International Journal of Heat and Mass Transfer 59 (2013) 423432

Contents lists available at SciVerse ScienceDirect

International Journal of Heat and Mass Transfer

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

A numerical investigation of laminar ow of a water/alumina nanouid

P. Farias Alvario, J.M. Siz Jabardo , A. Arce, M.I. Lamas Galdo
Escola Politcnica Superior, University of A Corua, Mendizbal s/n, 15403 Ferrol, A Corua, Spain

a r t i c l e

i n f o

Article history:
Received 13 July 2012
Received in revised form 12 December 2012
Accepted 14 December 2012
Available online 21 January 2013
Keywords:
Nanouids
CFD particles eld
Heat transfer enhancement

a b s t r a c t
The present paper reports the results of a numerical investigation of a tubular developing laminar ow of
a water/alumina nanouid under constant heat ux conditions. The nanouid is modelled as a water/alumina nanoparticles mixture. Brownian and thermophoretic diffusion effects have been considered
through the energy and nanoparticle concentration eld equations. Though these modes of diffusion have
been suggested extensively in the literature, their effect on momentum and energy transport has not yet
been numerically analyzed. In addition, the model includes temperature and concentration effects over
the transport properties of the nanouid. A new CFD solver has been developed for the solution of the
present set of governing equations. Model solutions have been sought for conditions corresponding to
two experimental investigations carried out elsewhere. Base uid (water) solutions have been used to
asses the accuracy of the numerical model, the results being reasonably comparable to the experimental
ones and adequately correlated by the Churchill and Ozoe correlation. Nanouid solutions have been
obtained for nanoparticles volumetric concentrations of up to 6%. A concentration boundary layer has
been observed to develop along the wall of the pipe, a region which is progressively depleted of nanoparticles by the action of thermophoretic diffusion. A slight heat transfer enhancement has been found especially at the higher nanoparticle concentrations, with the maximum enhancement being of the order of
5%. The dimensionless numerical results, based on cross section (local) properties, fall within 25% with
respect to the experimental ones. The Churchill and Ozoe correlation seems to adequately correlate
numerical results for the range of Graetz numbers of the present investigation and for the lower range
of nanoparticles concentration. Heat ux effects seem to play a potential role in nanouid heat transfer
at the tube wall though their signicance has not been consistently investigated under the present
investigation.
2012 Elsevier Ltd. All rights reserved.

1. Introduction
Heat transfer enhancement is not a closed topic in the eld of
nanouids momentum and energy transport investigation, though
several arguments have been used to explain it [1]. Favourable characteristics and adequacy for heat transfer processes has motivated
the investigation of nanouids internal ow heat transfer, specially
during the last decade. Heat transfer enhancement promoted by
nanouids has been observed mostly by experimental studies
involving both internal laminar and turbulent ow. In a recently
published study [2], Buongiorno performed an order of magnitude
analysis of the different mechanisms that plausibly intervene in
momentum and heat transfer of a owing nanouid. The nanouid was assumed a mixture of a base uid and nanoparticles. In
his analysis, Buongiorno introduced, among other heat transfer
mechanisms, the diffusion of nanoparticles by brownian and thermophoretic effects. One of the most important contributions by this
Corresponding author.
E-mail addresses: mjabardo@udc.es, mjabardo@cdf.udc.es (J.M. Siz Jabardo).
0017-9310/$ - see front matter 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijheatmasstransfer.2012.12.033

study is the recognition of the importance of the effects of the

nanouid transport properties on heat transfer in the region
close to the heating wall under turbulent ow conditions.
A literature survey has revealed a signicant amount of publications dedicated to the investigation of internal ow heat transfer of
nanouids from the experimental point of view. Table 1 presents a
list of experimental studies considered as reference for the investigation reported herein. As one should expect, the list is not exhaustive given the number of nanouids heat transfer investigations
that have been reported in the literature. Considering Table 1 as
reference, it may be stated that most of the research is focused
on developed turbulent ow and developing laminar ow, with
test sections being mostly electrically heated. Volumetric concentrations of nanoparticles varied up to the order 10%, but generally
concentrations were lower than 6%, with one of the studies being
performed with such a low concentration as 0.01%.
The two main physical parameters that control nanouids
transport effects are the viscosity and conductivity. Both increase
with the nanoparticle concentration though the viscosity increment with respect to that of the base uid is signicantly higher

424

P. Farias Alvario et al. / International Journal of Heat and Mass Transfer 59 (2013) 423432

Nomenclature
~
U
C
D
d
dp
kB
L
q00
T
Tm
y

nanouid velocity
specic heat
diffusion coefcient
tube diameter
typical particles size (Diameter)
Botzmann constant, 1.3806504  1023/;(J/K)
Pipe length
heat ux
R
temperature
qnf C nf TU dA
mean temperature T m RA
q C U dA
A nf nf
distance from the wall

Dimensionless groups
T
h
dimensionless temperature h TTwwT
m
Gz
graetz number Gz RePr
x
d
NBT
Brownian/thermophoretic diffusion relation. See Eq. (8).
Nu
nusselt number Nu hd
j
Pr
prandtl number Pr Cjl
Re
Reynolds number Re qlUd

than that of the conductivity. As a result, one should expect a higher pressure drop increment than heat transfer with respect to the
base uid. This is not a clear cut aspect in the literature and will
not be treated in the present context, which is aimed at the heat
transfer investigation.
There is a generalized consensus in that the heat transfer coefcient increases with the nanoparticle concentration. The reported
studies generally relate the thermal performance of the nanouid
in terms of the heat transfer enhancement achieved with respect
to the base uid. Given the general assumption of considering
the nanouid as being a mixture of nanoparticles and base uid,
comparisons with the base uid have been performed in terms of
the internal ow governing dimensionless parameters, that is, Nusselt, Reynolds and Prandtl numbers. The reported results for both
turbulent and laminar regimes are far from being consistent. In
the case of turbulent ow, Pak and Cho [3] obtained a slight nanouid Nusselt number increment with respect to the classical Dittus
and Boelter correlation. Pak and Cho [3] proposed a correlation that
differs from the Dittus and Boelter one in the coefcient, 0.021 instead of 0.023, and the Prandtl number exponent, 0.5 instead of 0.4.
Pak and Cho [3] results have not been conrmed by Williams et al.
[5] whose nanouid results seem to be adequately correlated by
the Dittus and Boelter correlation. In contrast with the previous
studies, Xuan and Li [4] results indicate a signicant effect of nanoparticles concentration on the Nuselt number, for given Re and Pr
numbers. As a result, Xuan and Li [4] proposed a dimensionless
correlation that explicitly includes the nanoparticle concentration
in addition to a nanoparticle Pclt number which is referred to
the particle diameter.
The laminar ow trends are not different from those observed
under turbulent ow. In fact, Wen and Ding [7], for developing
laminar ow, found signicant heat transfer coefcient (dimensional) increments for the nanouid with respect to the base uid
for the same Reynolds number. They also found that, for the same
Reynolds number, Nu increases with the nanoparticle concentration and with respect to the Shahs correlation [13] for the average
Nusselt number from the entrance to the particular cross section.
Heris et al. [9] and [10] results display signicant increments of
the nanouid Nusselt number with respect to the base uid, for
the same Pclt number, with the differences being higher for
higher nanoparticle concentration. For the developing laminar region, Rea et al. [11], based on the same average velocity, obtained

