Technical Note

Natural convection around a radial heat sink

Seung-Hwan Yu, Kwan-Soo Lee
*
, Se-Jin Yook
School of Mechanical Engineering, Hanyang University, 17 Haengdang-dong, Seongdong-gu, Seoul 133-791, Republic of Korea
a r t i c l e i n f o
Article history:
Received 15 October 2009
Received in revised form 11 February 2010
Accepted 11 February 2010
Available online 19 March 2010
Keywords:
Natural convection
Heat sink
Circular base
Correlation
a b s t r a c t
This paper presents the details of an experimental and numerical investigation of natural convection in a
radial heat sink, composed of a horizontal circular base and rectangular ns. The general ow pattern is
that of a chimney; i.e., cooler air entering from outside is heated as it passes between the ns, and then
rises from the inner region of the heat sink. Parametric studies are performed to compare the effects of
three geometric parameters (n length, n height, and number of ns) and a single operating parameter
(heat ux) on the thermal resistance and the average heat transfer coefcient for the heat sink array. In
addition, a correlation is proposed to predict the average Nusselt number for a radial heat sink.
2010 Elsevier Ltd. All rights reserved.
1. Introduction
Light-emitting diode (LED) lights have recently attracted the
attention of the illumination industry, due to their lower power
consumption, longer life, and smaller, more durable structure com-
pared to other light sources. However, their use presents a thermal
problem, since about 70% of their total energy consumption is
emitted as heat. An efcient heat sink design is essential to solve
this problem. Natural convection heat sinks are appropriate for
LED lights, considering their overall advantages. However, natural
convection heat sinks commonly have rectangular bases, whereas
LED lights are generally circular. It is therefore desirable to inves-
tigate natural convection heat transfer via a heat sink with a circu-
lar base.
Numerous experimental [14] and numerical [5] studies of
rectangular n or pin n heat sinks have been carried out. Starner
and McManus [1] experimentally investigated natural convection
heat transfer from four heat sinks of differing dimensions, with
the heat sinks oriented vertically, at a 45 angle, and horizontally.
Welling and Woolbridge [2] conducted an experimental study of
vertically oriented rectangular ns of constant length attached to
a vertical base. They found that there exists an optimal n height,
corresponding to a maximum rate of natural convection heat
transfer, for any given n spacing. Harahap and Mcmanus [3] per-
formed experiments to calculate the average heat transfer coef-
cients for two different n lengths, and established a correlation
with non-dimensional parameters and relevant n dimensions.
However, most of these studies were concerned with heat sinks
with rectangular bases, which might be inefcient for cooling cir-
cular LED lights.
In this study, natural convection from a heat sink with a circular
base and rectangular ns is numerically and experimentally ana-
lyzed, and the thermo-ow pattern is observed. The effects of the
number of ns, n length, n height, and heat ux on the thermal
resistance and the average heat transfer coefcient are investi-
gated. A correlation is proposed to predict the average heat transfer
coefcient for this type of heat sink, as a function of heat sink
dimensions and heat ux.
2. Mathematical modeling
Fig. 1 shows a radial heat sink consisting of a circular base and
rectangular ns. The ns were arranged radially at regular inter-
vals. The heat sink base was oriented horizontally. The heat sink
was made of aluminum, whose properties are listed in Table 1.
2.1. Governing equations
For the numerical analysis, the following assumptions were
imposed.
(1) The ow was steady, laminar, and three-dimensional.
(2) Aside from density, the properties of the uid were indepen-
dent of temperature.
(3) Air density was calculated by treating air as an ideal gas.
(4) Radiation heat transfer was negligible.
The governing equations were as follows.
0017-9310/$ - see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijheatmasstransfer.2010.02.032
* Corresponding author. Tel.: +82 2 2220 0426; fax: +82 2 2295 9021.
E-mail address: ksleehy@hanyang.ac.kr (K.-S. Lee).
International Journal of Heat and Mass Transfer 53 (2010) 29352938
Contents lists available at ScienceDirect
International Journal of Heat and Mass Transfer
j our nal homepage: www. el sevi er . com/ l ocat e/ i j hmt
2.1.1. Air side
Continuity equation:
@qu
@x

