigan, Ann Arbor, MI 48109, USA
c
Computing, Environment and Life Sciences Directorate, Argonne National Laboratory, 9700 S. Cass Avenue, Argonne, IL 60439, USA

a r t i c l e i n f o a b s t r a c t

Article history: Mixed convection flow and heat transfer characteristics in a lid-driven cavity with a circular body inside
Received 30 March 2013 are studied numerically using a finite element formulation based on the Galerkin method of weighted
Received in revised form 1 July 2013 residuals. Comparisons of streamlines, isotherms and average Nusselt number are presented to show
Accepted 6 July 2013
the impact of the Richardson number, non-dimensional radius of the cylinder, and the location of the cyl-
Available online 1 August 2013
inder on the transport phenomena within the cavity. The results of this investigation show that the pres-
ence of the cylinder results in an increase in the average Nusselt number compared with a case with no
Keywords:
cylinder. This result is observed for both, an adiabatic and isothermal, boundary condition imposed on the
Lid-driven cavity
Mixed convection
cylinder. The average Nusselt number increases with an increase in the Richardson number for all non-
Circular cylinder dimensional radius of the cylinder studied in this work. Moreover, the optimal heat transfer results are
obtained when placing the cylinder near the bottom wall for various Richardson numbers. For dominant
natural convection (Ri P 2.5), the average Nusselt number increases with an increase in the non-dimen-
sional radius for 0.05 < ro/H < 0.2. Further increase in the non-dimensional radius does not change the
Nusselt number at a particular Ri. For dominant mixed convection, the average Nusselt number increases
with an increase in the radius of the cylinder for various Richardson numbers.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction in vertical pipes is of interest to sodium cooled reactors [8] or en-

Mixed convection in enclosures with a moving boundary is an
interesting problem largely owing to the enhancement in the flow
mixing and recirculation on account of the shear stress generated
by the moving surface, and its overall effect on the imposed ther-
mal gradient. This phenomenon occurs in a number of engineering
applications from solar power, microelectronics to nuclear reactors
[1–7]. In the mixed convection mode of heat transfer, features of
both forced and natural convection are important. In addition to
the numerous engineering applications, investigation of various
configurations under which mixed convection heat transfer occurs,
is of interest as fundamental problems in heat-transfer and fluid
flow. The impact of parameters such as Rayleigh number, Richard-
son number and Grashof number (Ri has effect of Re on it) on the
overall Nusselt number can lead to better understanding and new-
er engineering applications requiring enhanced heat and mass
transfer. For instance, turbulent mixed-convection heat transfer
⇑ Corresponding author at: Frankel Vascular Mechanics Laboratory, Biomedical
Engineering Department, University of Michigan, Ann Arbor, MI 48109, USA.
Tel.: +1 734 647 8033.
E-mail address: khanafer@umich.edu (K. Khanafer).
hanced air-conditioning for improved indoor air quality [9].
Introduction of an obstruction in an enclosure impacts the flow
field and heat transfer. Several studies dealing with enhancing nat-
ural convection in obstructed cavities are documented in the liter-
ature [10–13]. Literature on the body inserted lid-driven cavity is
sparse [14–24]. Dagtekin and Oztop [14] inserted an isothermally
heated rectangular block in a lid-driven cavity at different posi-
tions to simulate the cooling of electronic equipments. They found
that dimension of the body was the most effective parameter on
mixed convection flow. MHD mixed convection in a lid-driven cav-
ity along with joule heating and a centered heat conducting circu-
lar block was studied by Rahman et al. [15–18]. The numerical
results indicated that the Hartmann number, Reynolds number
and Richardson number had strong influence on the streamlines,
isotherms, average Nusselt number at the hot wall and average
temperature of the fluid in the enclosure. Islam et al. [22] studied
mixed convection in a lid driven cavity with an isothermally
heated square blockage and isothermal cold wall temperature im-
posed on all cavity walls. The top wall of the cavity was assumed to
move rightward with a constant speed. Their results showed that
Richardson number, blockage ratio and blockage placement eccen-
tricities have an effect on the average Nusselt number measured

0017-9310/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijheatmasstransfer.2013.07.023
 K. Khanafer, S.M. Aithal / International Journal of Heat and Mass Transfer 66 (2013) 200–209 201