Greek symbols
a
particles volumetric concentration
d
boundary layer thickness
j
thermal conductivity
l
viscosity
q
density
Subscripts

a
B
b
bf
nf
p
T
w
x

outer edge of particles boundary layer

Brownian
reference value for volumetric particles concentration,
temperature and Pr (uniform at inlet section)
base uid
nanouid
particles
thermophoresis
at the wall
local position

heat transfer coefcients increments with respect to the base uid

of 27% for 6% volumetric nanoparticle concentration, for alumina/
water nanoauid, and 3.5% increment for 3.5% concentration, for
a zirconia/water nanouid. However, Rea et al. [11] dimensionless
parameters, based on nanouid transport properties, are adequately correlated by an expression very close to the one proposed
by Shah [13] for the local Nusselt number under developing laminar ow.
Recently Anoop et al. [12] published an experimental study
dealing with the undeveloped laminar ow of an alumina/water
nanouid in a 4.75 mm internal diameter and 1200 mm length
copper tube. The investigation involved nanoparticles of two different diameters: 45 nm and 145 nm. The nanouid heat transfer
enhancement with respect to the base uid varied with both the
distance from the tube entrance and the particle diameter. It was
determined that ner nanoparticles enhancement is higher. Anoop
et al. [12] proposed a dimensionless heat transfer correlation in
terms of the undeveloped laminar ow conventional parameters
along with the nanoparticles diameter and concentration. The proposed correlation tted experimental data within 20%.
The observed inconsistencies in the reported experimental results regarding both the level and quality of heat transfer enhancement obtained with internal ow of nanouids indicate the need
for further research which could be performed either from the
experimental or theoretical (numerical) point of views. Several
models have been introduced to capture the patterns of nanoparticles distribution in the ow eld. Behzadmehr et al., (2007) [14],
proposed an approach under turbulent ow regime based on the
classical two phase eld equations. Their paper seems to be the
rst numerical approach to include nanoparticles distribution in
the internal ow of a nanouid. In their analysis, Behzadmehr
et al. [14] included effects of nanoparticles drift and proposed a
model for the mass balance of particles in the ow. The drift with
respect to the continuous phase is based on the assumption that
the nanoparticles are driven by both their own inertia and gravity.
However, previously, in the aforementioned paper, Buongiorno [2],
through an order of magnitude analysis, concluded that the nanoparticle drift could safely be neglected even for the smallest turbulent eddies. In addition, Buongiorno suggested negligibly small
gravity effects over the nanoparticles what clearly contradicts the
assumptions made by Behzadmehr et al. [14], who, on the other
hand, included brownian diffusion effects in their analysis through

425

P. Farias Alvario et al. / International Journal of Heat and Mass Transfer 59 (2013) 423432
Table 1
Several experimental references.
Author

Nanouid composition

Flow regime

Pak and Cho [3]

Base uid: water

Nanoparticles:Al2O3 (13 nm.)
and TiO2 (27 nm.)
a(%): up to 10%

Turbulent

Xuan and Li [4]

Base uid: water

Nanoparticles:Al2O3 (46 nm.)
and TiO2 (60 nm.)
a(%): up to 2%

Turbulent

Williams et al. [5]

Base uid: water

Nanoparticles:Al2O3 (13 nm.)
and ZrO2 (27 nm.)
a(%):
0.9 and 3.6% for Al2O3
0.2 and 0.9% for ZrO2

Turbulent

Djajadiwinata et al. [6]

Base uid: water

Nanoparticles:CuO (28 nm.)
a(%): 0.01%

Turbulent

Wen and Ding [7]

Base uid: water

Nanoparticles:Al2O3 (27/56
nm.)
a(%): up to 1.6%

Laminar

Yang et al. [8]

Base uid:
Transmission oils
Synthetic oils
Nanoparticles: graphite, aspect
ratio: 0.02
2 and 2.5 % weight

Laminar

Heris et al. [9]

Base uid: water

Nanoparticles: Al2O3 (20 nm.)
a(%): up to 2.5%

Laminar

Rea et al. [11]

Base uid: water

Nanoparticles: Al2O3 and ZrO2
a(%):
up to 6.0% for Al2O3
up to 1.32% for ZrO2

Laminar

Anoop et al. [12]

Base uid: water

Nanoparticles: Al2O3 (45 nm
and 150 nm)
a(%): 1 to 6%

Laminar

similar analysis, avoiding any reference to the nanoparticle distribution equation. Kalteh et al. [21] sought a numerical solution of a
nanouid owing in an isothermal parallel plates microchannel
under laminar ow regime conditions. Their model assumes that
the nanouid is a two-phase mixture of nanoaparticles and base
uid. The proposed model consists of the typical average eld
equations under steady ow conditions along with such effects
over the nanoparticles as drift, drag, added mass and particle/particle interaction. The model also requires the introduction of an
apparent viscosity of the particles phase. One of the most signicant paper conclusions is that the nanouid can be considered as
a homogeneous mixture. In addition, heat transfer enhancement
is observed for higher Reynolds numbers and volumetric nanoparticle concentrations and for smaller nanoparticles diameter.
The present paper reports a numerical heat transfer study of the
internal hydrodynamic and thermally developing laminar ow of a
nanouid, assumed as a mixture of the base uid and nanoparticles, aiming at devising the transport mechanisms that govern
the heated wall heat transfer. Temperature and concentration
dependence of the nanouid transport properties has been included in the numerically obtained solution of the eld equations.
One of the eld equations is the nanoaparticles distribution equation which includes brownian and thermophoretic effects as suggested by Buongiorno [2]. The solution of the coupled set of eld
equations provides the development of the hydrodynamic, thermal
and concentration boundary layers along with heat transfer coefcients. Numerical results are evaluated through experimental data
obtained elsewhere.
2. Governing equations

a model proposed by Xuan and Li [15]. Mirmasoumi and Behzadmehr [16] applied the previous two-phase ow model, keeping
nanoparticle inertia effects, to developing laminar ow at the relatively low Reynolds number of 300. According to Mirmasoumi and
Behzadmehr [16] results, nanoparticles migrate toward the wall,
probably carried both by natural convection and brownian diffusion, a trend that clearly contradicts the one proposed by Buongiorno [2] for a heated nanouid.
Savithiri et al. [17] performed a similar scaling analysis as the
one by Buongiorno [2] aiming at determining which dynamic
nanoparticle/base uid interaction effects were signicant. It was
found that slip mechanisms dominate in the range of higher nanoparticle diameters (10 to 80 nm) whereas Brownian and drag
forces dominate in the lower range (less than 10 nm). In addition,
Brownian forces dominate in the lower nanoparticles concentration range and higher temperatures. Magnus effect was observed
to be signicant at higher temperatures. Savithiri et al. [17] also
investigated relative effects of nanoparticles diameter, concentration, and shape.
Oztop et al. [18] proposed a model based on the vorticity and
energy conservation equations. Brownian diffusion is included in
the model through the Xuan and Lis [15] model, though no reference is made to the nanoparticle distribution equation and their
distribution in the ow eld. Subsequent studies, [19,20] perform