@qv
@y

@qw
@z
0 1
Momentum equations:
@qu
2

@x

@quv
@y

@quw
@z

@P
@x
l
@
2
u
@x
2

@
2
u
@y
2

@
2
u
@z
2
!
2
@qvu
@x

@qv
2

@y

@qvw
@z

@P
@y
l
@
2
v
@x
2

@
2
v
@y
2

@
2
v
@z
2
!
gq q
a
3
@qwu
@x

@qwv
@y

@qw
2

@z

@P
@z
l
@
2
w
@x
2

@
2
w
@y
2

@
2
w
@z
2
!
4
Energy equation:
@quT
@x

@qvT
@y

@qwT
@z

k
c
p
@
2
T
@x
2

@
2
T
@y
2

@
2
T
@z
2
!
5
2.1.2. Solid side
Energy equation:
@
2
T
@x
2

@
2
T
@y
2

@
2
T
@z
2
0 6
The density of air was calculated from the ideal gas law,
q
P
atm
R
c
=M
w
T
7
where M
w
of air is 28.966 kg/kmol.
Periodic boundary conditions were adopted in accordance with
the geometry of the heat sink (Fig. 1). Because of the number of
grids and the computational time involved, only a single n was
considered, as shown in Fig. 2.
2.2. Numerical procedure and validation
The numerical simulation was conducted using Fluent V6.3, a
commercially available CFD code based on the nite volume meth-
od. The grid dependence was investigated by varying the number
of grid points from 22,680 to 285,714. We selected 65,016 grid
points; additional grid points produced a change of less than
0.5% in the average heat sink temperature for the reference model
of n = 20 and r
o
= 75 mm.
The numerical results were validated with experimental data by
comparing the differences between the ambient and heat sink tem-
peratures. The geometric parameters of the experimental model
were n = 20, r
o
= 75 mm, L = 55 mm, H = 21.3 mm, and t = 2 mm.
Fig. 3 compares the temperature differences between the experi-
mental and numerical results in terms of the heat ux applied to
the heat sink base. This implies that the present numerical model
can correctly predict the natural convection ow around a radial
heat sink.
Fig. 1. Radial heat sink with a circular base and rectangular ns.
Nomenclature
b spacing between ns, mm
c
p
coefcient of heat capacity, J/(kg C)
F view factor
h heat transfer coefcient, W/m
2
K
H n height, mm
k thermal conductivity, W/m C
L n length, mm
M
w
gas molecular weight, kg/kmol
Nu Nusselt number, hL/k
n number of ns in the normal direction
Pr Prandtl number
p pressure, N/m
2
_ q heat ux, W/m
2
R
c
universal gas constant
R
TH
thermal resistance, C/W
Ra* modied Rayleigh number,
q
2
gbcppr
2
o
r
2
i
_ qL
3
lLk
2
r radius, mm
T temperature, K or C
t n thickness, mm
u x-component of velocity, m/s
v y-component of velocity, m/s
w z-component of velocity, m/s
Greek symbols
e emissivity
l dynamic viscosity, N/m
2
s
h angle,
q density, kg/m
3
r StefanBoltzmann constant, 5.67 10
8
W/m
2
K
4
Subscripts
avg average
f uid (air)
i inner
o outer
s solid (heat sink)
Table 1
Air and heat sink properties.
Material c
p
(J/
kg C)
l (N/m
2
s) k (W/m C) q (kg/
m
3
)
Air 1005.585 1.834 10
5
2.643 10
5
Eq. (7)
Heat sink
(aluminum)
2800 193 880
2936 S.-H. Yu et al. / International Journal of Heat and Mass Transfer 53 (2010) 29352938
3. Results and discussion
Parametric studies were carried out by numerically investigat-
ing the effects of the number of ns, n length, n height, and heat
ux on the thermal resistance and the heat transfer coefcient.
Based on these results, a correlation was proposed to predict the
Nusselt number for a heat sink with a horizontal circular base
and rectangular ns.
3.1. Thermo-ow characteristics
There are two ows, i.e., vertical and horizontal ows, around
the radial heat sink. The vertical ow is in the upward direction,
since air is heated by the heat sink (which is maintained at a higher
temperature) and becomes lighter than the surrounding air. The
horizontal ow is created by air entering from outside the heat
sink to make up for the vertical ow in the inner region. Therefore,
the overall ow pattern is chimney-like. The temperature of heat
sink maintains almost uniformly high because of high conductivity
of aluminum. The heat transfer rate in the outer region of the heat
sink was higher than in the inner region. This was because the tem-
perature difference between the air and the heat sink decreased as
the cool air proceeded towards the inner region of the heat sink.
3.2. Parametric study
The effects of the number of ns, n length, n height, and heat
ux on the thermal resistance and the heat transfer coefcient
were investigated. The reference model is n = 20, r
o
= 75 mm,
L = 55 mm, H = 21.3 mm, t = 2 mm, and _ q 700 W=m
2
.
The effect of the number of ns on the thermal resistance and
heat transfer coefcient is shown in Fig. 4(a). The average heat
transfer coefcient decreased as the number of ns increased, since
the ow rate of the cooler air entering the spaces between the ns
decreased and the air was heated more quickly on account of the
reduced space between ns. However, when the number of ns
was less than 36, the thermal resistance of the heat sink decreased
with increasing n, since the effect of the increased heat transfer
surface area was larger than the effect of the decreased heat trans-
fer coefcient. When the number of ns was greater than 36, the
thermal resistance of the heat sink increased with increasing n,
since the heat transfer coefcient was very small. Consequently,
there exists optimum number of ns that gives the minimum ther-
mal resistance.
Fig. 4(b) shows the effect of the n length. As the n length in-
creased, the thermal resistance and average heat transfer coef-
cient decreased. The thermal resistance leveled off and reached a
steady value when the n was longer than 55 mm. This was be-
cause the air temperature in the inner region was almost the same
as the heat sink temperature, and hence any additional n length
beyond 55 mm did not contribute to the heat transfer rate.
Fig. 4(c) indicates the effect of the n height. A lower thermal
resistance resulted from the increased heat transfer surface area
created by the incremented n height. However, the change in
the heat transfer coefcient was relatively small, since the velocity
of the air entering from outside increased very little with increas-
ing n height.
Fig. 4(d) illustrates the effect of the heat ux applied to the heat
sink base. The decrease in thermal resistance due to increasing
heat ux resulted in a greater rising air velocity, which in turn in-
creased the ow rate of the cooler air entering from outside.
Accordingly, the average heat transfer coefcient increased almost
linearly, thanks to the enhanced effect of natural convection.
3.3. Correlation
A correlation for predicting the Nusselt number for a heat sink
with a horizontal circular base and rectangular ns was derived
as a function of the parameters investigated in the previous sec-
tion, as well as other geometric parameters, and was obtained from
numerical data. This formula is based on the correlations for rect-
angular heat sinks obtained in previous studies [1,3], using average
n spacing and the modied channel Rayleigh number,
Nu 0:195Ra