Nomenclature

g gravitational acceleration x, y cartesian coordinates

Gr Grashof Number = gb(Th  TC)H3/m2 X, Y dimensionless cartesian coordinates, (x, y)/H
H cavity side length
Nu Nusselt number Greek symbols
P dimensionless fluid pressure ¼ p=ðqu2o Þ a thermal diffusivity of the fluid
Pr Prandtl number = m/a b coefficient of thermal expansion of fluid
Re Reynolds number = uoH/m dx, dy dimensionless coordinates of the porous block
Ri Richardson number = Gr/Re2 m kinematic viscosity
ro radius of the cylinder h dimensionless temperature = (T  TC)/(Th  TC)
ro/H non-dimensional radius of the cylinder q density
T temperature W dimensionless stream function
t time
u velocity in x-direction Subscripts
uo lid speed C cold
U dimensionless horizontal velocity = u/uo
h hot
v velocity in y-direction
V dimensionless vertical velocity = m/uo

along the solid block. Oztop et al. [23] studied mixed convection and placed at nine prescribed dimensionless locations (dx, dy) with
heat transfer characteristics in a lid-driven differentially heated the default case being at the center of the cavity (0.5H, 0.5H). No
cavity having a circular body. In that study, flow was driven by slip boundary condition is imposed on all cavity walls and cylinder
the left hot wall moving upward with a constant speed while the surfaces. The top wall is moving towards the right with a constant
other walls were assumed stationary. The horizontal walls were speed uo. The vertical walls of the cavity are considered adiabatic
assumed adiabatic while the right wall was maintained at a lower while the top wall is maintained at a colder temperature TC than
temperature than the left wall. Three different boundary condi- the bottom wall, which is maintained at Th. Different thermal
tions were imposed on the inner cylinder, namely, adiabatic, iso- boundary conditions for the circular body are imposed, namely,
thermal or conductive. Their results showed that the direction of adiabatic or isothermal (TC). Moreover, the cavity is filled with
the moving lid was found to be the most important parameter air (Pr = 0.7) as the working fluid and the thermophysical proper-
on flow and temperature characteristics within the cavity. Mixed ties of the working fluid are taken to be constant except for the
convection heat transfer in a lid-driven cavity along with a heated density variation, which is assumed to vary linearly with temper-
circular hollow cylinder positioned at the center of the cavity was ature according to the Boussinesq approximation.
analyzed numerically by Billah et al. [24]. The left cold wall was as- Based on the above considerations, the following variables are
sumed to move upward with a constant speed. The horizontal introduced to render the analysis dimensionless
walls of the cavity were adiabatic while the right wall was main- x y u v
tained at a cold temperature. This study showed that the flow field X¼ ; Y¼ ; U¼ ; V¼
H H uo uo
and temperature distribution were strongly depending on the cyl-
ðT  T C Þ p
inder diameter and the solid–fluid thermal conductivity ratio. h¼ ; P¼ 2 ð1Þ
Earlier studies on mixed convection in lid-driven enclosures
ðT h  T C Þ quo
with an obstacle considered either heated obstacles or a differen- uo H gbðT h  T C ÞH3 m
Re ¼ ; Gr ¼ ; Pr ¼
tially heated cavity. Many engineering applications such as chem- m m2 a
ical reactors, mixing tanks and boilers have obstructions (such as
where P is the dimensionless pressure, Re the Reynolds number, Gr
stirrers with blades) wherein the lower surface of the reactor/boi-
the Grashof number and Pr the Prandtl number. According to the
ler is at a higher temperature as compared to the upper surface.
above assumptions and dimensionless variables, the normalized
The heat and mass transfer characteristics of such engineering
applications are difficult to evaluate from a numerical standpoint.
However, simplified configurations can provide useful engineering
insights and scaling information. With this goal in mind, the cur-
rent numerical investigation aims at exploring the effects of the
Richardson number, size and location of a circular obstruction on
the momentum and energy transport processes in the lid-driven
cavity heated from below. Two temperature boundary conditions
on the cylinder were investigated, namely, adiabatic and constant
temperature. The results of this investigation are presented in
terms of streamlines, isotherms, and average Nusselt number at
the hot surface of the cavity.

2. Mathematical formulation

The physical system under study with the system of coordinates

is depicted in Fig. 1. The flow inside the cavity with height H is con-
sidered steady, laminar, incompressible, and two-dimensional. A
circular cylinder of radius ro (=0.2H) is inserted within the cavity Fig. 1. Schematic of the model with coordinates and boundary conditions.
202 K. Khanafer, S.M. Aithal / International Journal of Heat and Mass Transfer 66 (2013) 200–209

governing equations for flow and temperature within the cavity are
expressed in the following format
@U @V
þ ¼0 ð2Þ
@X @Y
!
@U @U @P 1 @2U @2U
U þV ¼ þ þ ð3Þ
@X @Y @X Re @X 2 @Y 2
!
@V @V @P 1 @2V @2V Gr
U þV ¼ þ þ þ 2h ð4Þ
@X @Y @Y Re @X 2 @Y 2 Re