The study reported herein deals with the ow of a nanouid in

the entrance region of a tube under laminar ow conditions. The
ow is assumed hydrodynamic and thermally undeveloped. The
entrance conditions are assumed to be the conventional ones, that
is, uniform velocity, temperature and nanoparticle concentration
at the test section entrance. The nanouid is considered as a mixture, designated by the subscript nf, of a base uid, designated by
the subscript bf, and nanoparticles, designated by the subscript p.
The following are the governing equations:
 Mixture continuity:

@ qnf
~ 0
r  qnf U
@t

 Momentum:

~
@qnf U
~ P r  l r
~U
~U
~U
~ r
~r
~T 
r  qnf U
nf
@t

 Energy:

@qnf C nf T
~
r  qnf C nf UT
@t
~ T q C p DB r
~a  r
~ T DT
r  jnf r
p

~T  r
~T
r
T

!
3

 Particles distribution:

~
@a
~a r  DB r
~ a DT rT
r  U
T
@t

!
4

Though the expressions of the set of governing equations are

the conventional ones, some peculiarities in the present form, particularly those related to the nanoparticles concentration equation,

426

P. Farias Alvario et al. / International Journal of Heat and Mass Transfer 59 (2013) 423432

require further explanation. The following are some explanatory

notes related to this set of equations:

Table 2
Transport properties correlations for the nanouid and the base uid.
Equation

 Brownian and thermophoretic diffusion effects are included in

the Mixture Energy equation, the second right hand side term
of Eq. (3), and the nanoparticle Distribution equation, right
hand side term of Eq. (4).
 The following is the assumed expression for the Brownian diffusion coefcient, DB, [2]:

DB

kB T
3plbf dp

qbf = Cq0 + Cq1T + Cq2T2

lbf = Cl0 + Cl1T + Cl2T2 + Cl3T3 + Cl4 T4
jbf = Cj0 + Cj1T + Cj2T2
Cbf = CC0 + CC1T + CC2T2
qnf = qbf(1  a) + qpa
h i
Al a

lnf lbf e Bl a
jnf = jbf(1 + Aja + Bja2)
qnfCnf = qbfCbf(1  a) + qpCpa
See Table 4 for values of coefcients for water and water/alumina nanouid.

 The assumed expression for the thermophoretic diffusion coefcient, DT, is the following [2]:

DT Scoef

lbf
lbf
jbf
a 0:26
a
qbf
2jbf jp qbf

 As previously mentioned, the nanouid transport properties

variation with temperature and nanoparticles concentration
are considered in the present analysis along with the dependence of the base uid on the temperature. Table 2 presents
the assumed expressions for the mixture transport properties [5,11]. It must be stressed that the transport properties are coupled with the governing equations since
they depend on the temperature and nanoparticles
concentration.
 There is another important issue related to the governing
equations, namely the boundary conditions. In the case of
Eqs. (1)(3), the boundaries are treated as usual, and are generally available in all CFD software packages. In addition, they
do not need a special treatment to use and adapt them to the
present case of nanouids. The last equation, Eq. (4), requires
unusual boundary conditions. There are no special comments
regarding symmetry, periodicity, xed value, etc. However,
since the tube wall is impermeable, the nanoparticles mass
ow rate there must be zero. Thus, considering the fact that
there is no particle transport by convection at the wall and
the ow is in steady state, the nanoparticles diffusion by thermotheresis is equal and opposite to the Brownian diffusion,
that is,

~ aj 
r
w

~ Tj
DT r
w
DB T

It should be noticed that this boundary condition, Eq. (7), strongly

depends on the solution eld. As a result, the algorithm is expected to be of lower convergence rate than conventional CFD
solvers. In addition to this convergence problem, another potential
problem might arise related to negative diffusion coefcients that
must be circumvented since this can be the cause of diverging
solutions.
 Through a dimensional analysis, Buongiorno [2] obtained the
following dimensionless relation between Brownian and Thermophoretic diffusion that he designated by NBT:

NBT

ab DB T a
DTDT

ab DB T a jnf
q00 d

a DT

3. Numerical algorithm
The open source code OpenFoam [22] has been adjusted to
comply with the following requirements related to solution procedure of the set of governing equations:
 The solver must work with the present form of Eq. (1) since the
Boussinesq approximation is not adequate. It has been determined that for a velocity divergence free ow, the mass ow
rate is not conserved. The use of this assumption has led to
errors of up to about 20% for a nanouid owing in a pipe at
Reynolds number of 750. Errors of this order in the velocity eld
result in low accuracy Nusselt numbers.
 The nanoparticles distribution equation, Eq. (4), had to be overrelaxed to prevent oscillations that could lead to the nal divergence of the whole process.
 A fast convergence should not be expected since boundary conditions like Eq. (7) depend themselves on the solution.
 Special care must be taken in keeping all diffusion coefcients
higher than zero every time and over the whole control volume.
In the design of this solver any kind of summation to the diffusion coefcients has been avoided. Usually most solvers overcome this problem by introducing an articial diffusion.
 In the last case of Table 5 the thermophoresis diffusion coefcient, DTT , has achieved values varying from 1.99662  1013 up
to 6.73357  1013 m2 s1 K1. The brownian diffusion coefcient, DB, assumes values from 1.09306  1011 up to
2.75834  1011 m2 s1.
 In order to avoid possible spurious values when calculating both
the uid properties and the diffusion coefcients, values of the
nanoparticles volumetric concentration must be positive and
lower than twice the bulk concentration during the numerical
procedure. However, values of particles concentration have
not reached these limits in the solutions shown in the present
study.
The procedure for the pressurevelocity coupling is the PISO
algorithm over a user dened number of loops.
4. Numerical model. Case set up

where the subscript b designates bulk nanouid properties. This

dimensionless parameter controls the nanoparticles concentration
in the region close to the wall. In the case of heating walls, the
temperature gradient tends to diffuse nanoparticles away from
the wall by thermophoresis, whereas Brownanian diffusion, as a
result of the concentration gradient, tends to move them closer
to the wall. Thus, the lower the value of NBT, the lower must be
the average nanoparticles concentration in the region close to
the wall.

