Available online at www.sciencedirect.

com

International Communications in Heat and Mass Transfer 35 (2008) 657 – 665

www.elsevier.com/locate/ichmt

Natural convection heat transfer enhancement in horizontal

concentric annuli using nanofluids ☆
E. Abu-Nada ⁎, Z. Masoud, A. Hijazi
Department of Mechanical Engineering, Hashemite University, Zarqa 13115, Jordan
Available online 6 February 2008

Abstract

Heat transfer enhancement in horizontal annuli using nanofluids is investigated. Water-based nanofluid containing various volume fractions of
Cu, Ag, Al2O3 and TiO2 nanoparticles is used. The addition of the different types and different volume fractions of nanoparticles were found to
have adverse effects on heat transfer characteristics. For high values of Rayleigh number and high L/D ratio, nanoparticles with high thermal
conductivity cause significant enhancement of heat transfer characteristics. On the other hand, for intermediate values of Rayleigh number,
nanoparticles with low thermal conductivity cause a reduction in heat transfer. For Ra = 103 and Ra = 105 the addition of Al2O3 nanoparticles
improves heat transfer. However, for Ra = 104, the addition of nanoparticles has a very minor effect on heat transfer characteristics.
© 2008 Elsevier Ltd. All rights reserved.

Keywords: Nanofluids; Heat transfer enhancement; Natural convection; Concentric annulus

1. Introduction different from conventional particles (millimeter or micro-scale)

in that they tend to remain suspended in the fluid and no
Heat transfer within horizontal annuli has many engineering sedimentation occur which causes no increase in pressure drop
applications such as heat exchangers, solar collectors, thermal in the flow field [2].
storage systems, and cooling of electronic components. Several The past decade has witnessed extensive work on
applications use natural convection as the main heat transfer convective heat transfer using nanofluids. Studies on the
mechanism. Therefore, it is important to understand the thermal enhancement of heat transfer characteristics in forced convec-
behavior of such systems when only natural convection is in tion applications were conducted by a number of researchers.
effect so that methods to enhance heat transfer characteristics in On the other hand, heat transfer enhancement in natural
such systems can be devised. convection applications has received little attention. Examples
The geometric shape of the cylindrical annulus creates non- of the work conducted on natural convection heat transfer
uniformity in heat transfer within the annulus. With a better include the work of Khanafer et al. [3] who studied Copper–
understanding of the flow field, it is possible to devise methods Water nanofluids in a two dimensional rectangular enclosure.
for heat transfer enhancement. An innovative technique for They found that the heat transfer rate increased by increasing
improving heat transfer is using ultra fine solid particles in the the percentage of the suspended particles. Similar enhance-
base fluids, which has been used extensively in the past ten ment was acheived experimentally by Nnanna et al. [4] for Cu
years. The term nanofluid refers to fluids in which nano-scale nanoparticles in ethylene glycol and by Nnanna and Routhu
particles are suspended in the base fluid [1]. The particles are [5] for Alumina–Water nanofluids. However, different experi-
mental findings were reported by Putra et al. [6] on the natural
convection of Al2O3– and CuO–Water nanofluids inside a
cylindrical enclosure heated from one side and cooled from the
☆
Communicated by W.J. Minkowycz. other. They found that the natural convection heat transfer
⁎ Corresponding author. coefficient was lower than that of pure water. Wen and Ding
E-mail address: [email protected] (E. Abu-Nada). [7] found that the natural convection heat transfer coefficient
0735-1933/$ - see front matter © 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.icheatmasstransfer.2007.11.004
658 E. Abu-Nada et al. / International Communications in Heat and Mass Transfer 35 (2008) 657–665