@h @h r2 h
U þV ¼ ð5Þ
@X @Y Pr Re
The appropriate forms of the dimensionless boundary conditions
used to solve the governing equations (2)–(5) along the cavity
boundaries are
U ¼ 0;
V ¼ 0; h¼1 Y ¼ 0; 0<X<1
U ¼ 1;
V ¼ 0; h¼0 Y ¼ 1; 0<X<1 ð6Þ
@h
U¼V¼ ¼0 X ¼ 0; 1; 06Y 61
@x Fig. 2. Grid distribution.
The boundary conditions used in the present problem for the cylin-
der are:
Table 1
For isothermal : h ¼ 0 ð7aÞ Grid convergence study for various Richardson numbers (Re = 100, Pr = 0.7, ro/
@h H = 0.2).
For adiabatic : ¼0 ð7bÞ
@n Ri Nu
2820 nodes 19,520 nodes 43,680 nodes
2.1. Evaluation of Nusselt number
0.01 2.92 2.92 2.93
1 3.45 3.50 3.50
The local Nusselt number along the heated bottom wall of the 5 4.71 4.70 4.70
cavity is calculated based on a TC as a reference temperature: 10 5.16 5.04 5.06
 
@h
Nuy ¼  ð8aÞ
@Y y¼0

The above equation defines convective heat flux along the cavity. Table 2
The average Nusselt number is calculated by integrating as Comparison of the average Nusselt number at the top wall between various studies at

Z 1 Z 1 Gr = 104 and 106 in the absence of any obstruction.
@h
Nu ¼  Nux dX ¼  dX ð8bÞ Re Gr = 104 Gr = 106
0 0 @Y
Present Ref. [1] Ref. [2] Present Ref. [1] Ref. [2]
100 1.38 1.34 – 1.02 1.02 –
3. Numerical scheme 400 3.76 3.62 3.82 1.17 1.22 1.17
1000 6.56 6.29 6.50 1.72 1.77 1.81
A finite element formulation based on the Galerkin method was
implemented in the present investigation (ADINA 8.8.3, ADINA R&D,
Inc., Watertown, MA) to solve the governing equations in a lid-dri-
Table 3
ven cavity heated from below with a circular cylinder and subject to
Comparison of the average Nusselt number at the lower hot surface of a lid driven
the boundary conditions described above. The finite element code cavity with a circular cylinder between the present results and that of Nek5000 CFD
used in the present work is well documented [25] and the scheme solver (Re = 100, Pr = 0.7, ro/H = 0.2).
is briefly explained here. The finite element equations are obtained
Ri Isothermal cold cylinder Adiabatic cylinder
by establishing a weak form of the governing equations using the
Presenta Nek5000b %Relative Presenta Nek5000b %Relative
Galerkin procedure. The continuity equation, the momentum equa-
error error
tion and the energy equation are weighted with the virtual quanti-
0.01 2.92 2.91 0.34 2.24 2.21 1.34
ties of pressure, velocities and temperature. The governing
1 3.50 3.47 0.86 3.33 3.32 0.30
equations are integrated over the computational domain. The con- 5 4.69 4.67 0.43 4.06 4.06 0
tinuum domain is first divided into a set of simply shaped, non- 10 5.04 4.99 1.0 4.50 4.40 2.22
overlapping regions called elements, within each of which the un- a
ADINA 8.8.3 CFD.
known variables ux, uy, uz, p, and T are approximated by using the b
Open-source CFD solver based on the spectral element method (Argonne
following equations National laboratory).

ux  uhx ¼ /T ½U x 
uy  uhy ¼ /T ½U y  where /, w, and H are the interpolation functions for velocity, pres-
uz  uhz ¼ /T ½U z  ð9Þ sure, and temperature, respectively, and are local functions of the
nodal coordinates for that element as well as the independent vari-
p  ph ¼ wT ½P ables. The vectors [Ux], [Uy], [Uz], [P], and [T] consist of the values of
T  T h ¼ HT ½T the respective variables at the nodes of the element. Substituting
 K. Khanafer, S.M. Aithal / International Journal of Heat and Mass Transfer 66 (2013) 200–209 203

weighted sense by making the residuals orthogonal to the interpo-

lation functions of each element as follows:
R R 9
w  E1 dV ¼ V w  f1 dV ¼ 0 >
=
RV R
V
/  E2 dV ¼ V /  f2 dV ¼ 0 ð11Þ
R R >
;
V
H  E3 dV ¼ V H  f3 dV ¼ 0