4.1. Nanouid selection

The nanouid selected for this study is made up of water, as a
base uid, and alumina (Al2O3) nanoparticles. Transport properties
of alumina and the nanoparticles assumed diameter are shown in
Table 3. Water properties have been assumed to be temperature
dependent and their variation with temperature has been determined by curve tting data from the 1995 Formulation for the
Thermodynamic Properties of Ordinary Water Substance for General and Scientic Use, issued by The International Association

427

P. Farias Alvario et al. / International Journal of Heat and Mass Transfer 59 (2013) 423432
Table 3
Alumina properties used in the solver.

Table 5
Cases numerically studied.

Concept

Value

Author (Exper.)

aBULK

Particles diameter dp
Particles density qp
Particles conductivity jp
Specic heat Cp

46 nm
3.920  103 kg/m3
40.0 W/m K
0.880 k J/kg K

Wen et al. [7]

Wen et al. [7]
Wen et al. [7]
Wen et al. [7]
Wen et al. [7]
Wen et al. [7]
Rea et al. [11]
Rea et al. [11]
Rea et al. [11]
Rea et al. [11]

Pure water (0.00%)

Pure water (0.00%)
Pure water (0.00%)
1.60 %
1.60 %
1.60 %
2.76 %
2.76 %
6.00 %
6.00 %

Table 4
Numerical coefcients for expressions of transport properties of base uid and
nanouid (from Table 2).
Concept
Coefcient
Coefcient
Coefcient
Coefcient
Coefcient
Coefcient
Coefcient
Coefcient
Coefcient
Coefcient
Coefcient
Coefcient
Coefcient
Coefcient
Coefcient
Coefcient
Coefcient
Coefcient

Cq0
Cq1
Cq2
CC0
CC1
CC2
Cl0
Cl1
Cl2
Cl3
Cl4
Cj0
Cj1
Cj2
Al
Bl
Aj
Bj

Value

Units (S.I.)

759.3
1.8475
3.52615  103
5.52022  103
8.42432
1.32352  102
0.437721
5.0587  103
2.20399  105
4.28147  108
3.12531  1011
0.748304
7.42386  103
9.64742  106
4.91
0.2092
4.5503
0

kg m3
kg m3 K1
kg m3 K2
k J kg1 K1
k J kg1 K2
k J kg1 K3
kg m1 s1
kg m 1 s1 K1
kg m1 s1 K2
kg m1 s1 K3
kg m1 s1 K4
W m1 K1
W m1 K2
W m1 K3
[]
[]
[]
[]

for the Properties of Water and Steam (IAPWS) [23]. The curve tting coefcients for each property can be found in Table 4.
The numerical results from the present investigation have been
evaluated by experimental data from the investigation by Wen
et al. [7] and Rea et al. [11]. Both studies have been carried out under developing internal laminar ow regime. Table 5 displays the
characteristics of the nanouid for both investigations along with
the maximum nanoparticles volumetric concentrations. Rea et al.
[11] performed their investigation with alumina and zirconia
nanoparticles, whereas Wen et al. [7] worked with alumina, the
ones that have been considered in the present investigation.
Though the test section in the Wen et al. [7] study was horizontal
whereas Rea et al. [11] used a vertical one, pipe orientation has
been neglected since natural convection effects are marginal for
the range of Reynolds number considered in the present study. A
nanoparticle average diameter of 46 nm has been used in the present calculations, a value which is the average nanoparticle diameter in both reference studies.
4.2. Geometry selection
The pipe diameter of the test section simulated in the present
numerical study is 4.5 mm, which is the same as that of both the
Wen et al. [7] and Rea et al. [11] investigations. The length of the
test section depends on the reference study, being equal to
970 mm for the Wen et al. [7] and 1010 mm for Rea et al. [11].
4.3. Boundary conditions and physical aspects
It must be stressed that the aim of the present study is to
numerically determine the variation of the local (cross section)
Nusselt number. For that purpose the following boundary conditions have been used in order to comply with the experimental
conditions reported by Wen et al. [7] and Rea et al. [11]:

Reynolds
750
1250
1750
750
1250
1750
555
1113
553
940

 Pipe entrance section:

Constant and uniform velocity depending on the chosen Reynolds number.
As usual in this kind of models, the pressure gradient is
assumed equal to zero.
Constant and uniform temperature which, in the present
case, was set equal to 283 K for all the Wen et al. [7] cases.
A uniform inlet temperature is provided by Rea et al. [11].
This temperature varied from case to case.
Constant and uniform nanoparticles volumetric concentration. In the present paper, just pure water and the uniform
values of Table 5.
 Solid wall:
No slip for the velocity
As usual in this kind of models, the pressure gradient is
assumed equal to zero.
A xed and uniform value of the heat ux along the tube
wall, q00 , which is not equivalent as assuming a uniform temperature gradient since the thermal conductivity of the
nanouid varies along the tube. This boundary condition is
written according to the following expression:

~ Tj
r
w

q00

jnf

The heat ux has been assumed as the one corresponding to

the maximum test section power reported by Wen et al. [7],
whereas the energy supply for each Rea case has been determined from information provided by Rea et al. [11].
Eq. (7) has been used for particles concentration.
 Pipe exit section: The axial gradient of all the signicant physical parameters (velocity, temperature, concentration) has been
assumed to be equal to zero.
4.4. Mesh and numerical aspects
The numerical model has been solved as two-dimensional with
axial symmetry. The discretization is performed in a 5 wedge
geometry. The mesh is structured and developed in two blocks.
The rst one models the development of the boundary layers from
the entrance to a section L/20 apart. This block contains 90 longitudinal (ow direction) cells and 250 radial ones, overall grid increase rate is 2.0 longitudinally and 15.0 in the radial
direction.The second block is connected to the rst one and completes the rest of the pipe length. The radial discretization is the
same as the one in the rst block whereas the longitudinal direction contains 1235 equally spaced cells. Since the Rea et al. [11]
test section length was a bit larger than the Wen et al. [7]
(1010 mm against 970 mm), a new block with similar discretization was attached to the second one. The mesh convergence has
been checked only for the rst block. The criterion for mesh convergence consisted in checking the maximum wall temperature

