International Journal of Heat and Mass Transfer 49 (2006) 3500–3513

www.elsevier.com/locate/ijhmt

Numerical simulation of liquid metals in differentially heated

enclosure undergoing orthogonal rotation
M.F. Baig *, M. Zunaid
Department of Mechanical Engineering, Aligarh Muslim University, Aligarh 202001, India

Received 6 September 2005

Available online 19 May 2006

Abstract

In this work a numerical investigation has been performed to examine the characteristics of mixed convective heat transfer in square
enclosures undergoing orthogonal rotation i.e. rotation axis and gravity axis are orthogonal to each other. A semi implicit finite differ-
ence code on a collocated grid is used to solve the momentum and energy equations subject to Boussinesq approximation. The study is
carried out for a wide range of operating parameters such as Rayleigh number (Ra), Taylor number (Ta), Rotational Rayleigh number
(Raw) for a fixed Prandtl number (Pr). The numerical experiments have been carried out for a fixed Pr = 0.01, Ra varies from 105 to 107
while Ta and Raw vary from almost 0 to 109. Results reveal that significant increase or decrease in heat transfer rates can be achieved by
the rotational effects, mainly influenced by centrifugal force.
 2006 Elsevier Ltd. All rights reserved.

Keywords: Liquid metals; Coriolis force; Centrifugal force; Orthogonal rotation; Convection; Heat transfer

1. Introduction cient in elliptical geometry ducts, rotating about a parallel

The effect of cavity rotation on flow stability as well as
heat transfer is particularly very important concerning
the flow of liquid metals inside a cavity, which is the stan-
dard material for the manufacture of single wafer crystals
in the semiconductor industry. Though a vast amount of
work, both theoretical and experimental, exists in the open
literature related to heat and mass transfer in rotating ducts
[1–4] for cases of either air and water, yet little or no work
has been done, for low Prandtl number fluids especially in
configurations like square ducts undergoing orthogonal
rotation.
Some of the results pertaining to either air and water in
rotating ducts, channels etc. are summarised herewith. Ngu-
yen et al. [5] have shown that the effect of counter rotating
the cavity results in growth of good quality crystals. Maha-
devappa et al. [6] found that the heat transfer is more effi-
*
Corresponding author. Tel.: +91 9897517071.
E-mail address: drmfbaig@yahoo.co.uk (M.F. Baig).
axis situated parallel to and away from the central axis, then
in rectangular ducts under similar operating conditions.
Mixed convection heat and mass transfer have been studied
by Lee et al. [7] considering uniform temperature and con-
centration conditions along the porous duct walls. They
concluded that the combined thermal and solutal buoyan-
cies enhance the heat and mass transfer rate along the wet-
ted wall. Theoretical and experimental examination of the
effect of Coriolis force on the flow structures for adiabatic
rotating channels was performed by Hwang and Jen [8]
and it was found that at low rotation rates, a double vortex
secondary flow appears in the transverse plan of the channel
while at large rotation rates, an instability occurs in the
form of longitudinal rolls in the interior of the channel.
Fann and Yang [9] studied the effect of centrifugal buoy-
ancy on the fluid flow and heat transfer characteristics in
radially rotating channels for steady laminar flow and they
came to the conclusion that the Coriolis force causes the
generation or decay of one or multiple pairs of transverse
velocities. Turbulent heat convection with fully developed

0017-9310/$ - see front matter  2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijheatmasstransfer.2006.02.046
 M.F. Baig, M. Zunaid / International Journal of Heat and Mass Transfer 49 (2006) 3500–3513 3501

Nomenclature

L
AR H aspect ratio x, y dimensional coordinate, m
Cp specific heat at constant pressure, kJ X, Y dimensionless coordinate
kg K
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 2
CEF centrifugal force, Raw Prh X þ Y Greek symbols
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
COR Coriolis force, Pr TaðU 2 þ V 2 Þ a coefficient of thermal expansion, K1
DT temperature difference between hot and cold
g magnitude of acceleration due to gravity, m s2 walls
H depth of fluid layer, m
s dimensional time, s
i, j indices in the X and Y directions T c
h non-dimensional temperature, T DT
L width of fluid layer, m
thermal diffusivity, ms
2
Nu space-averaged Nusselt number j
N kg
pm motion pressure, m 2
q density, m3
q0 reference density
Pm dimensionless motion pressure
kinematic viscosity, ms
2
m
Pr Prandtl number, jm
3 Xd dimensional speed of rotation, rad
s  
Ra Rayleigh number, agHmjDT 2
ðX2 H ÞH 3 aDT X non-dimensional rotation speed, Xd Hj
Raw rotational Rayleigh number, d ma
t dimensionless time
Subscripts
T dimensional temperature, K
c cold wall
4X2 H 4
Ta Taylor number, md2 h hot wall
TBF thermal buoyancy force, RaPrh m motion
v (u, v) dimensional velocity field in x and y direc- 0 reference
tions, ms
? perpendicular
V (U, V) dimensionless velocity field in X and Y
directions

velocity and temperature fields was studied by Mori et al. The principal motivating factor to study this complex
[10] in a circular pipe rotating around an axis perpendicular flow of rotating liquids metals is due to the scarcity of lit-
to its own axis. They found that the increase of Nusselt erature available pertaining to this configuration, to the
number is not significant for turbulent regions of the flow
as compared to laminar regions. Yan [11] numerically stud-
ied the characteristics of laminar mixed convection in rect-
angular ducts rotating about a parallel axis with water film U=V=0 and ∂θ/∂Y = 0

evaporation along the porous ducts walls and concluded

that heat and mass transfer is enhanced because of the for-
(c)
mation of secondary vortices in the duct. Y
Song et al. [12], studied the fluid flow and heat transfer
model for high speed rotating heat pipe, with substantial
centrifugal accelerations. They found that the natural con-
vection in the liquid film becomes more significant at higher U=V=0 U=V=0
accelerations and fluid loadings. Yang and Wang [13] stud-
and and
ied the bifurcation structure and stability of combined free Ω X
and forced convection in a rotating curved duct of square θ = 1.0 θ=0
cross-section. Results showed that as the relative strength
of buoyancy force decreases, temporal oscillations change
from periodic to non-periodic and finally chaotic. Ker and
Lin [14] in their study have emphasized that the study of (b)
flow behavior for case of liquid metals in rotating cavities
needs further investigation. The major difference between (a)
flow of liquid metals to comparatively larger Prandtl num-
U=V=0 and ∂θ/∂Y = 0
ber fluids such as air, is that the former fluids have higher
density and hence are subjected to large centrifugal forces Fig. 1. Schematic diagram of a square enclosure showing boundary
per unit volume as compared to the latter. conditions and location of spatial points (a), (b) and (c).
3502 M.F. Baig, M. Zunaid / International Journal of Heat and Mass Transfer 49 (2006) 3500–3513

best of our knowledge. The present analysis takes into to thermal and rotational buoyancy forces. By assuming
account both the Coriolis and centrifugal forces combined the Boussinesq approximation in both the gravitational
with thermal buoyancy in order to analyse their effects on and centrifugal forces, the thermal, centrifugal and Corio-
dynamics of flow and heat transfer. lis forces acting on the flow are respectively, equal to
q0ga(T  T0), q0a(T  T0)Xd · Xd · r and 2q0Xd · V.
2. Mathematical formulation As the gravity, centrifugal and Coriolis forces all lie
within the cross-sectional plane of the cavity and on further
The schematic diagram of the physical system under assuming a long square duct thereby neglecting the
investigation is depicted in Fig. 1. Initially at time s < 0, end-effects of 3D cavity flow, the problem can be analysed
the liquid inside the enclosure is rotating anti-clockwise as a two-dimensional flow problem. Thus it is felt that two-
as a solid body at a constant angular speed X, and is iso- dimensional calculations provide a good platform for
thermal at temperature T0. At time s P 0, the vertical understanding the basic flow dynamics and the roles played
side-walls are suddenly raised and lowered to uniform tem- by the time-varying gravitational buoyancy, rotational
peratures Th = T0 + DT/2 and Tc = T0  DT/2 while the buoyancy and Coriolis forces in affecting the flow struc-
top and bottom walls are thermally insulated. Thus, the tures. Due to Boussinesq approximation the problem and
liquid metal inside the cavity is simultaneously subjected consequently the results are valid for configurations which