This procedure yields a system of equations for each element, which

can be written as

Fig. 3. Comparison of the variation of the local Nusselt number along the heated
bottom wall (isothermal cylinder) between present results and that of Nek5000 for
different Richardson numbers (Re = 100, Pr = 0.7, ro/H = 0.2).
these basis functions into the governing equations and boundary
conditions yields a residual (error) in each of the equations as
follows:
Continuity : f 1 ð/; U x ; U y ; U z Þ ¼ E1
Momentum : f 2 ð/; w; H; U x ; U y ; U z ; P; TÞ ¼ E2
Energy : f 3 ð/; H; U x ; U y ; U z ; TÞ ¼ E3
where E1, E2, and E3 are the residuals resulting from the use of the
finite element approximations. The Galerkin form of the method of
weighted residuals seeks to reduce these errors to zero in a
@u
M þ KðuÞ  u ¼ F ð12Þ
@t
Where u is the column vector of the unknown variables, F is the
force vector (incorporating the boundary conditions), M is the mass
matrix, K is the stiffness matrix (representing the diffusion and con-
vection of energy). The representation of the entire continuum re-
gion of interest is obtained through an assemblage of elements.
A non-uniform grid was used to capture the rapid changes in
the dependent variables especially in the vicinity of the walls
where major gradients occur inside the boundary layer as depicted
in Fig. 2. The Newton–Raphson method was used to solve the dis-
cretized equations in the fluid region. The solution was assumed to
have converged when the relative change in the dependent vari-
able between two successive iterations was <104 as shown below
ð10aÞ X .X
cþ1
jki;j  kci;j j cþ1
jki;j j 6 104 ð13Þ
ð10bÞ
ð10cÞ where kci;j represents any particular dependent variable at iteration
c. Different numerical experiments with various mesh sizes were
performed to achieve grid-independent results. Three different
non-uniform grid systems with the following number of nodes:
2820, 19,520 and 43,680 were examined to calculate the average

Present Streamlines Islam et al. [22]

Isotherms

Fig. 4. Comparison of the streamlines and isotherms between the present results and that of Islam et al. [22] (h/H = 0.25, Re = 100 and Ri = 0.1).
204 K. Khanafer, S.M. Aithal / International Journal of Heat and Mass Transfer 66 (2013) 200–209

Nusselt number along the heated bottom wall. Table 1 shows that a
non-uniform grid system of 19,520 nodes gave grid-independent
results and hence was used for all the results presented in this
paper.
4. Code validation
The present numerical scheme was validated against many
numerical results available in the literature. Table 2 demonstrates
an excellent comparison of the average Nusselt number measured
at the top wall of a lid-driven cavity between various studies at
Gr = 104 and 106 in the absence of any obstruction. Table 3 illus-
trates an excellent comparison of the average Nusselt number be-
tween the present results and that of Nek5000 CFD solver (Open-
source solver; Argonne National Laboratory) in a lid-driven cavity
heated from below with a circular cylinder. Moreover, Fig. 3 shows
excellent comparison of the local Nusselt number variation along
the bottom heated surface obtained using ADINA and Nek5000.
As an additional check on the accuracy of the present scheme,
Fig. 4 illustrates a comparison of streamlines and isotherms be-
tween the present results and that of Islam et al. [22] in a square
cavity with an isothermally heated square blockage at Reynolds
number Re = 100 and a Richardson number Ri = 0.1.The results
shown in Tables 2 and 3 along with Figs. 3 and 4 provide sufficient
confidence in the present numerical scheme.
5. Results and discussion
Fig. 5 shows the effect of the presence of the circular cylinder in
the lid-driven cavity for various Richardson numbers and cylinder

Fig. 5. Effect of the presence of the circular cylinder on the streamlines for various boundary conditions and Richardson numbers (Re = 100, Pr = 0.7, ro/H = 0.2).
 K. Khanafer, S.M. Aithal / International Journal of Heat and Mass Transfer 66 (2013) 200–209 205

wall boundary conditions, namely, adiabatic and isothermal. For
the isothermal cylinder wall case, the wall temperature is main-
tained at Tc. Re, Pr and ro/H were kept constant at 100, 0.7 and
0.2, respectively for these simulations. The results were compared
to a case without a cylinder in the cavity. It is instructive to study
Dw (difference between the maximum and minimum value of the
stream function) as a function of Ri. For all the cases shown (no cyl-
inder and the two wall BCs) Dw increases monotonically with Ri. A
higher Ri represents a higher influence of natural convection and
thus a higher value of buoyancy induced flow rate. For every value
of Ri studied, the Dw for the two different boundary conditions,
namely, isothermal and adiabatic differ by less than 1% while the
Dw for the case with no cylinder is about 33% higher. It must be
noted that while the overall blockage of the cylinder is only about
12.5% (p  0.22 = 0.125) of the cavity area, the drop in the overall
Dw is about 33%. This implies that introduction of the blockage
has a significant impact on the flow characteristics. It also signifies
that the thermal boundary condition on the cylinder surface has a
negligible effect.
Fig. 6 shows the effect of the cylinder on the isotherms for var-
ious Ri. Two cases, namely, isothermal cylinder wall (at Tc) and adi-
abatic cylinder wall are shown (along with a case with no
cylinder). As expected, a higher Ri implies a higher natural convec-
tion, leading to a greater ‘‘homogenization’’ of the temperature
field in the cavity. The presence of the isothermal cylinder leads
to a ‘‘thermal shielding’’ effect which results in the fluid on the
right side of the cavity being at a lower temperature as compared
to the no cylinder and adiabatic cylinder case. This is an important
effect to consider when using a blockage in engineering applica-
tions using mixed-convection effects. Figs. 7 and 8 show the effect