Nomenclature Recently, Trisaksri and Wongwises [9], and Wang and

Mujumdar [10] conducted a literature review on the general heat
b source term transfer characteristics of nanofluids. Daungthongsuk and
cp specific heat at constant pressure Wongwises [2] performed a comprehensive review of con-
D diameter of inner cylinder vective heat transfer of nanofluids. Based on literature reviews,
g gravitational acceleration no work has focused on natural convection heat transfer
h local heat transfer coefficient enhancement within a concentric cylindrical annulus. There-
k thermal conductivity fore, the goal of this work is to investigate heat transfer
L gap between inner and outer cylinder, i.e., characteristics of natural convection in the annulus between
L = ro − ri horizontal concentric cylinders using different types of nano-
Nu Nusselt number fluids. The problem will be investigated numerically by solving
n particle shape factor the Navier–Stokes and energy equations (NSE) using the finite
Pr Prandtl number, volume technique. Heat transfer characteristics will be analyzed
qw heat transfer at the cylinder wall using a wide range of volume fractions of nanoparticles at
Ra Rayleigh number, various Rayleigh numbers.
~
r radial coordinate measured from the inner
cylinder surface 2. Problem description and governing equations
r nondimensional radial distance,
~ Due to geometrical symmetry, Fig. 1, only one half of the
T dimensional temperature
T nondimensional temperature, annulus is simulated. The annulus between the two cylinders is
u dimensional tangential velocity filled with water based nanofluid. Four types of nanoparticles
U nondimensional tangential velocity, (Al2O3, Cu, Ag, and TiO2) are investigated. It was assumed that
v dimensional radial velocity the flow is steady, and that the base fluid and the nanoparticles
V nondimensional radial velocity, are in thermal equilibrium and no slip occurs between them.
Thermo-physical properties of the nanofluid are assumed to be
Greek symbols constant except for the density variation, which is approximated
α thermal diffusivity by the Boussinesq model. The thermo-physical properties of the
β thermal expansion coefficient base fluid and the different types of nanoparticles as well as air
ε numerical tolerance which were used for code validation are given in Table 1. The
η coordinates in computational plane inner cylinder surface is maintained at a constant temperature Ti
φ nanoparticle volume fraction which is higher than the outer cylinder temperature To. The
ϕ transport quantity governing equations for the laminar natural convection in terms
ν kinematic viscosity of stream function-vorticity formulation are the vorticity
θ angle measured from the lower plane equation, the energy equation, and the kinematics equation
Ψ nondimensional stream function, respectively:
ψ dimensional stream function    
Ω nondimensional vorticity, A AW A AW
x x
ω dimensional vorticity A r̃ Ah
  Ah  A r̃  
ρ density A A Ax A 1 Ax
¼ nf r̃ þ
μ dynamic viscosity qnf A r̃ A r̃ Ah r̃ Ah !
ξ coordinate in computational plane


uqs bs þ ð1 uÞqf bf AT̃ AT̃
þ g r̃sinh þ cosh ¼ 0
qnf A r̃ Ah
Subscripts
f fluid ð1Þ
i inner
nf nanofluid     !
A AW A AW A AT̃
o outer T̃ T̃ ¼ anf r̃
s solid A r̃ Ah Ah A r̃ A r̃ A r̃
!
A 1 AT̃
þ anf ð2Þ
Ah r̃ Ah
in a vessel composed of two discs using TiO2 nanoparticles
decreases by increasing the volume fraction of nanoparticles.
   
A AW A 1 AW
Jou and Tzeng [8] simulated natural convection heat transfer of r̃ þ ¼ r̃x ð3Þ
A r̃ A r̃ Ah r̃ Ah
Copper–Water nanofluids in a two dimensional enclosure.
They reported an increase in heat transfer by the addition of where anf ¼ knf
.
nanoparticles. ðqcp Þnf
 E. Abu-Nada et al. / International Communications in Heat and Mass Transfer 35 (2008) 657–665 659

for studying heat transfer enhancement using nanofluids

[13–16].
The radial and tangential velocities are given by the
following relations, respectively:
1 Aw Aw
v¼ and u ¼ ð8Þ
r̃ Ah A r̃
The following dimensionless group is introduced
xL2 w vL uL T̃ To
X¼ ;W ¼ ;V ¼ ;U ¼ ;T ¼ ; and
af af af af Ti To
r̃ ri
r¼
L
Also, the following transformation is used:
 