428

P. Farias Alvario et al. / International Journal of Heat and Mass Transfer 59 (2013) 423432

at the end of the rst block. According to this criterion, the mesh
was considered acceptable when wall temperature differences
from one mesh to the immediately rened one were lower than
0.01 K. The CPU time, for each case study considered herein, running in 8 parallel 1.2 Gz processors, was around four weeks. It must
be emphasized that the designed solver includes gravity as a
momentum source though, for the cases considered in the present
paper, gravity has been set to zero in order to avoid natural convection effects.
5. Results
5.1. Assessment of the numerical procedure
The numerical results have been assessed as previously mentioned by experimental data from the Wen et al. [7] and Rea
et al. [11] experimental investigations. A preliminary check out
of the numerical results was performed through a comparison with
analytical solutions for single phase water owing under developing laminar ow. However, no analytical solution has been found
for comparison purposes in the literature for the simultaneous
hydrodynamic and thermal development of laminar ow in a pipe.
Some of the available solutions assume Langhaar velocity proles
as starting point [24,25]. Others are the result of either numerical
or curve tting procedures [26,13,27]. In addition, these developing laminar ow solutions assume constant transport properties
of the uid, an aspect not contemplated in the present paper,
which includes the temperature effect over the properties. Thus,
the comparison of numerical results with available correlations
could be considered acceptable if the deviations of uid temperatures (along the test section) from the average one keep within reasonable limits. One of the correlations considered for comparison
purposes is the one proposed by Churchill and Ozoe [27], which assumes the following expression:

8
8
93 913
2>
>
>
>
>
>
>
>
>
>
pGz
>
>
>
<
<
=
=>
4
Nux
19:4
1
1
1




 p 2 6 >
h
 p 2 3 >
>
>

2 i12
>
>
>
>
Gz
>
>
>
4
>
: 1 Pr 3 1 4Gz
>
;>
4:364 1 29:6
:
;
29:6
0:0207
10
Wen et al. [7] referred their data to the Shahs correlation [13],
though using its average Nu version, whereas Rea et al. [11] proposed a correlation for the local Nu very close to the Shahs one,
which is written as:

8
 1 13
1
>
>
1
if Gz
6 0:00005
1:302 Gz
>
>
<
 1 13
1
if 0:00005 6 Gz
6 0:0015
Nux 1:302 Gz  0:5
>
>
h
i
>
 1  0:506 41 1
>
3
1
: 4:364 8:68 10
Gz
if Gz
e
P 0:0015
Gz
11
The numerical dimensionless parameters have been determined for
transport properties based on the average temperature of the uid
over the particular cross section. In addition, the local heat transfer
coefcient has been determined from the Newtons cooling law at
the particular cross section with the uid temperature being the local average temperature. The average temperature has been determined according to the conventional procedure. Effects of
temperature and particle concentration on the density and specic
heat were considered for both water (pure uid) and nanouid in
each cross section. It must be added at this point that due to the
variation of the viscosity from the center of the pipe to the wall,
the velocity will tend to be higher than that for a constant viscosity
in the region close to the wall, with the opposite occurring at the

center of the pipe. Thus the velocity prole resulting from the present numerical solution must be slightly deformed with respect to
both the developing and the parabolic developed velocity proles.
This trend must lead to a slight heat transfer enhancement with respect to the constant property solutions over the whole length of
the pipe even under developing ow conditions.
A comparison of heat transfer experimental results from the reference investigations with those obtained in the present numerical
study is shown in Fig. 1 in terms of the conventional dimensionless
parameters: Nu and Gz numbers. The curves representing the correlations by Curchill and Ozoe [27] and Shah [13] are overlaid in
the same plot. As can be noted, the results from the present investigation have been labelled with different symbols corresponding
to different Reynolds numbers. As should be expected, all the
numerical data fall into a single well behaved curve, which is closely reproduced by the Churchill and Ozoe [27] correlation. Though
slightly higher than Rea et al. [11] results in the range of higher Gz,
present numerical data reproduce reasonably well those from that
investigation whereas generally under predict those from the Wen
et [7] study. In general, all data plotted in Fig. 1 tend to coincide for
lower values of the Graetz number, with higher differences being
observed at higher values of this dimensionless parameter. The
range of high values of Gz corresponds to regions of the tube closer
to the entrance, a region characterized with high rates of heat
transfer, for which, published data, either experimental or theoretical, present higher differences. Wen et al. [7] results are clearly
above the rest, including the Shahs correlation ones. It is interesting to note at this point that in their paper, Wen et al. [7] claim that
their results are closely correlated by the Shahs correlation, which
is a correct statement, though they referred to the correlation corresponding to the average Nu number from the entrance to the
particular cross section, which is higher than the local Nu.
As a concluding remark, it can be stated that, given the observed
trends in the plot of Fig. 1, the numerical procedure of the present
investigation presents reliable results that conrm its adequacy as
an analysis tool of nanouids heat transfer despite the added difculty of the inclusion of nanoparticles and its diffusion effects in
the region close to the wall.
5.2. Nanouid results
As previously mentioned, the cases that have been considered
in the numerical investigation reported herein are those related
to the experimental studies carried out by Wen et al. [7] and Rea
et al. [11], whose operational conditions are reproduced in Table 5.
Accordingly, the nanouid to be considered is made up of water
and alumina, whose volumetric concentrations are also shown in
Table 5.
The development of the concentration boundary layer might
signicantly affect the transport properties in the region close to
heated wall, and, as a result, affect the rate of heat transfer. Thus,
it would be interesting to show how the nanoparticles concentration behaves along the tube. The radial variation of the concentration and its development along the length of the tube is shown in
the plot of Fig. 2 for a bulk volumetric concentration of 6.0% and a
Reynolds number of 940. The nanoparticle volumetric concentration with respect to the bulk one at the entrance (6.0%) is plotted
against the dimensionless distance from the wall at cross sections
located at several distances from the entrance of the pipe. The
increment of concentration with the distance from the wall at each
cross section presents a boundary layer resemblance. It can be
noted that the initial uniform concentration at the entrance section
progressively is deformed, with nanoparticles being displaced from
the wall region towards the center of the tube, thus depleting the
wall region of nanoparticles. This is an expected result if thermophoretic diffusion effects are considered.

429

P. Farias Alvario et al. / International Journal of Heat and Mass Transfer 59 (2013) 423432

Nu [-]

15

0.040

Churchill
Shah
Numeric Re=750
Numeric Re=1250
Numeric Re=1750
Wen et al (Water)
Rea et al (Water)

x = 0.0cm
x = 5.0cm
x = 16.2cm
x = 30.0cm
x = 44.0cm
x = 58.0cm
x = 89.0cm
x = 100.0cm

0.035
0.030
0.025

y/d [-]

20

10

0.020
0.015
0.010

0.005
0
0.001

0.01

0.1

1/Gz [-]
Fig. 1. Comparison of numerical results with experimental ones from [7,11].