Fig. 6. Effect of the presence of the circular cylinder on the isotherms for different boundary conditions and Richardson numbers (Re = 100, Pr = 0.7, ro/H = 0.2).
206 K. Khanafer, S.M. Aithal / International Journal of Heat and Mass Transfer 66 (2013) 200–209

Fig. 7. Effect of varying Richardson number on temperature along a horizontal

plane (Y = 0.15) of the cavity (Re = 100, Pr = 0.7, ro/H = 0.2).

of Richardson number on the variation of the temperature along

different locations namely, Y = 0.15 (below the cylinder), X = 0.15
(left of the cylinder) and X = 0.85 (right of the cylinder), respec-
tively. Fig. 7 show that for both cases (isothermal and adiabatic cyl-
inders), the higher the value of Ri, the lower the value of
temperature at any given horizontal location. It is also seen that
for any given Ri, the temperature of the adiabatic case is higher
than the isothermal case. Additionally, for both cases (isothermal
and adiabatic cylinders), the difference between the maximum
and minimum temperatures in the horizontal direction decreases
with decreasing Ri. This is to be expected as lower Ri indicates a

Fig. 9. Effect of a fixed circular cylinder on (a) local Nusselt number along the
heated wall and (b) the average Nusselt number for various Richardson numbers
(Re = 100, Pr = 0.7, ro/H = 0.2).

Ri = 0.01
(a)
higher Re, implying greater mixing of the fluid leading to less var-
iation in temperature.
Fig. 8(a) and (b) shows the effect of Richardson number on the
θ temperature along vertical planes of the cavity, namely X = 0.15
Ri = 1 Ri = 5
(left of cylinder) and X = 0.85 (right of cylinder), respectively. The
motion of the top surface moving from left to right induces a clock-
X = 0.15 wise flow effect. This flow pattern has an interesting effect on the
temperature gradient on the left and right sides of the cylinder. The
Isothermal Cylinder Ri = 0.01 temperature gradient is affected both by the Re and boundary con-
Adiabatic Cylinder
dition imposed on the cylinder. As the Re increases (lower Ri), the
thermal gradient on the left becomes more uniform for both the
isothermal and adiabatic cases. This can be explained by the en-
hanced fluid mixing on account of the higher Re. It is seen that
Isothermal Cylinder (b) the isothermal BC leads to a near-linear dT/dy for Ri = 0.01 (high
Adiabatic Cylinder
Re). The thermal gradient for the adiabatic BC also more uniform
as compared to Ri = 1 and Ri = 5, but less pronounced as compared
to the isothermal case. The thermal shielding effect due to the iso-
thermal BC is evident in the temperature profiles shown on the
Ri = 0.01 X = 0.85
θ right side (X = 0.85) of the cylinder. It is seen that for Ri = 0.01,
there is practically no temperature gradient for about 60% of the
Ri = 1 distance from the top surface. The adiabatic case has a slight dT/
dy at Ri = 0.01 from Y = 0.5 to Y = 0 as compared to the adiabatic
case.
Ri = 5
The effect of varying Richardson number on the local Nusselt
number measured along the bottom heated surface is demon-
strated in Fig. 9(a). Local Nusselt number is found to increase with
an increase in the Richardson number. For centrally placed isother-
Fig. 8. Effect of varying Richardson on the temperature along a vertical plane of the mal cylinder, Richardson number of 10 (dominant natural convec-
cavity (Re = 100, Pr = 0.7, ro/H = 0.2). tion) exhibits the largest local Nusselt number variation along the
 K. Khanafer, S.M. Aithal / International Journal of Heat and Mass Transfer 66 (2013) 200–209
Fig. 10. Effect of varying Richardson number on the velocity (U and V) and
temperature along a horizontal plane of the cavity (Re = 100, Pr = 0.7, ro/H = 0.2).
bottom surface compared to other configurations. Fig. 9(a) illus-
trates that no cylinder configuration exhibits the lowest local Nus-
selt number variation along the bottom surface compared to
isothermal and adiabatic centrally placed cylinders for various
Richardson numbers. This can be attributed to a fact that the cen-
trally placed cylinder causes an increase in the flow velocity be-
tween the cylinder and the bottom surface and consequently
increases the temperature gradient in the vertical direction.
Fig. 9(b) shows the average Nu as a function of Ri at the heated
surface of the cavity. For each of the cases considered, namely, no
cylinder, adiabatic cylinder and isothermal cylinder – the Nu in-
creases monotonically with Ri but tends to an asymptotic value
for higher Ri. For a given Ri, the average Nu for the isothermal wall
207
Fig. 11(a). Effect of varying the size of the cylinder on the streamlines for various
Richardson numbers using Re = 100 and Pr = 0.7.
case is higher than the adiabatic case and no cylinder case. For
example, the centrally placed isothermal cylinder causes an in-
crease in the velocity in the region between the bottom heated sur-
face and the cylinder and consequently increases heat transfer as
depicted in Fig. 10. The increase in the velocity is due to a small
gap between the bottom wall and the cylinder. Fig. 10 illustrates
the effect of varying the Richardson number on the velocity and
temperature profiles along a horizontal plane placed at Y = 0.15
from the bottom surface. The velocity components were found to
increase with an increase in the Richardson number while the tem-
perature decreases. Fig. 10 shows that the horizontal velocity com-
ponent reaches a maximum value in the region below the center of
the cylinder while the vertical velocity component is nearly zero
below the center of the cylinder. This is to be expected since the
motion of the lid causes a clockwise flow. On the right, the flow
vectors are directed towards the bottom plate while towards the
left they are directed towards the top plate. At Y = 0.5 (mid-plane)
the vertical component of velocity goes to zero indicating a near
horizontal flow in the negative X-direction. The increased velocity
with increasing Ri causes a greater convective flow of the heat from
the bottom surface leading to higher Nusselts number.
The effect of changing the size of the cylinder on the stream-
lines and isotherms for various Richardson numbers (isothermal
cylinder wall) are depicted in Figs. 11(a) and 11(b) respectively.
Fig. 11(a) shows that as the cylinder radius increases, the passage
width between the cylinder and the top wall of the cavity deceases
and accordingly the flow behavior changes. For dominant forced
convection mechanism (Ri = 0.01), the main recirculation vortex
starts to elongate horizontally as the non-dimensional radius of
the cylinder increases (ro/H = 0.2) and eventually breaks up into
208 K. Khanafer, S.M. Aithal / International Journal of Heat and Mass Transfer 66 (2013) 200–209