11 12
9 10
78 13 Level PSi Level Theta
6 11

4
5 14 15 5.88656 12 15 0.9375
0.4 4 12 0.4
9 10 14 4.07412 14 0.875

86
23
3

12
67 8 11 11
2 5 13 2.26169 13 0.8125

10
11

12 0.44925 12 0.75
3

11 3
1
14 15

10 9
7
0.2 0.2
4

11 -1.36319 10
11 0.6875
9 8 10
4

10 -3.17562 10 0.625

11
9

51
2

9 -4.98806 9 0.5625
1
7 68
5

9
3

8 -6.8005 8 0.5
Y

0 0
9
876 10

4
7 -8.61294 8 7 0.4375
8
109

10 12 8
2

6 -10.4254 7 6 0.375

6
11

32
4

7
5 -12.2378 5 0.3125
5

-0.2 4 -14.0502 -0.2 4 0.25

9
3

6
13 11

5
7

3 -15.8627 3 0.1875
4

5
1514

67 2
8 2 -17.6751 5 2 0.125
4 3 1 -19.4876 1 0.0625
-0.4 -0.4

1
13 1 10 9 4
11

3 5 5
11

786 9 3
4

12 11 10

-0.5 -0.25 0 0.25 0.5 -0.5 -0.25 0 0.25 0.5

X X
(a) t = 0.10 (d) t = 0.10

12
15

10 11
12
Level PSi Level Theta
1

9 8 7 9 15 15 4.71187 15 0.9375
11

0.4 5 6 8 0.4
3
10

12
14 2.91575 14 0.875
11

2
10

4
43

13 11
1213 14
6

11 13 1.11963 11 13 0.8125
2 5
109

12 -0.6765 12 0.75
7

10
7
4

0.2 11 -2.47262 0.2 11 0.6875

12

12
7
5
10

9
8

10 -4.26875 10 0.625
6
9

11

9
9 -6.06487 9 0.5625
6

3 8
10
8

8 -7.861 8 0.5
Y

Y
15

0 4 0
9 8

10

1
23

7 -9.65712 7 0.4375
8

6 -11.4532 6 0.375
3
11

2
8
6

7
13 14

5 4
1

5 -13.2494 5 0.3125
11

3
12
7
9

-0.2 4 -15.0455 -0.2 4 0.25

6
2

5 6
2

2 3 -16.8416 3 0.1875
3

12
1

5
12

2 -18.6377 2 0.125
7

4
101
8
6

2
6
9

-0.4 3 4 13 1 -20.4339 -0.4 1 0.0625

4

11 12 10 8 6 5 8
9 7 10
4
15

11 12

-0.5 -0.25 0 0.25 0.5 -0.5 -0.25 0 0.25 0.5

X X
(b) t = 0.30 (e) t = 0.30
13
10 11 Level PSi Level Theta
14

12
8 7 12
13

6 9
0.4 5 15 2.58781 0.4 15 0.9375
4

4
12 1 14

2 8
3

3 14 0.978625 14 0.875
9

23
5

4 12
10
11

11
1311

13 -0.630562 13 0.8125
3

11
1

12 -2.23975 12 0.75
5 8
10
7

7
13

10
0.2 11 -3.84894 0.2 10 11 0.6875
9
9

15

9
10 -5.45812 10 0.625
9
12
8

9 -7.06731 9 0.5625
6

1011

8
5
12

7 8
8 -8.6765 8 0.5
4
Y

0 0
2

7 -10.2857 7 0.4375
3

7
11

14

4
6
8

6 -11.8949 6 0.375
13 12
3

6 4
2 2

5
13
1110
1

5 -13.5041 5 0.3125
3
2

-0.2 4 -15.1132 -0.2 6

4 0.25
1
7
9
7

1
4

3 -16.7224 3 0.1875
5
13

6 5
9

5 2 -18.3316 2 0.125
9

10 8 6
15

11
12
7

-0.4 5 1 -19.9408 -0.4 1 0.0625

14

4 3 4
11

12 9 7 6
10

12 10 8
13 11 4
6

13
-0.5 -0.25 0 0.25 0.5 -0.5 -0.25 0 0.25 0.5

X X
(c) t = 0.40 (f) t = 0.40

Fig. 2. (a), (b) and (c) show streamline contours while (d), (e) and (f) show isotherm contours at different time instants for Ra = 107, Ta = 8.16 and
Raw = 1.02 · 102.
 M.F. Baig, M. Zunaid / International Journal of Heat and Mass Transfer 49 (2006) 3500–3513 3503

5
0.97

0.93

θ
-5

0.89
-10

-15 0.85
0.1 0.2 0.3 0.4 0.1 0.2 0.3 0.4
t t
(a) (b)

1200
0.02

900

Power (θ)
Power (U)

600
0.01

300

0 0
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.025 0.05 0.075

f / fs f / fs
(c) (d)

Fig. 3. (a) and (b) show time history while (c) and (d) show power spectrum of horizontal velocity component (U) and temperature (h) at spatial point (a)
for Ra = 107, Ta = 8.16 and Raw = 1.02 · 102.

4 8 2 5 9 31
412 1 4
7
7 11 6
1510 137 2 Level COR 10 Level CEF
2
3
1 2

6 11 12
9

9 8 10 8 3
15 7.73906
4

8
4

0.4 5 76 9 5
0.4 15 0.676406
3

43 7
5 14 7.22313 14 0.631312
11

2 43 8
8

6
9
9

12

13 6.70719 13 0.586219
95

1213
2
76

12 6.19125 5
3 5

1 7 12 0.541125
6

5
8

0.2 11 5.67531
6

4
8 10

0.2 11 0.496031
7
1

3
6

10 5.15938
4
8

7
45

10 0.450937
1

6
4

8 10

9 4.64344 9 0.405844
1 7

1
9

8 4.1275
13

8 0.36075
63

9 3 5144
2

2
3
Y

0
Y
11 9

0
1

7 3.61156
10

7 0.315656
2

54
3

4
9 6

215 12

6 3.09563 6 0.270562
11
9
5

2
12
6 71 11
1

5 2.57969 5 0.225469
5

6 7

4 2.06375
4

1 4 0.180375
-0.2
2
3

-0.2
7

3 2 3 1.54781
3

3 0.135281
8

8
6
10
1

54
6

2 1.03188 2 0.0901875
6

10
7

8 10 1 0.515938 1 0.0450937
10
74

3 42 10
-0.4 67 7 6 5
-0.4 1
8

8
3

2
4

9 11
5

3
4

5 12 10 11
1

3 14 9
15

13 8
14

5 43
13

12
5

1 2 67 124 1

-0.5 -0.25 0 0.25 0.5 -0.5 -0.25 0 0.25 0.5

X X
(a) (b)