That thermophoretic effects are signicant even at the furthest

downstream sections from the entrance can be demonstrated by
the value of the dimensionless parameter NBT, previously referred.
In fact, at the test Section 1000 mm downstream from the entrance,
for the same conditions as those of Fig. 2, the wall temperature is
equal to 345.3 K, whereas the temperature at the edge of the concentration boundary layer, whose thickness (99%) has been estimated as being equal to 2.9  104 m, is equal to 329.8 K. The
local brownian and thermophoretic diffusion coefcients at the edge
of the concentration boundary layer are equal to 2.091  1011
m2 s1 and 1.239  1010 m2 s1, with the corresponding NBT value
being equal to 0.217. This value clearly illustrates the relative dominance of thermophoretic over Brownian diffusion effects.
Experimental results from the reference investigations are plotted along with the corresponding numerical data in the plot of
Fig. 3(a) in terms of the Nusslet and Graetz numbers and for different nanoparticles volumetric concentrations. The correlations by
Shah [13] and Churchill and Ozoe [27], based on average transport
properties of the nanouid, have been overlaid in the plot for comparison purposes. As observed in the case of water, deviations of
numerical results from the experimental ones lessen in the range
of lower Graetz numbers. Higher differences are observed in the
higher Graetz numbers range. It is interesting to note the signicant differences with Rea et al. [11] data for 6.0% of nanoparticles
concentration and Reynolds number of 553, for which the numerical results are clearly higher than the experimental ones. Given
the closeness of numerical and Rea et al. [11] results for the rest
of the data, this result is clearly peculiar. Generally, numerical results are higher than data from Rea et al. [11]. An opposite trend is
observed with respect to Wen et al. [7] data for a nanoparticle concentration of 1.6%. It is interesting to note in the plot of Fig. 3(a)
that experimental data points are slightly displaced to the left
(higher Graetz number) with respect to the corresponding numerical one. This is due to the fact, already mentioned, that numerical
transport properties of the nanouid are averaged at the particular
cross section whereas the experimental ones are determined as the
average over the whole test section.
A general trend of the numerical with respect to experimental
results for the case studies considered herein can be obtained from
the plot of Fig. 3(b), where numerical Nusselt numbers are plotted
against the experimental ones. Data for 0% nanoparticle concentration (pure water) have been included in the plot. As noted, most of
numerical data fall within 25%. Data points from Rea et al. [11]
investigations corresponding to the 6% concentration and Reynolds
number of 553 along with two data points from Wen et al. [7] fall
outside this range.

0.000
0.6

0.7

0.8

0.9

1.1

/b [-]
Fig. 2. Particles concentration proles for Re = 940 and ab = 0.060.

The next step in the present numerical analysis is to investigate

the heat transfer enhancement produced by the presence of nanoparticles in the base uid. The plot of Fig. 4(a) could be used for
that purpose. As in previous plots, the Shah [13] and Churchill
and Ozoe [27] correlations are overlaid in that plot along with
the numerical data points. The heat transfer enhancement related
to the nanoparticles concentration is clearly displayed in that plot,
where the highest Nusselt numbers are attained under the highest
nanoparticle concentration (6%), diminishing for the other two
concentrations. As in the case of pure water, the Churchill and Ozoe
[27] correlation closely correlates the lower concentrations (0, 1.6,
and 2.7%) data. However, at the higher concentrations, the data
points clearly deviate upwards from that correlation.
The numerical Nusselt number is plotted against the Churchill
and Ozoe [27] correlation in Fig. 4(b), where data for water are
overlaid. The heat transfer enhancement promoted by the nanoparticles is clearly observed, especially at the higher concentrations. Heat transfer enhancements of up to 5.1% with respect to
water have been achieved.
As previously observed, the Churchill and Ozoe [27] correlation
ts quite adequately data for pure water and low nanoparticle concentrations, at least for the range of Graetz numbers of the present
study. However, numerical data tend to overcome the correlation
for lower values of Graetz, corresponding to lower values of the
Nusselt number. This trend is a clear indication that the actual fully
developed laminar ow Nusselt number might attain values a bit
larger than the accepted one of 4.36, which is obtained under constant transport properties. In fact, the presence of nanoparticles
and local temperature variations affect the transport properties
which in turn might affect the heat transfer rate at the heated
wall.
The cross section variation of the transport properties is illustrated in Fig. 5 where the local Prandtl number with respect to
the bulk one for the exit cross sections of different cases from
the present study is plotted against the dimensionless distance
from the wall. It can be noted that though the Prandtl number of
pure water is of the order of 40% that of the bulk uid, its variation
in the region close to the wall is small compared with that of the
nanouids. It is clear that Prandtl gradient at the wall increases
with the nanoparticles concentration. The region where the Prandtl
number variation is more signicant clearly delimits the concentration boundary layer, whose edge is very close for the three
cases. This statement is conrmed by the nanoparticles concentration variation with distance from the wall, not shown for space
saving purposes. It is interesting to emphasize that the Prandtl
number in the region close to the wall is affected mostly by the

430

P. Farias Alvario et al. / International Journal of Heat and Mass Transfer 59 (2013) 423432

25

15

Churchill
Shah
Num. 6.00% Re=940
Num. 6.00% Re=553
Num. 2.76% Re=1113
Num. 2.76% Re=555
Num. 1.60% Re=1750
Num. 1.60% Re=1250
Num. 1.60% Re=750

20

Nu [-]

20

Nu [-]

25

Churchill
Shah
Num. 6.00% Re=940
Rea 6.00% Re=940
Num. 6.00% Re=553
Rea 6.00% Re=553
Num. 2.76% Re=1113
Rea 2.76% Re=1113
Num. 2.76% Re=555
Rea 2.76% Re=555
Num. 1.60% Re=1750
Wen 1.60% Re=1750
Num. 1.60% Re=1250
Wen 1.60% Re=1250
Num. 1.60% Re=750
Wen 1.60% Re=750

15

10

10

5
0.001

0.01

0.001

0.1

1/Gz [-]

0.01

0.1

1/Gz [-]

(a)

(a)
12

20

5%

11
10

14

Num. Nu [-]

Num. Nu [-]

16

-25

12
10

Water Re=750
Water Re=1250
Water Re=1750
NF (6.00%) Re=940
NF (6.00%) Re=553
NF (2.76%) Re=1113
NF (2.76%) Re=555
NF (1.60%) Re=1750
NF (1.60%) Re=1250
NF (1.60%) Re=750

8
6
4

0%
+1
%
+5

+2

18

10

12

14

16

18

9
8
Water Re=750
Water Re=1250
Water Re=1750
NF (1.60%) Re=750
NF (1.60%) Re=1250
NF (1.60%) Re=1750
NF (2.76%) Re=555
NF (2.76%) Re=1113
NF (6.00%) Re=553
NF (6.00%) Re=940

7
6

20

Exp. Nu [-]

(b)

10

11

12

Churchill Nu [-]

(b)

Fig. 3. (a) Local Nusselt number versus the inverse of the Graetz number; (b)
numerical versus experimental data.