  n r
bþ1 1 r
ð2a bÞ þ ð2a þ bÞ
b 1
r¼  1
 n r
0
 ð9Þ
Fig. 1. Sketch of problem geometry. b þ 1 1 r
ð2a þ 1Þ@1 þ A
b 1
h ¼ pg
The viscosity of the nanofluid is approximated by the
viscosity of a base fluid μf containing diluted suspension of This transformation ensures a fine grid (a stretched grid)
fine spherical particles and is given by [11]: close to the inner and to the outer cylinders surfaces, which can
resolve the steep velocity and temperature gradients at these
Af surfaces. This grid stretching method results in considerable
Anf ¼ ð4Þ
uÞ2:5
:
ð1 savings in terms of the grid size and the computational time
[17]. By using the mentioned transformation the physical
The density of the nanofluid is given by: domain (r, θ) is transformed into a uniformm grid in the com-
qnf ¼ ð1 uÞqf þ uqs ð5Þ putational (ξ, η) domain (see Fig. 2).
By using the dimensionless parameters and the transforma-
The heat capacitance of the nanofluid is expressed as: tion given in Eq. (9), the governing equations are rewritten as:
    
1 A AW A AW




qcp nf ¼ ð1 uÞ qcp f þu qcp s ð6Þ X X ¼ sGðXÞ
Prp An Ag Ag An
The effective thermal conductivity of the nanofluid is ap- "
Ar
#
proximated by the Hamilton–Crosser model [12] as: An AT  ri  AT ð10Þ
þdRa cosðpgÞ þ r þ sinðpgÞ

p Ag L An
knf ks þ ðn 1Þkf ðn 1Þ kf ks u
¼
ð7Þ
kf ks þ ðn 1Þkf þ kf ks u  
1 A AW

A

AW

T T ¼ GðT Þ ð11Þ
where n is the nanoparticle shape factor which is set equal to 3 kp An Ag Ag An
for spherical nanoparticles. The use of this equation is restricted
1
to spherical nanoparticles where it does not account for other X¼
GðwÞ ð12Þ
shapes of nanoparticles. This model is found to be appropriate
Ar
An r þ rLi

Table 1
Thermo-physical properties
Pr ρ (kg/m3) cp (J/kg K) k (W/m K) β × 10− 5 (K− 1) α × 10− 5
(m2/s)
Base fluids Air 0.7 1.2 1006 0.026 340
Water 6.2 997.1 4179 0.613 21
Nano-particles Copper (Cu) 8933 385 401 1.67 11.7
Silver (Ag) 10500 235 429 1.89 17.4
Alumina (Al2O3) 3970 765 40 0.85 1.3
Titanium Oxide (TiO2) 4250 686.2 8.9538 0.9 0.31
660 E. Abu-Nada et al. / International Communications in Heat and Mass Transfer 35 (2008) 657–665

where After solving for Ψ, Ω, and T, further useful quantities are

obtained. For example, Nusselt number can be expressed as:
r i  A2 Q A2 n Ar
 
1
GðQÞ ¼ Ar r þ
An
L An2
þ 1þr 2
Ar An
þr ð
A An
An Ar Nu ¼ Ar
 
2 ri  knf AT
: ð17Þ
Ar An
L kf An
Ar A2 n Ar A An ri  AQ A2 Q
  
þ þ
An Ar2 An An Ar L
An
An p2 r þ rLi Ag2

Þ The average Nusselt number on the inner cylinder surface is
evaluated as:
1 1 bs 1
s¼   ; d ¼ ð1 uÞ qf
þ u qs ; Z g¼1
ð1 uÞ 0:25
ð1 uÞ þ u qqs u qs þ 1 bf ð1 uÞ q þ 1 f Nuavg ¼ NuðgÞdg: ð18Þ
f
g¼0
knf
mf gbðTw Tl ÞL3 kf
Pr ¼ ; Ra ¼ ; and k ¼ 1/3rd Simpson's rule of integration is used to evaluate Eq. (18).
ðqcp Þs
:
af v f af
ð1 uÞ þ u The Nusselt number is used as an indicator of heat transfer
ðqcp Þf
enhancement where an increase in Nusselt number corresponds
The velocities are written in a non-dimensional form in the to enhancement in heat transfer.
computational domain as:

1 AW 4. Grid testing and code validation

V ¼ ð13Þ
p r þ rLi Ag


An extensive mesh testing procedure was conducted to
1 AW guarantee a grid independent solution. Two cases of Ra =
U¼ ð14Þ
Ar An 0.53 × 104 and Ra = 105 using Pr = 0.7 are tested for grid inde-
An
pendence with no nanoparticles in the flow field. The present
The dimensionless boundary conditions are as follows: code was tested for grid independence by calculating the
average Nusselt number around the inner cylinder surface. It
On the inner cylinder surface;W ¼ 0; was found that a grid size of 61 × 61 guarantees a grid inde-

g
pendent solution for both cases. The Nusselt number for the grid
1 A2 W
x¼ 2 ; and T ¼ 1: independent solution is compared with the results of Guj and
An2

Ar
Stella [21] and Shu et al. [22] for concentric horizontal annulus
An
and Ra = 0.53 × 104. The calculated average Nusselt number by
On the outer cylinder surface;W ¼ 0; ð15Þ the current code gives a value of 2.47010 which falls between
1A2 W the results obtained by Guj and Stella (Nuavg = 2.4220) and the
x¼  2 ; and T ¼ 0: results of Shu et al. (Nuavg = 2.5560). Furthermore, a grid inde-
Ar An2
An pendence test was carried out for the nanofluid case for water
AT as the base fluid, Al2O3 nanoparticles, φ = 0.10, Ra = 105, and
Symmetry lines :W ¼ 0; X ¼ 0; and ¼ 0: L/D = 0.8. It was found that a grid size of 61 × 61 guarantees a
Ag
grid independent solution.
Due the lack of experimental data for natural convection in an
3. Numerical implementation annulus with the presence of nanoparticles, the present numerical
solution is validated by the experimental results of Kuhen and
Eqs. (10) through (12), with the corresponding boundary Goldstein [23] using Ra = 4.57 × 104 and Pr = 0.7. Three tempera-
conditions given in Eq. (15), are solved using the finite volume ture profiles at three different angles are shown in [25]. The
approach [18,19]. The diffusion term in the vorticity, energy, present results are in good agreement with other published data.
and kinematic equations is approximated by a second-order
central difference which gives a stable solution. Furthermore, a
second order upwind differencing scheme is adopted for the 5. Results and discussion
convective terms. The resulting algebraic equations are solved
using successive over/under relaxation method. Successive The range of Rayleigh number, volume fraction of
under relaxation was used due to the non-linear nature of the nanoparticles, and L/D ratio are Ra = 103–105, 0 ≤ φ ≤ 0.1,
governing equations especially for the vorticity equation. A and 0.2 ≤ L/D ≤ 0.8, respectively. The Prandtl number of the
fourth order accurate formula for the vorticity boundary base fluid is kept constant at 6.2. Fig. 3 presents Nusselt number
condition at the wall is adopted [20]: distribution around the inner cylinder surface using various

volume fractions of Al2O3 nanoparticles. Fig. 3(a) shows the
108W1;j 27W2;j þ 4W3;j variation using L/D = 0.8 and Ra = 105. The figure shows that,
X0;j ¼  2 : ð16Þ
around the inner cylinder surface, there are locations where the
18ðDnÞ2 AnAr
Nusselt number distribution is affected by the presence of
 E. Abu-Nada et al. / International Communications in Heat and Mass Transfer 35 (2008) 657–665 661

Fig. 2. Comparison of the results of the present work (solid lines) and Kuhen and Goldstein [24] experimental results, experimental data points: □: 0°, ◊: 90°, ○: 180°
(Ra = 4.7 × 104, Pr = 0.706, and L/D = 0.8).

nanoparticles and other locations (such as the plume region) temperature gradient at the cylinder surface and the thermal
where Nusselt number is not affected. The intensity of the conductivity ratio (note that other parameters are constants).
plume increases with the addition of nanoparticles as shown in Fig. 5 quantifies the weight of these two parameters. It is clear
Fig. 4(a) and (b). The dependency of Nusselt number on the from Fig. 5(a) that as the percentage of the Al2O3 nanoparticles
nanoparticle volume fraction can be further explained by increases, the temperature gradient decreases. This decrease is
examining Eq. (17). The equation shows that the Nusselt due to the enhanced effective thermal conductivity of the
number is influenced by two parameters, which are the nanofluid which is accompanied by an increase in the thermal