15
6

13 Level TBF
3
1
12

0.4 15 93750
12 9
14 87500 Hot
13 81250
Cold .............
11

11 13
11
12

12 75000
14

10
10

0.2 11 68750
10

9
15

10 62500
2

9
8 9 56250 11
13

8 50000
6

Nu

8
Y

0
13

7 43750
5
7

6 37500
9 11
6

7 9
5 31250
8

4 25000
1214

-0.2
7

6
3 18750
10

2 12500 7
6

5
5

2
15

1 6250
-0.4
13

5
3

7
1

4 5
6

-0.5 -0.25 0 0.25 0.5 0.1 0.2 0.3 0.4

X t
(c) (d)

Fig. 4. Depicts spatial variation of (a) Coriolis force, (b) centrifugal force, (c) thermal buoyancy force at t = 0.4 and (d) variation of Nusselt number with
time for Ra = 107, Ta = 8.16 and Raw = 1.02 · 102.
3504 M.F. Baig, M. Zunaid / International Journal of Heat and Mass Transfer 49 (2006) 3500–3513

have small DT and small scale heights compared to the where

large-scale geophysical systems. opm op
The governing equation of mass, momentum and energy ¼  þ q0 X2d x  q0 g sin Xd s
ox ox
can be expressed in the dimensional form for 2D incom- opm op
pressible flow as follows: ¼  þ q0 X2d y  q0 g cos Xd s ð4Þ
oy oy
ou ov DT
þ ¼0 ð1Þ ¼ jr2 T
ox oy Ds
Du 1 opm No-slip boundary conditions are imposed for the velocity
¼ þ aðT  T 0 Þg sin Xd s þ mr2 u þ 2Xd v components at all the walls. T is assumed to satisfy thermally
Ds q0 ox
conducting boundary conditions at the vertical side walls,
 X2d aðT  T 0 Þx ð2Þ
Dv 1 opm while Neumann boundary conditions oT oy
¼ 0 are assumed
¼ þ aðT  T 0 Þg cos Xd s þ mr2 v  2Xd u for the horizontal top and bottom walls as shown in Fig. 1.
Ds q0 oy
On performing non-dimensionalization using reference
 X2d aðT  T 0 Þy ð3Þ length, velocity, time and pressure scales as H ; Hj ;

10
12 1314 Level PSi Level Theta

8
11 12

11

5
10 98 10 15 6.22562 9 15 0.9375

4
0.4 7 9 0.4 10

14
6

11
14 3.48725 14 0.875
11

6
8

5 13 0.748875 13 0.8125
7

11 9
12

73
12 -1.9895 12 0.75

10
8 10

13
3
2

0.2 11 -4.72787 0.2 11 0.6875

5
6

9
12
6

12
5

10 -7.46625 15 8 10 0.625
1

10

9 -10.2046 9 0.5625
7

11
9
12

7
4
9

3
5

8 -12.943 8 0.5
Y

0 0
9

5
8
7 -15.6814 7 0.4375

4
8
4
2

6 -18.4197 6 0.375
14
10

6 7
5 -21.1581 7 5 0.3125
11

8
-0.2 -0.2
5

8 10

4 -23.8965 4 0.25

3
12
3

13

14 3 -26.6349 3 0.1875
8
6

4 2 -29.3732 2 0.125
11 13

1 2
5
12 15

12 9 7 6 7 5
10 1 -32.1116 1 0.0625
-0.4 10 9
-0.4 11 6 5
8 4
13 11
12

-0.5 -0.25 0 0.25 0.5 -0.5 -0.25 0 0.25 0.5

X X
(a) t = 0.20 (d) t = 0.20 2
4

13 Level PSi Level Theta

12

11
14

5
6

10 12
8 11
14

0.4 7 15 3.21187 15 0.9375

14 3

0.4
8

5 9
13

10

14 0.34975 14 0.875
1

11
13

6
9
12

10
8

7
4 13 -2.51238 13 0.8125
6
34

12 -5.3745 12 0.75
10
2

11
8
10

12
3

0.2 11 -8.23663 0.2 11 0.6875

13

2
1
9

10 -11.0988 10 0.625
8

9
11

14 15

9 -13.9609 9 0.5625
7
3 42 5
8
12
9

8 -16.823 8 0.5
Y

0 0
6
7
6
5

8
7

7 -19.6851 8 7 0.4375
10
13

6 -22.5473 6 0.375
11

13

3
8

5 -25.4094 5 0.3125
7
4
7

-0.2 6
4 -28.2715 -0.2 4 0.25
10

10

1
11
8

14 2
12

3 -31.1336 3 0.1875
5

11 3
5 4 7
2 -33.9958 8 2 0.125
9 9
1

8
-0.4 1 -36.8579 -0.4 7 1 0.0625
15

8 7 6
13

14 9
3
1415

7
6

10 6 5
12

11
2

12
13 14

-0.5 -0.25 0 0.25 0.5 -0.5 -0.25 0 0.25 0.5

X X
(b) t = 0.40 (e) t = 0.40

14
12 13 Level PSi Level Theta
11 10 11
15 2.24031 15 0.9375
14

0.4 8 0.4 9
14

9 7 10
9

14 -1.15838 8 14 0.875
10
13

9
11

4 6 8
13 -4.55706 13 0.8125
23
12

15

3
12

5
10

12 -7.95575 12 0.75
13
6

2
5

0.2 0.2
14

11 -11.3544 11 0.6875
12
8

1
14

13

10 -14.7531 10 0.625
7

1
65 7
10

10

9 -18.1518 9 0.5625
8

8
11

9
4

8 -21.5505 8 0.5
Y

0 0
11

9
14

7 -24.9492 7 0.4375
14

8
2

6 -28.3479 6 0.375
10
11
9

7
6
2

1
5 -31.7466 5 0.3125
12 1

3
9

32

-0.2 -0.2
15
1213 14

4 -35.1453 4 0.25
3

5
11

5
8 7 3 -38.5439 3 0.1875
1312

4 7 7
8
10 6 2 -41.9426 7 2 0.125
8 10 6 6
15
91

1 -45.3413 1 0.0625
1
1

-0.4 -0.4
5

14 15 9
12
14

11
13 14

-0.5 -0.25 0 0.25 0.5 -0.5 -0.25 0 0.25 0.5

X X
(c) t = 0.60 (f) t = 0.60

Fig. 5. (a), (b) and (c) show streamline contours while (d), (e) and (f) show isotherm contours for different time instants for Ra = 107, Ta = 8.16 · 104 and
Raw = 1.02 · 106.
 M.F. Baig, M. Zunaid / International Journal of Heat and Mass Transfer 49 (2006) 3500–3513 3505

10
0.99

0 0.96

θ
-5

-10 0.93

-15

0.9
0.2 0.4 0.6 0.8 0.2 0.4 0.6 0.8

t t
(a) (b)

800 0.02

0.0175

600 0.015

0.0125

Power (θ)
Power ( U)

400 0.01

0.0075

200 0.005

0.0025

0 0
0.01 0.03 0.05 0.07 0.01 0.02 0.03 0.04 0.05

f / fs f / fs
(c) (d)

Fig. 6. (a) and (b) show time history while (c) and (d) shows power spectrum of horizontal velocity component (U) and temperature (h) at spatial point (a)
for Ra = 107, Ta = 8.16 · 104 and Raw = 1.02 · 106.