Fig. 4. (a) Local numerical Nusselt number versus the inverse of the Graetz
number; (b) numerical data compared with the Churchill and Ozoe correlation.

nanoparticles concentration with temperature effects being rather

limited as conrmed by the pure water trend.
The Prandtl number variation shown in Fig. 5 must affect the
cross section velocity and temperature distribution. This effect
has been suggested by Buongiorno [2] in his order of magnitude
analysis and modeling according to which the potential increment
of the heat transfer coefcient in a nanouid is closely related to
the reduced value of the Prandtl number in the viscous sublayer
under turbulent ow conditions. The plots of Fig. 6(a) and (b) display the variation of the dimensionless velocity and temperature
with the dimensionless distance from the wall for the same nanouids and cross sections of the plot of Fig. 5. The following are
some conclusions that can be drawn from these plots:

cases, including the one for pure water, deviates from the parabolic prole as a result of both temperature and concentration
effects over the transport properties, Fig. 6(a). Higher deviations
are noted as the nanoparticles concentration increases.
 The velocity gradient at the wall is clearly affected by the temperature and nanoparticles concentration effect over the transport properties especially over the viscosity. The higher velocity
gradient corresponds to the higher nanoparticles concentration.
The effect of the other concentrations is difcult to discern.
 The effect of the nanoparticles concentration is apparent in the
case of the dimensionless temperature variation, Fig. 6(b). Two
trends are clearly discernable in this plot: (i) the dimensionless
temperature at the center of the tube diminishes with the nanoparticles concentration; and (ii) the temperature gradient at the
wall increases with the concentration.

 All the cases considered in Figs. 5, 6(a) and (b) correspond to

experimental conditions of the studies considered herein, including the heat ux, which might differ from case to case. A consistent investigation of heat ux effects over the momentum and
energy transport at the wall has not been pursued in the present
study though it might play some role given its potential effect
over the nanoparticles concentration distribution.
 Given the order of magnitude of the Prandtl number in all the
cases of Fig. 5, the ow must be dynamically developed at the
given cross sections. However, the velocity distribution in all

The plot of Fig. 7 is similar to that of Fig. 6(b). However, instead

of the exit sections, the plot includes those sections that present a
common value of the inverse of the Graetz number, in this case of
the order of 0.0094 for each concentration. It can be noted that the
cross section (with the same Graetz number) is displaced downstream as the concentration increases due to the increment of
the Prandtl number. The curves corresponding to each concentration are very close to each other, especially the lower range ones,

431

P. Farias Alvario et al. / International Journal of Heat and Mass Transfer 59 (2013) 423432

0.050
x = 100.0 cm. (6.00%) Re=940
x = 100.0 cm. (2.76%) Re=1113
x = 97.0 cm. (1.60%) Re=750
x = 97.0 cm. (0.00% Water) Re=750

0.040

x = 100.0 cm. (6.00%) Re=940 1/Gz=0.0094

x = 46.6 cm. (2.76%) Re=1113 1/Gz=0.0094
x = 31.7 cm. (1.60%) Re=750 1/Gz=0.0094
x = 20.2 cm. (0.00% Water) Re=750 1/Gz=0.0094

0.5

0.03
0.025
0.02

0.030

y/d [-]

y/d [-]

0.4

0.020

0.3

0.015
0.01
0.005

0.2

0
0

0.010

0.1

0.2

0.3

0.4

0.5

0.6

0.15

0.2

0.2

0.4

0.6

0.8

1.2

1.4

[-]

Pr/Pr b [-]

Fig. 7. Dimensionless temperature proles for cross sections with Graetz number of
the same order.

Fig. 5. Prandtl proles.

0.2

(in fact, it is equal to the Nusselt number), it can be concluded that

the Nusselt number is not an exclusive function of the Graetz number but depends on the concentration too, increasing with it. This is
a contrasting result with the typical laminar heat transfer correlations, such as the Churchill and Ozoe [27] and Shah [13], which are
expressed in terms of the Graetz number. In addition, the observed
trends in Fig. 7 conrm the heat transfer enhancement promoted
by the nanoparticles beyond the strict effect over the transport
properties. Similar qualitative trends were suggested for turbulent
ow by Pak and Cho [3] and Xuan and Li [4].

0.1

6. Conclusions

x = 100.0 cm (6.00%) Re=940

x = 100.0 cm (2.76%) Re=1113
x = 97.0 cm (1.60%) Re=750
x = 97.0 cm (0.00% Water) Re=750
Parabolic profile

0.5

0.4

y/d [-]

0.1

0.1

0.000

0.3

0.5

1.5

Ux /Uo [-]

(a)
x = 100.0 cm. (6.00%) Re=940
x = 100.0 cm. (2.76%) Re=1113
x = 97.0 cm. (1.60%) Re=750
x = 97.0 cm. (0.00% Water) Re=750

0.5

0.4

y/d [-]

0.05

0.3

0.2

0.1

0.2

0.4

0.6

0.8

1.2

1.4

1.6

[-]

(b)
Fig. 6. (a) Dimensionless velocity proles near the outlet section; (b) dimensionless
temperature proles near the outlet section.

though the one corresponding to the higher concentration clearly

deviates from the others. Since the dimensionless temperature gradient at the wall is closely related to the heat transfer coefcient

A numerical investigation of the internal undeveloped laminar

ow heat transfer of a water/alumina nanouid has been carried
out based on a physical model previously proposed by Buongiorno
[2]. According to this model, the nanouid behaves as a mixture
where nanoparticles are submitted to brownian and thermophoretic diffusion effects. It has been determined that these diffusion
effects alter the nanoparticles concentration at the wall region,
affecting, as a result, the momentum and heat transfer in a level
that depends on the average nanoparticles concentration. Temperature and concentration effects over the transport properties of the
nanouid have been included in the numerical solution, an aspect
that has not generally been contemplated in previous numerical
solutions. The following is a summary of the conclusions that have
been drawn from the present investigation:
 The development of a concentration boundary layer along the
tube wall has been observed that extends from the entrance cross
section down to the exit one, progressively depleting the wall
region of nanoparticles. This boundary layer affects the velocity
and temperature proles with respect to those for pure uid.
 The temperature effect over the transport properties modies
the velocity and temperature proles with respect to the constant properties solutions. The deviation widens with the nanoparticles concentration, becoming signicant for concentrations
of 6%. Temperature and concentration effects over the transport
properties affect, as should be expected, the developed ow
which in this case does not attain the parabolic prole. In addition, the asymptotic Nusselt number deviates from the constant
property value, 4.36, in the present constant heat ux case.
 Present numerical solutions point toward an effective heat
transfer enhancement by nanouid with respect to the base
uid, the enhancement being more signicant at nanoparticles

432

P. Farias Alvario et al. / International Journal of Heat and Mass Transfer 59 (2013) 423432

volumetric concentrations higher than 2.76%. In that respect,

Curchill and Ozoe [27] correlation, with transport properties
averaged over the test section, seems to t reasonably well lower
concentration numerical results for the range of Graetz numbers
considered in the present study. Heat transfer enhancement is
clear at higher concentrations for which the numerical Nusselt
number deviate upward from this correlation.
 As a nal remark, a word on heat ux effects. The present
numerical investigation has focused on laminar ow solutions
for constant heat ux. The heat ux of each of the case studies
considered herein has been the one corresponding to the reference experimental investigation. Single phase ow solutions for
constant transport properties clearly are independent of the
wall heat ux. However, this might not be the case when nanoparticles diffusion effects are included in the model. Diffusion of
nanoparticles is affected by the temperature/concentration gradients in the wall region and consequently, by the heat ux.
Thus a comprehensive study of heat ux effects for both heating
and cooling processes is needed to expand the investigation of
nanouids heat transfer phenomena.