Fig. 3. Nusselt number distribution around inner cylinder surface using various volume fractions of Al2O3 nanoparticles (a) Ra = 105, L/D = 0.8 (b) Ra = 104, L/D = 0.8
(c) Ra = 103, L/D = 0.8 (d) Ra = 104, L/D = 0.4.
662 E. Abu-Nada et al. / International Communications in Heat and Mass Transfer 35 (2008) 657–665

diffusivity (see Fig. 5(b) and (c)). The increase in thermal nanoparticles for Ra = 103 always increase Nusselt number and
diffusivity causes the drop in the temperature gradients. By this increase become more pronounced in the plume region. The
inspecting Fig. 5(a) and (b) it is clear that outside the plume behavior in the plume region for Ra = 103 is different from that of
region, the increase in thermal conductivity ratio is greater than Ra = 105. The inertia forces at Ra = 103, are small and the
the decrease in temperature gradient. For example, at θ = 0 and conduction heat transfer become more significant due to the
when the volume fraction of the nanoparticles is increased from addition of nanoparticles. This will reduce the size of the plume
zero to 10%, the temperature gradient is decreased by approxi- and the size of the boundary layer which will increase the value
mately 20% and the thermal conductivity ratio of the nanofluid is of Nusselt number, Fig. 4(c) and (d). Fig. 3(d) shows that the
increased by about 30%. Accordingly, the Nusselt number addition of nanoparticles has an adverse effect on Nusselt
increases by increasing the volume fraction of the nanoparticles. number for Ra = 104 and L/D = 0.4.
However, within the plume region, the reduction in temperature An interesting comparison between various nanofluids is
gradient becomes approximately equal to the enhancement in the presented in Fig. 6. The figure shows that nanoparticles with
effective thermal conductivity which explains the independency of high thermal conductivity cause significant enhancement in heat
the Nusselt number from the volume fraction of the nanoparticles in transfer characteristics. However, within the plume region the
the plume region. effect of the nanoparticles type is less pronounced. The figure
Fig. 3(b) shows that the region that experience enhancement also shows the case of Ra = 10 3 and L/D = 0.8 where
in heat transfer becomes narrower. For this Rayleigh number, the nanoparticles with high thermal conductivity cause more
plume region and the thermal boundary layer become wider. For enhancement of heat transfer characteristics. However, an
example, for θ N 110°, the nanoparticles cause a decrease in interesting observation which is worth mentioning is that
Nusselt number. Fig. 3(c) shows that the presence of although Al2O3 has a low thermal conductivity, it shows higher

Fig. 4. Temperature isotherms using Al2O3 nanoparticles (a) Ra = 105, L/D = 0.8, φ = 0 (b) Ra = 105, L/D = 0.8, φ = 0.10 (c) Ra = 103, L/D = 0.8, φ = 0 (d) Ra = 103,
L/D = 0.8, φ = 0.10.
 E. Abu-Nada et al. / International Communications in Heat and Mass Transfer 35 (2008) 657–665 663

Fig. 5. (a) Temperature gradient at the cylinder surface using Al2O3 nanoparticles, Ra = 105, L/D = 0.8, φ = 0.10 (b) Effect of volume fraction of Al2O3 nanoparticles on
thermal conductivity ratio (b) Effect of volume fraction of Al2O3 nanoparticles on thermal diffusivity.