Fig. 12. Effect of varying the non-dimensional radius on the average Nusselt
number for various Richardson numbers (Re = 100 and Pr = 0.7).

Table 4
Effect of varying the location of the cylinder on the average Nusselt number using
Ri = 0.01, Re = 100, Pr = 0.7, and ro/H = 0.2.

Vertical placement (dx, dy) Nu
Top (0.3, 0.7) 1.83
(0.5, 0.7) 2.26
(0.7, 0.7) 1.87
Middle (0.3, 0.5) 2.43
(0.5, 0.5) 2.92
(0.7, 0.5) 2.42
Bottom (0.3, 0.3) 5.03
(0.5, 0.3) 5.19
(0.7, 0.3) 4.99
Fig. 11(b). Effect of varying the size of the cylinder on the isotherms for various
Richardson numbers (Re = 100 and Pr = 0.7).

Table 5
two small vortices between the cylinder and the top wall for ro/ Effect of varying the location of the cylinder on the average Nusselt number using
Ri = 5, Re = 100, Pr = 0.7, and ro/H = 0.2.
H = 0.3. As the Richardson number increases, Fig. 11(a) shows an

absence of vortices within the cavity for Richardson number of 5 Vertical placement (dx, dy) Nu
for various non-dimensional radius of the cylinder. This is to be ex-
Top (0.3, 0.7) 3.17
pected since the natural convection mechanism becomes domi- (0.5, 0.7) 2.89
nant with increasing Ri and there is not much flow induced (0.7, 0.7) 3.03
effect. Fig. 11(b) illustrates the influence of changing the radius Middle (0.3, 0.5) 3.50
of the cylinder on the isotherms for various Richardson numbers. (0.5, 0.5) 4.70
Fig. 11(b) shows that the isotherms are more clustered near the (0.7, 0.5) 3.45
bottom wall as the radius of the cylinder increases, indicating the Bottom (0.3, 0.3) 5.81
development of thin boundary layers with large normal tempera- (0.5, 0.3) 5.52
ture gradients along the bottom wall. (0.7, 0.3) 5.83