Acknowledgements
The authors gratefully acknowledge the support given to the
investigation reported herein by the Galician High Performance
Computing Center (CESGA) and the Xunta de Galicia, Spain,
through the Grant No. 10REM265008PR.
References
[1] S.K. Das, S.U.S. Choi, A review of heat transfer in nanouids, Adv. Heat Transfer
41 (2009) 81197.
[2] J. Buongiorno, Convective transport in nanouids, J. Heat Transfer 128 (2006)
240250.
[3] B.C. Pak, Y.I. Cho, Hydrodynamic and heat transfer study of dispersed uid with
submicron metallic oxide particles, Exp. Heat Transfer 11 (2) (1998) 151170.
[4] Y. Xuan, Q. Li, Investigation of convective heat transfer and ow features of
nanouids, J. Heat Transfer 125 (2003) 151155.
[5] W.C. Williams, J. Buongiorno, L.-W. Hu, Experimental investigation of
turbulent convective heat transfer and pressure loss of alumina/water and
zirconia/water nanoparticle colloids (nanouids) in horizontal tubes, J. Heat
Transfer 130 (2008) 042412-1042412-7.
[6] E. Djajadiwinata, H.A. Al-Ansary, K. Al-Dakkan, A. Bagabas, A. Al-Jariwi, M.F.
Zedan, Turbulent convective heat transfer and pressure drop of dilute CuO
(copper oxide)-water nanouid inside a circular tube, 3rd Micro and Nano
Flows Conference, Thessaloniki, Greece, 2224 August, 2011.

[7] D. Wen, Y. Ding, Experimental investigation into convective heat transfer of

nanouids at entrance region under laminar ow conditions, Int. J. Heat Mass
Transfer 47 (2004) 51815188.
[8] Y. Yang, Z.G. Zhang, E.A. Grulke, W.B. Anderson, G. Wu, Heat transfer properties
of nanoparticle-in-uid dispersions (nanouids) in laminar ow, Int. J. Heat
Mass Transfer 48 (2005) 11071116.
[9] S.Z. Heris, M.N. Esfahany, S. Etemad, Experimental investigation of convective
heat transfer of al_2o_3/water nanouid in circular tube, Int. J. Heat Fluid Flow
28 (2006) 203210.
[10] S.Z. Heris, S.G. Etemad, M.N. Esfahany, Convective heat transfer of a cu/water
nanouid owing through a circular tube, Exp. Heat Transfer 22 (4) (2009)
217227.
[11] U. Rea, T. McKrell, L.-W. Hu, J. Buongiorno, Laminar convective heat transfer
and viscous pressure loss of aluminawater and zirconiawater nanouids,
Int. J. Heat Mass Transfer 52 (2009) 20422048.
[12] K. Anoop, T. Sundararajan, S. Das, Effect of particle size on the convective heat
transfer in nanouid in the developing region, Int. J. Heat Mass Transfer 52
(2009) 21892195.
[13] R.K. Shah, Thermal entry length solutions for the circular tube and parallel
plates, in: Proceedings of 3rd National Heat and Mass Transfer Conference 1,
1975 (HMT1175).
[14] A. Behzadmehr, M. Saffar-Avval, N. Galanis, Prediction of turbulent forced
convection of a nanouid in a tube with uniform heat ux using a two phase
approach, Int. J. Heat Fluid Flow 28 (2007) 211219.
[15] Y. Xuan, Q. Li, Heat transfer enhancement of nanouids, Int. J. Heat Fluid Flow
21 (2000) 5864.
[16] S. Mirmasoumi, A. Behzadmehr, Effect of nanoparticles mean diameter on
mixed convection heat transfer of a nanouid in a horizontal tube, Int. J. Heat
Fluid Flow 29 (2008) 557566.
[17] S. Savithiri, A. Pattamatta, S.K. Das, Scaling anlysis for the investigation of slip
mechanisms in nanouids, Nanoscale Res. Lett. 6 (1) (2011) 6:471.
[18] H.F. Oztop, E. Abu-Nada, Numerical study of natural convection in partially
heated rectangular enclosures lled with nanouids, Int. J. Heat Fluid Flow 29
(2008) 13261336.
[19] M. Izadi, A. Behzadmehr, D. Jalali-Vahida, Numerical study of developing
laminar forced convection of a nanouid in an annulus, Int. J. Therm. Sci. 48
(2009) 21192129.
[20] E. Ebrahimnia-Bajestan, H. Niazmand, W. Duangthongsuk, S. Wongwises,
Numerical investigation of effective parameters in convective heat transfer of
nanouids owing under a laminar ow regime, Int. J. Heat Mass Transfer 54
(2011) 43764388.
[21] M. Kalteh, Abbas Abbassi, M. Saffar-Avval, J. Harting, Eulerianeulerian twophase numerical simulation of nanouid laminar forced convection in a
microchannel, Int. J. Heat Fluid Flow 32 (2011) 107116.
[22] OpenFOAM Foundation, OpenFoam users guide, December 2011. http://
www.openfoam.org/docs/.
[23] S. Klein, Engineering equation solver. <www.fChart.com>.
[24] R. Manohar, Analysis of laminar-ow heat transfer in the entrance region of
circular tubes, Int. J. Heat Mass Transfer 12 (1969) 1522.
[25] D. Ulrichson, R.A. Schmith, Laminar-ow heat transfer in the entrance region
of circular tubes, Int. J. Heat Mass Transfer 8 (1964) 253258.
[26] W.M. Kays, Numerical solutions for laminar-ow heat transfer in circular
tubes, J. Heat Transfer Trans. ASME (1955) 12651274.
[27] S. Churchill, H. Ozoe, Correlations for laminar forced convection with uniform
heating in ow over a plpate and in developing and fully developed ow in a
tube, J. Heat Transfer Trans. ASME (1973) 7884.