3 12 9 10 5 118 7 43 12 1
Level COR Level CEF
2
13

6
3

8
11
12

6 2 3
0.4
4 11 15 830.344 7 7
15 6764.06
5
4

0.4
9
12

12 13

9 11
14 774.988
4
8 7

5
2

10 10 6 14 6313.13
5 78 6
5

13 719.631 13 5862.19
5

4 5
6

4 3
9 10

2 12 664.275
8
7 8

12 5411.25
34
6

8
103

9
5

2
0.2 11 608.919
4

0.2 3 11 4960.31
2
5
11

10 553.563
4

10 4509.38
2
219

9 498.206
2

9 4058.44
1
9

1 1
8 442.85 8 3607.5
6

9
3

11

3
3
7
5
Y

12

0
Y

7 387.494 0 7 3156.56
10
78 6
4

6 332.138
10 11
10

6 2705.63
7

43
5

4
12

2
4
3

5 276.781 5 2254.69
6
54 3

1
9

1 4 221.425 2
4 1803.75
2

78

10 8 9

-0.2 7 -0.2
9

9
4
5

2 1 3 166.069
8 3 1352.81
11 1

1121 54 3
4
5

2 110.713 2 901.875
1 1 62

5
0

6 14 3 10
7 8
10 6 1 55.3563 1 450.938
5 9 7
-0.4 8
-0.4
5 9

7
1

11
3

12

5
6

4
11
13

4 13 4
3 59 2
12
14 11 2
6 12 71 8 43 3
1

2 10 3
2

1
3

-0.5 -0.25 0 0.25 0.5 -0.5 -0.25 0 0.25 0.5

X X
(a) (b)
12
12

Level TBF
14

15 93750
7

3 14
0.4
Hot
10

14 87500
8

11
11
15

10 13 81250
Cold ............
9

12 75000 12
5
9

0.2 11 68750
12 13

10 62500
6

10
9

9 56250
4
7

8 50000
14

21

Nu
7
8
Y

0 7 43750
8 8
3
11 10

6 37500
7
15

5 31250
8

-0.2 4 25000 6
3 18750
5
9
13 12

8
2 12500
6

7
7 1 6250 4
8 6
4

-0.4 7
14 5
2
1

10
2
3

-0.5 -0.25 0 0.25 0.5 0.0001 0.5001 1.0001

X t
(c) (d)

Fig. 7. Depicts spatial variation of (a) Coriolis force, (b) centrifugal force, (c) thermal buoyancy force at t = 0.4 and (d) variation of Nusselt number Nu
with time for Ra = 107, Ta = 8.16 · 104 and Raw = 1.02 · 106.
3506 M.F. Baig, M. Zunaid / International Journal of Heat and Mass Transfer 49 (2006) 3500–3513

2
H2
j
; and qj
H2
respectively, the governing equations, initial The initial conditions are
and boundary conditions, become At t < 0 U ¼V ¼0 and h ¼ 0:5 for all X and Y
oU oV
þ ¼0 ð5Þ While the boundary conditions become
oX oY
DU oP m
¼ þ Prr2 U þ RaPrh sinðXtÞ þ Ta0:5 PrV At t P 0
Dt oX
oh
 Raw PrX h ð6Þ X ¼ 0; 1; Y ¼ 0; U ¼ V ¼ 0; ¼0
oY
DV oP m oh
¼ þ Prr2 V þ RaPrh cosðXtÞ  Ta0:5 PrU X ¼ 0; 1; Y ¼ 1; U ¼ V ¼ 0; ¼0
Dt oY oY
 Raw PrY h ð7Þ Y ¼ 0; 1; X ¼ 0; U ¼ V ¼ 0; h¼1
Dh Y ¼ 0; 1; X ¼ 1; U ¼ V ¼ 0; h¼0
¼ r2 h ð8Þ
Dt
In Eq. (8) we have neglected viscous dissipation and The space-averaged Nusselt number at both the cold
pressure work due to incompressibility of the flow. and hot walls has been calculated using the relation