0:263
nb
avg
H

1:35
r
o
L

0:444
r
o
b
avg

0:142
r
o
H

1:4
8
where Ra

q
2
gbcppr
2
o
r
2
i
_ qL
3
lLk
2
; b
avg
= {(2pr
o
/n t) + (2p(r
o
L)/n t)}/2,
and the properties are based on the lm temperature.
The predicted correlation was consistent with the numerical da-
ta, with an error of less than 10%, when t = 2 mm, 20 6 n 6 36,
0 200 400 600 800
10
20
30
40
Computational results
Experimental results
T
a
v
g

q(W/m
2
)
Fig. 3. Comparison of the temperature differences between the experimental and
numerical results.
Fig. 2. Computational domain and dimensions.
S.-H. Yu et al. / International Journal of Heat and Mass Transfer 53 (2010) 29352938 2937
21.3 mm6 H 6 63.9 mm, 75 mm6 r
o
6 102 mm, 40 mm6 L 6
80 mm, and 300 W=m
2
_ q 1100 W=m
2
.
4. Conclusions
Natural convection from a radial heat sink was experimentally
and numerically investigated. The general ow pattern was like
that of a chimney; i.e., the cooling air entering from outside was
heated as it passed between the ns, and then rose from the inner
region of heat sink. Parametric studies were performed to compare
the effects of the number of ns, n length, n height, and heat ux
on the thermal resistance and the heat transfer coefcient. As the
number of ns, n length, and n height increased, the thermal
resistance and heat transfer coefcient generally decreased. How-
ever, there existed optimal values of the number of ns and n
length to obtain an effective low heat sink temperature. The ther-
mal resistance decreased and the heat transfer coefcient in-
creased in proportion to the heat ux applied to the heat sink
base. A correlation was proposed to predict the average Nusselt
number for a radial heat sink.
References
[1] K.E. Starner, H.N. McManus, An experimental investigation of free convection
heat transfer from rectangular n arrays, J. Heat Transfer 85 (2) (1963) 273
278.
[2] J.R. Welling, C.B. Wooldridge, Free convection heat transfer coefcients
from rectangular vertical ns, Trans. ASME J. Heat Transfer 87 (3) (1965)
439444.
[3] F. Harahap, H.N. McManus, Natural convection heat transfer from horizontal
rectangular n arrays, J. Heat Transfer 89 (1) (1967) 3238.
[4] R.T. Huang, W.J. Sheu, C.C. Wang, Orientation effect on natural convective
performance of square pin n heat sinks, Int. J. Heat Mass Transfer 51 (910)
(2008) 23682376.
[5] S. Baskaya, M. Sivrioglu, M. Ozek, Parametric study of natural convection heat
transfer from horizontal rectangular n arrays, Int. J. Therm. Sci. 39 (8) (2000)
797805.
24 32 40
2.25
2.50
2.75
R
T
H
(
o
C
/
W
)
n
2
4
6
h
a
v
g

(
W
/
m
2
K
)
R
TH
h
avg
40 50 60
2.50
2.75
3.00
R
T
H
(
o
C
/
W
)
L (mm)
4
5
6
7
h
a
v
g
(
W
/
m
2
K
)
R
TH
h
avg
(a) The effect of the number of fins (b) The effect of the fin length
20 30 40
1.5
2.0
2.5
3.0
R
T
H
(
o
C
/
W
)
H (mm)
4
5
6
h

a
v
g
(
W
/
m
2
K
)
R
TH
h
avg
200 400 600 800 1000 1200
2.0
2.5
3.0
3.5
R
T
H
(
o
C
/
W
)
q (W/m
2
)
4
5
6
7
h
a
v
g
(
W
/
m
2
K
)
R
TH
h
avg
(c) The effect of the fin height (d) The effect of the heat flux
Fig. 4. The results of parametric study. (a) The effect of the number of ns, (b) the effect of the n length, (c) the effect of the n height, (d) the effect of the heat ux.
2938 S.-H. Yu et al. / International Journal of Heat and Mass Transfer 53 (2010) 29352938