enhancement of heat transfer characteristics in the plume region It shows that for Ra = 105 and Ra = 103, the addition of
compared to the high thermal conductivity metals, such as Cu or nanoparticles increases the average Nusselt number. However,
Ag. The thermal conductivity of Al2O3 is approximately one for Ra = 104, a minor effect on the average Nusselt number is
tenth of Cu, as given in Table 1. However, a unique property of observed. Also, it is worth mentioning that for Ra = 104 and L/
Al2O3 is its low thermal diffusivity, Table 1. The reduced value D = 0.4, the addition of nanoparticles reduces the average
of thermal diffusivity leads to higher temperature gradients and Nusselt number where the average Nusselt number at φ = 0.1 is
therefore, more enhancement in heat transfer. Cu and Ag have equal to 3.855 and the Nusselt number at φ = 0 is 3.940. For this
high values of thermal diffusivity and, therefore, this reduces combination of Rayleigh number and L/D ratio, the thermal
temperature gradients which will affect the performance of Cu boundary layer becomes thicker (see Fig. 7(b) and (c)).
and Ag nanoparticles in the plume region. Fig. 6(c) shows the Therefore, temperature gradients become smaller and the
case of Ra = 104 and L/D = 0.4. In this case, nanoparticles such increase in thermal conductivity ratio is still small due to the
as TiO2 and Al2O3 reduce the Nusselt number and nanoparticles low thermal conductivity of Al2O3 nanoparticles. Accordingly,
such as Cu and Ag increase the Nusselt number. For TiO2 and the presence of nanoparticles decrease the Nusselt number.
Al2O3 nanoparticles, the decrease in temperature gradients is
greater than the increase in thermal conductivity ratio. Actually, 6. Conclusions
the increase in thermal conductivity ratio due to the presence of
TiO2 and Al2O3 is not large because of the relatively low value The main conclusions are summarized as follows:
of thermal conductivity of those nanoparticles. On the contrary,
Cu and Ag nanoparticles have high values of thermal - For high values of Rayleigh number and L/D ratio, an
conductivity which will increase the Nusselt number compared enhancement in heat transfer is achieved around the inner
to pure water case. However, the difference in heat transfer cylinder surface due to the presence of Al2O3 nanoparticles
enhancement by using different types of nanoparticles is small. except in the plume region.
Fig. 7 shows the variation of average Nusselt number with - For intermediate values of Rayleigh number (Ra = 104) the
different volume fractions of Al2O3 nanoparticles for L/D = 0.8. effect of Al2O3 nanoparticles on Nusselt number is less
664 E. Abu-Nada et al. / International Communications in Heat and Mass Transfer 35 (2008) 657–665

Fig. 6. Nusselt number distribution using different types of nanoparticles φ = 0.10 (a) Ra = 105, L/D = 0.8 (b) Ra = 103, L/D = 0.8 (c) Ra = 104, L/D = 0.4.

pronounced, and even a reduction in the Nusselt number is [3] K. Khanafer, K. Vafai, M. Lightstone, Buoyancy-driven heat transfer
enhancement in a two-dimensional enclosure utilizing nanofluids,
observed for L/D = 0.4.
International Journal of Heat and Mass Transfer 46 (2003) 3639–3653.
- For low Rayleigh number, the addition of Al2O3 nano- [4] A.G.A. Nanna, T. Fistrovich, K. Malinski, S.U.S. Choi, Thermal
particles causes an increase in Nusselt number all over the transport phenomena in buoyancy-driven nanofluids, Proc. ASME Int.
region around the inner cylinder surface. Mech. Eng. Congress RD&D Expo., IMECE2004-62059, Anaheim,
- For high Rayleigh number and high L/D ratio, nanopar- CA, 2004, pp. 1–8.
ticles with high thermal conductivity result in significant [5] A.G.A. Nnanna, M. Routhu, Transport phenomena in buoyancy driven
nanofluids – Part II, Proc. ASME Summer Heat Transfer Conf., SHTC—
enhancement in heat transfer. 72782, San Francisco, CA, 2005, pp. 1–8.
- For intermediate value of Rayleigh number, nanoparticles [6] N. Putra, W. Roetzel, S.K. Das, Natural convection of nanofluids, Heat and
with low thermal conductivity show adverse effect on the Mass Transfer 39 (8–9) (2003) 775–784.
Nusselt number, especially for L/D = 0.4. [7] D. Wen, Y. Deng, Formulation of nanofluids for natural convective
- For low Rayleigh number, nanoparticles with high thermal heat transfer applications, International Journal of Heat and Fluid Flow
26 (6) (2005) 855–864.
conductivity cause more enhancement in heat transfer. How- [8] R.-Y. Jou, S.-C. Tzeng, Numerical research of nature convective heat
ever, Al2O3 causes more enhancement in heat transfer in the transfer enhancement filled with nanofluids in a rectangular enclosure,
plume region. International Communications in Heat and Mass Transfer 33 (2006)
727–736.
[9] V. Trisaksri, S. Wongwises, Critical review of heat transfer characteristics
References of nanofluids, Renewable and Sustainable Energy Reviews 11 (2007)
512–523.
[1] S. Lee, S.U.S. Choi, J.A. Eastman, Measuring thermal conductivity of [10] X.-Q. Wang, A.S. Mujumdar, Heat transfer characteristics of nanofluids: a
fluids containing oxide nanoparticles, Journal of Heat Transfer 121 (1999) review, International Journal of Thermal Sciences 46 (2007) 1–19.
280–289. [11] H.C. Brinkman, The viscosity of concentrated suspensions and solutions,
[2] W. Daungthongsuk, S. Wongwises, A critical review of convective heat Journal of Chemical Physics 20 (1952) 571–581.
transfer nanofluids, Renewable and Sustainable Energy Reviews 11 (2007) [12] R.L. Hamilton, O.K. Crosser, Thermal conductivity of heterogeneous two-
797–817. component system, I&EC Fundamentals 1 (1962) 187–191.
 E. Abu-Nada et al. / International Communications in Heat and Mass Transfer 35 (2008) 657–665 665