9
Level PSi Level Theta

14
9 11 9 2
1011 15 25.3756 12
3 15 0.9375
0.4 0.4
1132

1
9 14 21.4312 5 14 0.875

12 13
10

7
1414

13 17.4869

10
8 13 0.8125
8

15
9

12
13

6
12 11
12 13.5425 12 0.75
0.2 0.2
5 7

4
11 9.59812 11 0.6875
6
3

8
6 11
7

10 5.65375 10 0.625
10
9

13
14

9
2
9 1.70937 9 0.5625
5

11
4

8 -2.235 8 0.5

2
Y

0 1 0

35 1
7 -6.17938 7 0.4375
3

7
10

8
6 -10.1238 6 0.375

10
12
8
7

6
5 -14.0681 5 0.3125
5

-0.2 -0.2
15

4 -18.0125 4 0.25

4
8

6
3
4

3 -21.9569 3 0.1875
7

1
9

2
3 2 -25.9013 9 2 0.125
8

1113

5 4 5 10
1 -29.8456
14

-0.4 -0.4 8 9 1 0.0625

10

8 7
6 5
7 6 3 2
8 8

-0.5 -0.25 0 0.25 0.5 -0.5 -0.25 0 0.25 0.5

X X
(a) t = 0.10 (d) t = 0.10

5 6
5
6 Level PSi 8 1 Level Theta
13

9
0.4
7
6 15 20.8185 0.4 3 15 0.9375
5

6 10 5
8 9
10 8 7 14 18.827 14 0.875
15

11 11
12 13 16.8355 13 0.8125
4

13
9

12 14.844 12 0.75
14

7
6
12

14
2

0.2 0.2 11
10
9

11 12.8525 11 0.6875
8
12

10 10.861 12 10 0.625
11
10

9 8.8695 9 0.5625
10

15

1
9
5

8 6.878 8 0.5
Y

0 0
13

15

12

6
11

13

7 4.8865 7 0.4375
8

5
11
14

7
8

6 2.895 6 0.375
7
10

13

5 0.9035 5 0.3125
13

15

-0.2 -0.2
4

12
4

14 4 -1.088 4 0.25
6 7
6

14
9

10

3 -3.0795 3 0.1875
5
9

2
12

3 12 13
3

2 -5.071 2 0.125
4

10

11
4

5
-0.4 8
7
6
4
1 -7.0625 -0.4 1 0.0625
2

11

1
4

15

-0.5 -0.25 0 0.25 0.5 -0.5 -0.25 0 0.25 0.5

X X
(b) t = 0.30 (e) t = 0.30

9 9 Level PSi Level Theta

10 15 12.3781 5 15 0.9375
1

11
0.4 10 12 0.4
13
14

14 14 10.4863 11 14 0.875
15

15
12

13 8.59438 12 13 0.8125
9

13 12
11

4
9 1

67

12 6.7025 12 0.75
0

0.2 0.2 13
11 4.81063 11 0.6875
9

14

13
8

15
14 10 2.91875 10 0.625
2
9

12 13 11
11
10 12 10 9 1.02688 9 0.5625
3

8 9
7 65 7 8 14
56 8 -0.865 8 0.5
Y

0 3 4 0
2
8

1 7 -2.75687 7 0.4375
5

1
11
13

13
1
3
4

6 -4.64875 6 0.375
15
7

12

5 -6.54062 5 0.3125
7 9
12 13
4

-0.2 2 -0.2 10
11

12
6

4 -8.4325 4 0.25
6

6
3 -10.3244 3 0.1875
4

5 8
14

1
2 -12.2162 11 2 0.125
8

2 3 4
-0.4 1 -14.1081 -0.4 10
9 5
3 1 0.0625
5 6 7
8
10

7 7 8 6
8
4

-0.5 -0.25 0 0.25 0.5 -0.5 -0.25 0 0.25 0.5

X X
(c) t = 0.40 (f) t = 0.40

Fig. 8. (a), (b) and (c) show streamline contours while (d), (e) and (f) show isotherm contours at different time instants for Ra = 107, Ta = 8.16 · 106 and
Raw = 1.02 · 108.
 M.F. Baig, M. Zunaid / International Journal of Heat and Mass Transfer 49 (2006) 3500–3513 3507

Z 1
1 oh mean Nu number changed by less than 3% while qualitative
Nu ¼  dY ð9Þ change in spatial patterns remains is insignificant. Hence in
H 0 oX
order to cut down the computational time, we ran all our
The numerical scheme used to solve the equations has simulation cases on a coarser grid of 71 · 71. We have con-
been explained in detail in Hasan and Baig [15]. In this sidered three spatial points (a), (b) and (c) having coordi-
scheme, the non-linear convective terms in Eqs. (6)–(8) nate locations (X, Y) = (0.4933, 0.4933), (0.2816,
have been evaluated using third-order accurate upwind 0.2816) and (0.372, 0.372) on which temporal history of
scheme in the interior with first-order upwind scheme near several dynamical variables has been recorded.
the walls. The viscous diffusion terms are discretized using
the three-point central differencing stencil, while time inte-
2.1. Validation
gration has been performed explicitly using Adam-Bash-
forth second-order accurate scheme in order to capture
The numerical code developed has been validated by
the unsteady physics associated with the oscillatory flow
solving the glazing problem i.e. buoyancy driven flow
regime. A non-uniform collocated mesh of 71 · 71 grid
point has been generated, with the minimum spacing of Table 1
0.002 near the walls and maximum spacing of 0.046 near Validation of code with numerical results of Nonino and Croce [16]
the center of the cavity. The mass continuity has been Ra jWmaxj jUmaxj jVmaxj Nuavg Reference
enforced by solving Poisson equation for pressure, using 5
10 9.652 35.668 68.367 4.483 Present
Strongly Implicit Procedure (SIP), as implemented in [17] 105 9.618 34.749 68.646 4.521 Benchmark [16]
with a significantly small L2-norm tolerance limit of 109. 106 16.948 65.589 221.488 8.749 Present
Regarding grid-independence, we simulated the flow on 106 16.817 64.827 220.630 8.825 Benchmark [16]
a grid of 91 · 91 keeping the location of first grid point 107 30.638 145.460 697.669 16.284 Present
constant and found that the integral parameters such as 107 30.165 148.590 699.670 16.522 Benchmark [16]

1.05
20

10
1

0.95
θ

-10
U

-20
0.9
-30

-40
0.85

9.9E-05 0.500099 1.0001 9.9E-05 0.500099 1.0001

t t
(a) (b)

2000 0.01

1500
Power (θ)
Power ( U)

1000 0.005

500

0 0
0.025 0.05 0.075 0.1 0.025 0.05 0.075 0.1
f / fs f / fs
(c) (d)

Fig. 9. (a) and (b) show time history while (c) and (d) show power spectrum of horizontal velocity component (U) and temperature (h) at spatial point (a)
for Ra = 107, Ta = 8.16 · 106 and Raw = 1.02 · 108.
3508 M.F. Baig, M. Zunaid / International Journal of Heat and Mass Transfer 49 (2006) 3500–3513

inside a square cavity with differentially heated vertical For Ra = 107 at low rotational speeds (Ta = 8.16 and
walls and horizontal adiabatic walls. The results have been Raw = 1.02 · 102), corresponding to non-dimensional rota-
compared with the benchmark solutions of Nonino and tion X = 0.014, the thermal buoyancy is the dominating
Croce [16]. The range of Ra considered is between 105 force and this is also evident from Fig. 2 which shows
and 107 while Pr is taken to be 0.71. Table 1 shows the multi-cellular roll pattern of stream function. The isotherms
comparison between the results of present study and show thin boundary layers near isothermal walls with stable
benchmark solution. The minimum and maximum grid stratification of the core. The dynamical variables U and h
spacing employed was 0.004 and 0.048 respectively, with in Fig. 3 exhibit a chaotic time-signal at both the co-ordi-
total number of grid points being equal to 55, in both spa- nate points (a) and (b). The corresponding power spectrum
tial directions as in the benchmark study. The maximum too exhibits a broad range of frequencies at both coordinate
percentage error in jWmaxj, jUmaxj, jVmaxj and Nuavg are points (a) and (b). These features characterize a strong tur-
respectively equal to 1.568%, 2.106%, 0.406% and 1.440% bulent flow in the cavity. The spatial distribution of forces
respectively. in Fig. 4 shows thermal buoyancy is the dominating force
compared to smaller magnitudes of Coriolis and centrifugal
3. Results and discussion forces. From Fig. 4(d), it is clear that spatially averaged Nu
shows oscillatory heat transfer at both the walls with a tem-
3.1. Effects of increasing centrifugal and Coriolis forces poral mean value of 8.5 approximately.
As the rotational speed is increased corresponding to
In order to analyse the effects of increasing centrifugal non-dimensional X = 1.428 (Ta = 8.16 · 104 and Raw =
and Coriolis forces, we have performed several runs of 1.02 · 106), the stream functions in Fig. 5 again show
the code with increasing rotational speed X, keeping the formation of multiple rolls of slightly larger magni-
Ra = 107 as constant. tude compared to the previous case. The isotherms depict

1 1
2

1 2 Level COR 9 Level CEF

1 3 1

11

2
4 3 3

1
5 15 7924.69 15 676406
5

0.4 3
0.4
2

6 5 1 14 7396.38 7
4 14 631313
2

3 5 4
13 6868.06
1

1 13 586219
46

2
3

13

5 6
5 6
2 12 6339.75 12 541125
2

4
6
7 2

10

0.2 11 5811.44 4
4

1 0.2 11 496031
8

2
1

5 4 3 10 5283.13 10 450938
4

6 7 5
3

9 8 9 4754.81 9 405844
2

11 10
1

7
8 4226.5

2 3
8 8 360750
3
11

0
Y

7 9
161 7

0
Y

6 8 7 7 3698.19

1
4 5 6 7 315656
9

3
7

5
2
2

3 4 6 3169.88 6 270563
4

4
5

1
13
12

3 615

5 2641.56
2
83 5

5 225469
2

4
15

6
5

5 5
4 2113.25
10

-0.2 -0.2 4 4 180375

6

3 1584.94 3 135281
3

10
14
10 4

1
8
2

2 1056.63
2 4 7

3
1 2 90187.5
3
4

4
7

2 1 528.313
5

1 45093.8
1
5

-0.4 3
1

-0.4
9

5 4 2
3 1
4 2 6
11

6 7 4
24 8 24
7

6 1 3 5 516 3

-0.5 -0.25 0 0.25 0.5 -0.5 -0.25 0 0.25 0.5

X X
(a) (b)