Fig. 12 shows the effect of size of cylinder on the average Nus-

selt number at the heated surface (bottom) for various Richardson
numbers (isothermal cylinder wall). The higher temperature gradi-
ent for higher values of non-dimensional radius leads to increased
average Nusselt number at the bottom wall (as discussed above).
The average Nusselt number increases with an increase in the
Richardson number for all non-dimensional radius of the cylinder
studied in this work. Moreover, for dominant mixed convection,
the average Nusselt number increases with an increase in the ra-
dius of the cylinder for various Richardson numbers. Fig. 12 shows
an interesting result associated with the size of the cylinder. For
dominant natural convection (Ri P 2.5), the average Nusselt num-
ber increases with an increase in the non-dimensional radius for
0.05 < ro/H < 0.2. Further increase in the non-dimensional radius
does not change the Nusselt number for Ri > 2.5.
In addition to the size (non-dimensional cylinder radius), the
location of the cylinder also plays an important role in modifying
the velocity and thermal fields in the lid-driven cavity. The
impact of changing the location of the cylinder on the average Nus-
selt number is shown in Tables 4 and 5 for different Richardson
numbers. The cylinder is placed at three vertical positions near
the top, middle, and bottom of the cavity. In this regard, nine dif-
ferent positions of the cylinder were obtained with their dimen-
sionless center point (dx, dy) being given in Tables 4 and 5. Two
Richardson numbers representing two limiting cases; dominant
forced convection (Ri = 0.01) and dominant natural convection
(Ri = 5) were considered using Re = 100, Pr = 0.7, and ro/H = 0.2.
For both cases, it is seen that placing the isothermal cylinder near
the bottom heated wall results in the highest heat transfer rate in
the cavity. This behavior can be attributed to the development of
thin thermal boundary layer over the bottom wall. It is also seen
that the least Nusselt number can be obtained when the cylinder
is placed near the top wall for both Richardson numbers. Placing
the cylinder nearer the top wall reduces the constriction of the
 K. Khanafer, S.M. Aithal / International Journal of Heat and Mass Transfer 66 (2013) 200–209
Table 6
CPU time for various Richardson numbers
(Re = 100, Pr = 0.7, ro/H = 0.2).
Ri CPU (s)
0.01 16.19
0.1 19.84
1 24.36
5a 62.18
10a 84.39
a
Results from previous Ri were used to
obtained the results for new Ri.
flow near the bottom wall leading to reduced heat transfer rates
(opposite effect as with more constriction near the bottom wall).
Robustness and computational time are important consider-
ations when using simulations for design and parametric studies.
The computation time required for each Richardson number is de-
picted in Table 6 (Re = 100, Pr = 0.7, ro/H = 0.2). One can notice that
the maximum time taken by the CFD simulations is about 84.39 s
(Ri = 10) on a single processor (Intel Core, 4 GB RAM, 2.4 GHz)
using a non-uniform grid. Converged results for Ri = 1, were used
as initial conditions for Ri = 5 for increasing the stability and
decreasing the computational time. Similarly, the converged re-
sults for Ri = 5 were used as initial conditions for Ri = 10.
6. Conclusions
Two-dimensional laminar mixed convection flow and heat
transfer characteristics in a lid-driven cavity with a circular body
inside is investigated numerically for various parameters, namely,
Rayleigh number, type of boundary condition, non-dimensional ra-
dius of the cylinder, and the location of the cylinder. The results of
this study reveal that the presence of the cylinder results in an in-
crease in the average Nusselt number compared with the case with
no cylinder. This observation indicates that the cylinder can be
used as a control parameter for enhancing heat transfer character-
istics in such a configuration. Results obtained in this study indi-
cate that the placing the cylinder near the bottom wall increased
the maximum average Nusselt number while moving the cylinder
near the top wall resulted in the lowest average Nusselt number on
the bottom wall. For dominant natural convection (Ri P 2.5), the
average Nusselt number (on the bottom wall) was found to in-
crease with an increase in the non-dimensional radius for
0.05 < ro/H < 0.2. Additional increase in the non-dimensional radius
does not change the Nusselt number at a particular Ri. For domi-
nant mixed convection, the average Nusselt number increased
with an increase in the radius of the cylinder for various Richard-
son numbers. This study also showed the impact of the boundary
condition on the inner cylinder on the overall thermal field in
the cavity. It was seen that imposing an isothermal boundary con-
dition on the inner cylinder leads to a ‘‘thermal shielding’’ effect
that results in the fluid on the right side of the cavity being at a