Fig. 7. (a) Average Nusselt number versus volume fraction of Al2O3 nanoparticles, L/D = 0.8 and φ = 0.10, (b) Temperature isotherms Ra = 104, L/D = 0.4 φ = 0
(c) temperature isotherms Ra = 104, L/D = 0.4, φ = 0.10.

[13] Y. Xuan, W. Roetzel, Conceptions for heat transfer correlations of [20] E. Weinan, Jian-Guo Liu, Vorticity boundary condition and related issues
nanofluids, International Journal of Heat and Mass Transfer 43 (2000) for finite difference schemes, Journal of Computational Physics 124 (1996)
3701–3707. 368–382.
[14] S.E.B. Maiga, C.T. Nguyen, N. Galanis, G. Roy, Heat transfer behaviors of [21] G. Guj, F. Stella, Natural convection in horizontal eccentric annuli:
nanofluids in a uniformly heated tube, Supperlattices and Microstructures numerical study, Numerical Heat Transfer 27 (1995) 89–105.
35 (2004) 543–557. [22] C. Shu, K.S. Yeo, Q. Yao, An efficient approach to simulate natural
[15] G. Roy, C.T. Nguyen, P.R. Lajoie, Numerical investigation of laminar flow convection in arbitrary eccentric annuli by vorticity stream function
and heat transfer in a radial flow cooling system with the use of nanofluids, formulation, Numerical Heat Transfer 38 (A) (2000) 739–756.
Supperlattices and Microstructures 35 (2004) 497–511. [23] T.H. Kuhen, R.J. Goldstein, An experimental and theoretical study of
[16] A. Akbarinia, A. Behzadmehr, Numerical study of laminar mixed natural convection in the annulus between horizontal concentric cylinders,
convection of a nanofluid in horizontal curved tubes, Applied Thermal Journal of Fluid Mechanics 74 (1976) 695–719.
Engineering 27 (2007) 1327–1337. [24] T.H. Kuhen, R.J. Goldstein, Numerical solutions to the Navier–Stokes
[17] J.D. Anderson, Computational Fluid Dynamic: The Basics with Applica- equations for laminar natural convection, International Journal of Heat and
tions, McGraw-Hill, New York, 1995. Mass Transfer 23 (1980) 971–979.
[18] S.V. Patankar, Numerical Heat Transfer and Fluid Flow, Hemisphere [25] P. Wang, R. Kahawita, T.H. Nguyen, Numerical computation of the natural
Publishing Corporation, Taylor and Francis Group, New York, 1980. convection flow about a horizontal cylinder using splines, Numerical Heat
[19] H.K. Versteeg, W. Malalasekera, An Introduction to Computational Fluid Transfer 17 (A) (1990) 191–215.
Dynamic: The Finite Volume Method, John Wiley & Sons Inc, New York,
1995.