Level TBF 20
4
15 93750
0.4 Hot
8

12 14 87500 18
11 1
0 13 81250 Cold ...............
2
7
9

12 75000 16
12

0.2 13 11 68750
35 6

13
14
14

10 62500
1

9 56250
15

12

8 50000 12
Y

Nu

0
4

14
13

7 43750
8
7 0 1

6 37500 10
1 1

12 13
13

5 31250
9

-0.2 11 4 25000 8
12 3 18750
5

6 2 12500 6
10

11
15

10
1 6250
1

8
-0.4 9
7
4
4
1412
9

2
-0.5 -0.25 0 0.25 0.5 0.0001 1.0001
X t
(c) (d)

Fig. 10. Shows spatial variation of (a) Coriolis force, (b) centrifugal force, (c) thermal buoyancy force at t = 0.4 and (d) variation of Nusselt number Nu
with time for Ra = 107, Ta = 8.16 · 106 and Raw = 1.02 · 108.
 M.F. Baig, M. Zunaid / International Journal of Heat and Mass Transfer 49 (2006) 3500–3513 3509

beginning of unstable stratification in the core signifying the enclosure. These rolls are primarily formed due to the
increase in convective motion in the core of the cavity. action of centrifugal force on the fluid. Isotherms in
The dynamical variables U and h in Fig. 6 exhibit chaotic Fig. 8(d)–(f) show generation of mushroom-shaped struc-
time-signal trace at all and the power spectra reinforce that ture in the core of the cavity which leads to strong unstable
view. The spatial variation of forces show (Fig. 7) stronger stratification of the core. The time history in Fig. 9 depicts
Coriolis and centrifugal forces near the walls but still their a stronger turbulent time-signal at all the co-ordinate
magnitude is quite small compared to thermal buoyancy points while the corresponding power spectra show an
force. From Fig. 7(d), it is clear that Nu shows oscillatory increase of two to three times more power than the previ-
heat transfer at the walls with the mean value around ous case. Fig. 10(a)–(c) shows that in the core of the cavity
5.0. Hence the heat transfer at both the walls is appreciably centrifugal and Coriolis forces are more dominant espe-
reduced due to 100-fold increase in X as compared to the cially there is large magnitude of centrifugal force from
previous case. centre of hot wall towards the cold wall which leads to for-
Further 10-fold increase of rotation corresponding to mation of mushroom-shaped isotherms. Fig. 10(d) shows
X = 14.285 (Ta = 8.16 · 106 and Raw = 1.02 · 108), results unsteady almost cyclic heat transfer at the walls. The min-
in drastic change in pattern formation of stream functions ima correspond to the time interval when thermal buoy-
rolls as is evident in Fig. 8(a)–(c). These figures show for- ancy is dominant while the maxima correspond to the
mation of counter-rotating rolls which occupy whole of duration when centrifugal force is the dominating force.

3
Level PSi Level Theta

5
13

11
15 13

2
12 1 4 15 15 -0.272903 15 0.9375
0.4 14 11
9
10 0.4
8 7 11 1 1
14 -0.631976 14 0.875
23 12
5 6 13 -0.991049 13 0.8125
3 4 8
12 -1.35012 11 12 0.75

6
9

1
0.2 2 11 -1.7092 0.2 11 0.6875

4
7
10
6

9
8
7

13
10 -2.06827 10 10 0.625
10

9
9 -2.42734 9 0.5625

10
4
1213

14
12
15

14
15

15
5

8 -2.78641 8 0.5
Y

0 0 8

3
5
11
11
14

7
9

3
5

7 -3.14549 7 0.4375

2
7
6 -3.50456 6 0.375
11
6
7

1
13
12

5 -3.86363 6 5 0.3125
4

-0.2 4 3 2 -0.2
12

4 -4.22271 4 0.25
8
9

6
7

5
11

8
9

4
13 3 -4.58178 3 0.1875

1
10

6 5
2 -4.94085 2 0.125
13

8 7
10 9
10

-0.4 14 145 1 -5.29993 -0.4 3 1 0.0625

12 11
14

15 1
12

13
15

-0.5 -0.25 0 0.25 0.5 -0.5 -0.25 0 0.25 0.5

X X
(a) t = 0.20 (d) t = 0.20
2

15 Level PSi Level Theta

14 13 15 -0.413494 15 0.9375
10

0.4 11 0.4
14

12
8
9 14 -0.857927 14 0.875
14
3

10 7 10 13 -1.30236 11 13 0.8125
13
12

6
6
8
15

5 4 12 -1.74679 12 0.75
12
11

15
15

10
13

0.2 11 -2.19123 0.2 11 0.6875

5

2
9
4
9

10 -2.63566 10 0.625
13
3
14

9 -3.0801 9 0.5625
1

1
3

7
8 7
11
9
6

11
Y

0 8 -3.52453 0 8 0.5
10

2
10

12
8

4
7

7 -3.96896 7 0.4375
8

8
13 14

6 -4.4134 7 6 0.375
8

14
10

5 -4.85783 6 6 5 0.3125
3

3 2
-0.2 4 -0.2
12

4 -5.30226 4 0.25
15
9

6
11

5
14 5

5
3 -5.7467 3 0.1875
1

12

15

4
7
7

8 9 2 -6.19113 2 0.125
13 11 10 13
-0.4 1 -6.63557 -0.4 1 0.0625
5

12
1

14
15
11

2
10

-0.5 -0.25 0 0.25 0.5 -0.5 -0.25 0 0.25 0.5

X X
(b) t = 0.60 (e) t = 0.60

15
Level PSi Level Theta
10

14 15 -0.413862 15 0.9375
0.4 13 12 0.4 12
10 14 -0.866205 14 0.875
8
9

9
15

11 14
14

7 13 -1.31855 11 13 0.8125
11

13

8 12 -1.77089 10 12 0.75
13

5 4
1
3
8

0.2 3 11 -2.22323 0.2 11 0.6875

6

9
12

10 -2.67557 10 0.625
15
12

1
4
5
7
15

9
14

10

9 -3.12792 9 0.5625
2
2

6
12
13
10

7
5

8 -3.58026 8 0.5
Y

0 0
8
10
9
11
2 2

7 -4.0326 8
7 0.4375
7

4
11

1 6 -4.48494 6 0.375
11
13
15

7
14
14

4
5 -4.93729 5 0.3125
13

-0.2 -0.2
10

3
14

9 4 -5.38963 4 0.25
8

5 6
13

8
7

6 3 -5.84197 3 0.1875
3

7 9 5
12 12 2 -6.29431 4 2 0.125
15

10
11
15

-0.4 13 1 -6.74666 -0.4 1 0.0625

2

15 14
12

6
8
0
9

-0.5 -0.25 0 0.25 0.5 -0.5 -0.25 0 0.25 0.5

X X
(c) t = 0.80 (f) t = 0.80

Fig. 11. (a), (b) and (c) show streamline contours while (d), (e) and (f) show isotherm contours for different time instants for Ra = 105, Ta = 8.16 · 104 and
Raw = 1.02 · 104.
3510 M.F. Baig, M. Zunaid / International Journal of Heat and Mass Transfer 49 (2006) 3500–3513