lower temperature as compared to the no cylinder and adiabatic
cylinder case.
References
[1] K. Khanafer, A.J. Chamkha, Mixed convection flow in a lid-driven enclosure
filled with a fluid-saturated porous medium, Int. J. Heat Mass Transfer 42
(1998) 2465–2481.
209
[2] N. Alleborn, H. Raszillier, F. Durst, Lid-driven cavity with heat and mass
transport, Int. J. Heat Mass Transfer 42 (1999) 833–853.
[3] K. Khanafer, K. Vafai, Double-diffusive mixed convection in a lid-driven
enclosure filled with a fluid-saturated porous medium, Numer. Heat Transfer,
Part A 42 (2002) 465–486.
[4] K. Khanafer, A. AlAmiri, I. Pop, Numerical simulation of unsteady mixed
convection in a driven cavity using an externally excited sliding lid, Eur. J.
Mech. B Fluids 26 (2007) 669–687.
[5] R. Iwatsu, J.M. Hyun, K. Kuwahara, Mixed convection in a driven cavity with a
stable vertical temperature gradient, Int. J. Heat Mass Transfer 36 (1993)
1601–1608.
[6] A. Al-Amiri, K. Khanafer, J. Bull, I. Pop, Effect of sinusoidal wavy bottom surface
on mixed convection heat transfer in a lid-driven cavity, Int. J. Heat Mass
Transfer 50 (2007) 1771–1780.
[7] A. Al-Amiri, K. Khanafer, Fluid–structure interaction analysis of mixed
convection heat transfer in a lid-driven cavity with a flexible bottom wall,
Int. J. Heat Mass Transfer 54 (2011) 3826–3836.
[8] J.D. Jackson, B.P. Axcell, A. Walton, Mixed convection heat transfer to sodium in
a vertical pipe, Exp. Heat Transfer 7 (1994) 71–90.
[9] A. Al-Amiri, K. Khanafer, Augmentation of a heat rejection mechanism in a
ventilated enclosure filled partially with a porous medium, J. Porous Media 15
(2012) 835–848.
[10] B.S. Kim, D.S. Lee, M.Y. Ha, H.S. Yoon, A numerical study of natural convection
in a square enclosure with a circular cylinder at different vertical locations, Int.
J. Heat Mass Transfer 51 (2008) 1888–1906.
[11] J.R. Lee, M.Y. Ha, A numerical study of natural convection in a horizontal
enclosure with a conducting body, Int. J. Heat Mass Transfer 48 (2005) 3308–
3318.
[12] J.M. Lee, M.Y. Ha, H.S. Yoon, Natural convection in a square enclosure with a
circular cylinder at different horizontal and diagonal locations, Int. J. Heat
Mass Transfer 53 (2010) 5905–5919.
[13] H.S. Yoon, M.Y. Ha, B.S. Kim, D.H. Yu, Effect of the position of a circular cylinder
in a square enclosure on natural convection at Rayleigh number of 107, Phys.
Fluids 21 (2009) 1–11.
[14] I. Dagtekin, H.F. Oztop, Mixed convection in an enclosure with a vertical
heated block located, in: Proceedings of ESDA2002: Sixth Biennial Conference
on Engineering Systems Design and Analysis, 2002, pp. 1–8.
[15] M.M. Rahman, M.A. Alim, MHD mixed convection flow in a vertical lid-driven
square enclosure including a heat conducting horizontal circular cylinder with
Joule heating, Nonlinear Anal. Modell. Control 15 (2) (2010) 199–211.
[16] M.M. Rahman, M.A.H. Mamun, R. Saidur, Shuichi Nagata, Effect of a heat
conducting horizontal circular cylinder on MHD mixed convection in a lid-
driven cavity along with joule heating, Int. J. Mech. Mater. Eng. 4 (3) (2009)
256–265.
[17] M.M. Rahman, M.A. Alim, M.M.A. Sarker, Numerical study on the conjugate
effect of joule heating and magneto-hydrodynamics mixed convection in an
obstructed lid-driven square cavity, Int. Commun. Heat Mass Transfer 37 (5)
(2010) 524–534.
[18] M.M. Rahman, M.M. Billah, M.A.H. Mamun, R. Saidur, M. Hassan uzzaman,
Reynolds and Prandtl numbers effects on MHD mixed convection in a lid-
driven cavity along with joule heating and a centered heat conducting circular
block, Int. J. Mech. Mater. Eng. 2 (2010) 163–170.
[19] Y. Shih, J. Khodadadi, K. Weng, A. Ahmed, Periodic fluid flow and heat transfer
in a square cavity due to an insulated or isothermal rotating cylinder, J. Heat
Transfer 131 (2009) 1–11.
[20] V.A.F. Costa, A.M. Raimundo, Steady mixed convection in a differentially
heated square enclosure with an active rotating circular cylinder, Int. J. Heat
Mass Transfer 53 (2010) 1208–1219.
[21] M.A.H. Mamun, M.M. Rahman, M.M. Billah, R. Saidur, A numerical study on the
effect of a heated hollow cylinder on mixed convection in a ventilated cavity,
Int. Commun. Heat Mass Transfer 37 (2010) 1326–1334.
[22] A.W. Islam, M.A.R. Sharif, E.S. Carlson, Mixed convection in a lid driven square
cavity with an isothermally heated square blockage inside, Int. J. Heat Mass
Transfer 55 (2012) 5244–5255.
[23] H.F. Oztop, Z. Zhao, B. Yu, Fluid flow due to combined convection in lid-driven
enclosure having circular body, Int. J. Heat Fluid Flow 30 (2009) 886–901.
[24] M.M. Billah, M.M. Rahman, U.M. Sharif, N.A. Rahim, R. Saidur, M.
Hasanuzzaman, Numerical analysis of fluid flow due to mixed convection in
a lid-driven cavity having a heated circular hollow cylinder, Int. Commun.
Heat Mass Transfer 38 (2011) 1093–1103.
[25] ADINA, Theory and Modeling Guide, vol. III: ADINA CFD & FSI, 2011, pp. 265–
278.