8 1
2 4
7
6 3 Level COR Level CEF
5 0 12 5 2 8
11 9 4 15 66.6562 7 15 67.6406
3 91
0.4 1 3 10 7 0.4

1
15 14 62.2125 14 63.1313
11 3 14

12
1 14 6

6
15
13 57.7687 13 58.6219

1
6 5

12
8
14 13
12 53.325 12 54.1125

1
4

7
8
11 10 12
0.2 11 48.8812 0.2 11 49.6031

9
6
15

14

5
3
7 10 44.4375 10 45.0938

10
11

8
7 24

12
4

12

2
98
9 39.9937 9 40.5844

52 9

3
1

15
13

11 13
5
2

10
8 35.55 8 36.075

3
11
59

6
4
3

2
4
Y

Y
0 0

2
10
7 31.1062 7 31.5656

15
10

7
31 10

14
6 26.6625 6 27.0563

4
11 9

13

6
5 22.2187 5 22.5469

12
8 7

1
12

4
8
10 4 17.775 4 18.0375

15

7
9

2
-0.2 -0.2

1
14 11

4 2

12
15 13 3 13.3312 3 13.5281
5

14

9
36
12
14

5
2 8.8875 2 9.01875

1011
8
63 1 4.44375 1 4.50938

2
8 119
13

2 5
13 0
-0.4 -0.4

3
4
1 10 1
7 11 15 14 7
9 12

13
12
4 25 8
36

15
1

-0.5 -0.25 0 0.25 0.5 -0.5 -0.25 0 0.25 0.5

X X
(a) (b)

10

1
Level TBF

4
0.4 15 937.5
5 Hot
3
14 875
13

13 812.5
Cold ................
11 8
12 750
15

12

0.2 10 11 687.5
14

10 625
6

9
10

9 562.5 6
8
9

8 500
1

Nu
8
Y

0 7 437.5
4
11

6 375
13

4
7 5 312.5
4 250
8

-0.2
15

6 3 187.5
12

6
9

2 125
14

2
10

5
1 62.5
-0.4 4
2

0
-0.5 -0.25 0 0.25 0.5 0.000198 0.500198 1.0002 1.5002

X t
(c) (d)

Fig. 12. Depicts spatial distribution of (a) Coriolis force, (b) centrifugal force, (c) thermal buoyancy force at t = 0.4 and (d) variation of Nusselt number
Nu with time for Ra = 105, Ta = 8.16 · 104 and Raw = 1.02 · 104.

1.05

1
-0.2

-0.4 0.95
θ
U

-0.6
0.9

-0.8

-1 0.85

9.9E-05 0.500099 1.0001 9.9E-05 0.500099 1.0001

t t
(a) (b)

6
0.001

4
Power ( U)

Power (θ)

0.0005

0 0
0.01 0.02 0.03 0.04 0.05 0.01 0.02 0.03 0.04 0.05

f / fs f / fs
(c) (d)

Fig. 13. (a) and (b) show time history while (c) and (d) show power spectrum of horizontal velocity component (U) and temperature (h) at spatial point (a)
for Ra = 106, Ta = 8.16 · 104 and Raw = 1.02 · 105.
 M.F. Baig, M. Zunaid / International Journal of Heat and Mass Transfer 49 (2006) 3500–3513 3511

21 7 3 8
410 9
Level COR Level CEF

13
6 12 115 9

5
15 6

14
5 9 12
0.4 13 15 15 237.281 0.4
7 15 676.406

2 4

3
14 1413 7 8
11 6

10
12 14 221.463 14 631.313

12
12 5
10 11 13

10 4 3
8
13 205.644 13 586.219

1
7 1
89

43 10
5

1
12 189.825 12 541.125

13

6
2

14
3 4 6 4

8
4

4
11
0.2 11 174.006 0.2 3 11 496.031

1
2 17
2

5
10 158.188 10 450.938

2
8 9

9
11
12

10

7
1211

5
9 142.369 9 405.844

5
6
11
12
6
1

3
4

3
8 126.55 8 360.75

14
10
14 13
95

2
12

7
1
Y

Y
0 7 110.731 0 7 315.656

13

10
13

8
2

1 7
7
6 94.9125 6 270.563

12
89
7
5 79.0938 5 225.469

5
3 4

43

3
2
2 1 4 63.275 4 180.375

8 10
6

6 4
119 6

6
-0.2 -0.2

3
1110

1
5 4

4
13
3 47.4563 3 135.281

14
2

11
12
2 31.6375 2 90.1875
1

5
3
5

9 8 7 10 2

2
6 9 11
10 11 1 1 15.8188 1 45.0938
8

7
12 1 14
4

3
5
3 13
-0.4 -0.4
2

43 13

2
7 14 12 7
1 11 15
2 96

12
5 810 43 1

8
-0.5 -0.25 0 0.25 0.5
-0.5 -0.25 0 0.25 0.5
X X
(a) (b)

Level TBF 10
13

12
11 15 9375
0.4 Hot
2
10 14 8750
15

9 1
13 8125 8 Cold .............
11

12 7500
6
34

0.2 10 11 6875
8
7

10 6250
14

5
9

9 5625 6
8

8 5000
10

Nu
13
Y

0
12
8

7 4375
6 3750
2
7

4
15

7 5 3125
9

4 2500
11

-0.2
1
6

8
3 1875
3
4

7
2 1250 2
14

1 625
6

6
-0.4
10

8
13
12

0
-0.5 -0.25 0 0.25 0.5 9.9E-05 0.500099 1.0001 1.5001
X t
(c) (d)

Fig. 14. Shows spatial distribution of (a) Coriolis force (b) centrifugal force (c) thermal buoyancy force at t = 0.4 and (d) variation of Nusselt number Nu
with time for Ra = 106, Ta = 8.16 · 104 and Raw = 1.02 · 105.

3.2. Effects of increasing thermal buoyancy force dinate points. Fig. 14 shows the spatial variation of forces
which still exhibits the domination of thermal buoyancy
These simulations have been performed keeping the force with the centrifugal force getting stronger in the core
non-dimensional rotational speed X = 1.428 (Ta = 8.16 · of the cavity. From Fig. 14(d) it is clear that Nu shows
104 and Raw = 1.02 · 104) constant while the Ra is varied oscillatory heat transfer at the walls with a mean value of
from 105 to 107. For the case when Ra = 105, the stream 4.0 approximately.
functions exhibit a single clockwise roll almost occupying
the whole cavity as shown in Fig. 11. The corresponding 3.3. Heat transfer analysis
isotherms depict thick boundary layers near the isothermal
walls with stable stratification in the core of the enclosure. The plots (a) and (b) in Fig. 15 shows the effect on heat
The time history depicts steady flow at all the co-ordinate transfer at fixed Ra = 107 as rotational speed X is increased
points with the power density spectrums further corrobo- slightly from 0 to 0.014. From these plots it is apparent that
rating that view. The spatial variation of forces in there is no effect on magnitude of Nu. Fig. 15(c) shows that
Fig. 12(a)–(c) show thermal buoyancy as the main domi- at X = 1.428, the magnitude of Nu has appreciably
nating force. From Fig. 12(d) it is clear that Nu shows decreased to a mean value of around 7.0. Further increase
steady heat transfer at the walls with a value of 2.8. of X to 14.285 results in significant change in Nu variation
Further increase of Ra to 106 for same X as in the pre- with time. The possible source of almost cyclic rise and fall
vious case, results in stream function rolls becoming more of Nu is due to alternate domination of thermal buoyancy
stronger but the the overall pattern remains the same. The and centrifugal forces. As X is increased further to a large
isotherms show a comparably thinner thermal boundary value of 71.428, the Nu shows an appreciable rise to a mean
layer near the walls and also there is onset of convective value around 16.0.
motion in the core of the cavity. The time history of veloc- These change can be explained that at lower to medium
ity component U as well as temperature h in Fig. 13 depicts rotational speeds there is disruption of heat transfer by
a quasi-periodic time signal at all the spatial points. The opposing interaction of thermal and rotational buoyancy
power spectra show quasi-periodicity with a fundamental which results in decrease of Nu. For rotational speeds lar-
frequency and two to three sub-harmonics at all the coor- ger than a critical speed, the heat transfer is governed by
3512 M.F. Baig, M. Zunaid / International Journal of Heat and Mass Transfer 49 (2006) 3500–3513

20 20
Hot Hot
18 . Cold ............... Cold .............
16
15
14

Nu
Nu
12

10 10

6
5
4
0.1 0.2 0.3 0.4 0.1 0.2 0.3 0.4
t t
(a) (b)

20
Hot Hot
15
Cold ............
Cold ..................
15

10
Nu

Nu

10

5
5

9.9E-05 0.500099 1.0001 9.9E-05 0.500099 1.0001

t t
(c) (d)

30
Hot
Cold ..................
25

20
Nu

15

10

5
0.1 0.2 0.3 0.4 0.5 0.6 0.7
t
(e)

Fig. 15. Variation of Nu for different Ta and Raw at Ra = 107: (a) Ta = 0, Raw = 0; (b) Ta = 8.16, Raw = 1.02 · 102; (c) Ta = 8.16 · 104, Raw = 1.02 · 106;
(d) Ta = 8.16 · 106, Raw = 1.02 · 108; and (e) Ta = 2.04 · 108, Raw = 2.55 · 109.

centrifugal forces and occurs not through the boundary 1. Increase in the thermal buoyancy force with increasing Ra
layers but by the mushroom-shaped structure and hence generates thinner boundary layers which in turn increases
there is rise of Nu. heat transfer at the isothermal walls of the cavity.
2. Primary roll pattern changes into a multi-cellular roll
4. Conclusions pattern as Ra is increased keeping X constant.
3. At higher value of X i.e. at larger values of Ta and Raw,
The following conclusions can be derived on basis of the two horizontally aligned counter-rotating rolls are
numerical experiments: obtained. This leads to generation of mushroom-shaped
 M.F. Baig, M. Zunaid / International Journal of Heat and Mass Transfer 49 (2006) 3500–3513
structure isotherms in the core of the cavity with unsta-
ble thermal stratification.
4. This significantly higher strength of counter-rotating
rolls generated by the centrifugal force in the core of
the cavity lead to enhanced heat transfer.
5. At the rotational speed X = 14.285 there is alternate
domination of centrifugal and thermal buoyancy forces
and this results in almost periodic variation of Nusselt
number with time.
6. At very high rotational speeds corresponding to the case
X = 71.428, there is a significant increase in Nusselt
number which is primarily due to domination of centrif-
ugal force.
References
[1] W. Nakayama, Forced convective heat transfer in a straight pipe
rotating around a parallel axis (2nd report, Turbulent region), Int. J.
Heat Mass Transfer 11 (1968) 1185–1201.
[2] Y. Mori, W. Nakayama, Forced convection heat transfer in straight
pipe rotating around a parallel axis (1st report, Laminar region), Int.
J. Heat Mass Transfer 10 (1967) 1179–1194.
[3] B. Weigand, H. Beer, Fluid flow and heat transfer in a rotating pipe
subjected to external convection, Int. J. Heat Mass Transfer 35 (1992)
1803–1809.
[4] A.R. Johnson, W.D. Morris, An experimental investigation into the
effects of rotation on the isothermal flow resistance in circular tubes
rotating about a parallel axis, Int. J. Heat Mass Transfer 13 (1992)
132–140.
[5] C.T. Nguyen, H. Bazzi, J. Orfi, Numerical investigation of the effects
of rotation on a germanium float zone under micro-gravity condi-
tions, Numer. Heat transfer, Part A 28 (1995) 2667–2685.
3513
[6] M. Mahadevappa, K.V.C. Rao, V.M.K. Sastri, Numerical study of
fully developed fluid flow and heat transfer in rectangular and
elliptical ducts rotating about a parallel axis, Int. J. Heat Mass
Transfer 39 (1996) 867–875.
[7] K.T. Lee, H.L. Tsai, W.M. Yan, Mixed convection heat and mass
transfer in vertical rectangular ducts, Int. J. Heat Mass Transfer 40
(1997) 1621–1631.
[8] G.J. Hwang, T.C. Jen, Convective heat transfer in rotating isothermal
ducts, Int. J. Heat Mass Transfer 33 (1990) 1817–1828.
[9] S. Fann, W.J. Yang, Hydrodynamically and thermally developing
laminar flow through rotating channels having isothermal walls,
Numer. Heat Mass transfer, Part A 22 (1992) 257–288.
[10] Y. Mori, T. Fudaka, W. Nakayama, Convective heat transfer in
rotating circular pipe (2nd report), Int. J. Heat Mass Transfer 14
(1971) 1807–1824.
[11] W.-M. Yan, Mixed convection heat and mass transfer in rectangular
ducts rotating about parallel axis, Int. J. Heat Mass Transfer 42
(1999) 2955–2965.
[12] F. Song, D. Ewing, C.Y. Ching, Fluid flow and heat transfer model
for high speed rotating heat pipes, Int. J. Heat Mass Transfer 46
(2003) 4393–4401.
[13] T. Yang, L. Wang, Bifurcation and stability of combined free and
convection in rotating curved ducts of square cross-section, Int. J.
Heat Mass Transfer 46 (2003) 613–629.
[14] Y.T. Ker, T.F. Lin, A combined numerical and experimental study of
air convection in a differentially heated cubic cavity, Int. J. Heat Mass
Transfer 39 (1996) 3193–3210.
[15] N. Hasan, M.F. Baig, Evolution to aperiodic penetrative convection
in odd shaped rectangular enclosures, Int. J. Numer. Methods Heat
Fluid Flow 12 (8) (2002) 895–915.
[16] C. Nonino, G. Croce, An equal-order velocity–pressure algorithm for
incompressible thermal flows, Numer. Heat Transfer, Part B 32
(1997) 17–35.
[17] J.H. Ferziger, M. Peric, Computational Methods for Fluid Dynamics,
third ed., Springer-Verlag, 2003, pp. 101–105.