International Journal of Heat and Fluid Flow 32 (2011) 107–116

Contents lists available at ScienceDirect

International Journal of Heat and Fluid Flow

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

Eulerian–Eulerian two-phase numerical simulation of nanofluid laminar

forced convection in a microchannel
Mohammad Kalteh a,b, Abbas Abbassi a,⇑, Majid Saffar-Avval a, Jens Harting b,c
a
Department of Mechanical Engineering, Amirkabir University of Technology, Hafez Ave., P.O. Box 15916-34311, Tehran, Iran
b
Department of Applied Physics, Eindhoven University of Technology, P.O. Box 513, 5600 MB Eindhoven, The Netherlands
c
Institute for Computational Physics, University of Stuttgart, Pfaffenwaldring 27, 70569 Stuttgart, Germany

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

Article history: In this paper, laminar forced convection heat transfer of a copper–water nanofluid inside an isothermally
Received 4 March 2010 heated microchannel is studied numerically. An Eulerian two-fluid model is considered to simulate the
Received in revised form 1 August 2010 nanofluid flow inside the microchannel and the governing mass, momentum and energy equations for
Accepted 5 August 2010
both phases are solved using the finite volume method. For the first time, the detailed study of the rel-
ative velocity and temperature of the phases are presented and it has been observed that the relative
velocity and temperature between the phases is very small and negligible and the nanoparticle concen-
Keywords:
tration distribution is uniform. However, the two-phase modeling results show higher heat transfer
Nanofluid
Microchannel
enhancement in comparison to the homogeneous single-phase model. Also, the heat transfer enhance-
Two-phase ment increases with increase in Reynolds number and nanoparticle volume concentration as well as with
Laminar decrease in the nanoparticle diameter, while the pressure drop increases only slightly.
Heat transfer Ó 2010 Elsevier Inc. All rights reserved.

1. Introduction
Emerging developments in MEMS (Micro-Electro-Mechanical-
Systems) make it possible to fabricate very small scale devices.
On the other hand, these small scale devices can generate a high
amount of heat flux that should be taken away by a cooling system
to guarantee their appropriate performance. One possible way to
cool these devises can be the use of so-called nanofluids. A nano-
fluid is a suspension of nano-sized (10–100 nm) metallic and
non-metallic solid particles in a conventional cooling liquid such
as water, ethylene glycol, or oil. The term nanofluid for the first
time was used by Choi (1995) for such a suspension. After that,
many researchers focused on studying the thermophysical and also
heat and fluid flow properties of nanofluids. Most of the studies
concentrated on the modeling of the effective thermal conductivity
of nanofluids (e.g. Xuan et al., 2003; Koo and Kleinstreuer, 2004;
Feng et al., 2007). Recently, researchers have focused on the nano-
fluid heat and fluid flow behavior.
There are many experimental studies for nanofluids on macro
and micro-scales (e.g. Wen and Ding, 2004; Heris et al., 2006; Jung
et al., 2009; Wu et al., 2009). Wen and Ding (2004) investigated the
heat transfer of an Al2O3–water nanofluid in the entrance region of
a 4.5 mm diameter copper tube under the constant heat flux con-
ditions. Their measurements showed enhancement in heat transfer
⇑ Corresponding author. Tel.: +98 2164543425; fax: +98 2166419736.
E-mail address: abbassi@aut.ac.ir (A. Abbassi).
especially in the entrance region of the tube. They described this
behavior as the particle migration effect (non-uniform nanoparti-
cle volume concentration) that reduces the thermal boundary layer
thickness. For an annular tube with a 6 mm inner diameter copper
tube and a 32 mm outer diameter stainless steel tube, Heris et al.
(2006) studied CuO and alumina nanoparticles in water. They com-
pared the experimental results with homogeneous model results
(single-phase correlations with nanofluid effective properties)
and reported that the homogeneous modeling under-estimates
the heat transfer enhancement, especially in higher volume
concentrations.
Jung et al. (2009) did experiments for Al2O3–water nanofluids in
rectangular microchannels. The particle diameter in their experi-
ments was 170 nm. With only 1.8% of volume concentration they
reported a 32% increase of the heat transfer coefficient in compar-
ison to single distilled water. Also, experiments on nanofluid heat
transfer in trapezoidal silicon microchannels have been performed
by Wu et al. (2009). For channels with a hydraulic diameter of
194.5 lm and an Al2O3–water nanofluid, they reported an increase
in the Nusselt number with increasing particle concentration, Rey-
nolds and Prandtl numbers, while the pressure drop increased
slightly when compared to pure water.
For the theoretical study of the pressure-driven nanofluid heat
and fluid flow commonly homogenous (single-phase) and two-
phase models are used. In homogeneous modeling it is assumed
that the particles and the base fluid have the same temperature
and velocity and thus, the single-phase equations along with the

0142-727X/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.ijheatfluidflow.2010.08.001
108 M. Kalteh et al. / International Journal of Heat and Fluid Flow 32 (2011) 107–116

Nomenclature

A defined in Eq. (25) U, V non-dimensional velocity in the x- and y-directions

B defined in Eq. (24) respectively
Cp specific heat at constant pressure, J kg1 K1 t time, s
Cd drag coefficient T temperature, K
dp nanoparticle diameter, m x, y axial and vertical coordinates respectively, m
Dh hydraulic diameter, m x* non-dimensional axial length (xD1 1 1
h Re Pr )
Fcol particle–particle interaction force, Pa m1 X, Y non-dimensional axial and vertical coordinates respec-
Fd drag force, Pa m1 tively
Fvm virtual mass force, Pa m1
G particle–particle interaction modulus, Pa Greek symbols
h heat transfer coefficient based on the mean tempera- b friction coefficient, kg m3 s1
ture, W m2 K1 C defined in Eq. (23)
hv volumetric heat transfer coefficient, W m3 K1 h non-dimensional temperature
hp liquid-particle heat transfer coefficient, W m2 K1 l viscosity, Pa s
H channel height, m q density, kg m3
k thermal conductivity, W m1 K1 u volume concentration
L channel length, m x defined in Eq. (25)
Nu Nusselt number (hDh/kl)
Nu average Nusselt number Subscripts
Nup Particle Nusselt number b bulk
p pressure, Pa eff effective
P non-dimensional pressure i phase index (=l, p)
Pr liquid Prandtl number in inlet
q00 wall convective heat flux, W m2 l liquid phase
Re Reynolds number (qluinDh/ll) m mean
ReH Reynolds number (qluinH/ll)  nf nanofluid
ul ql j~
V l ~
V p jdp p particle phase
Rep particle Reynolds number ll
pw pure water
u, v velocity components in the x- and y-directions respec-
w wall
tively, m s1

appropriate effective thermophysical properties (thermal conduc-
tivity, viscosity, specific heat and density) for the nanofluid are
solved. In this method, the accuracy of the models used as effective
thermophysical properties is very important. Most of the theoreti-
cal studies in this field are based on the homogeneous approach
(e.g. Koo and Kleinstreuer, 2005; Li and Kleinstreuer, 2008; Santra
et al., 2009). In addition to the pressure-driven nanofluid flows, it is
possible to use electroosmotic transport for nanofluids especially
in the micro-scale. In this case, the thickness of the electrical dou-
ble layer and effective electrical conductivity of the nanofluid can
affect the nanofluid heat transfer behavior (Chakraborty and Pad-
hy, 2008; Chakraborty and Roy, 2008).
In despite of the homogeneous modeling, in the two-phase
modeling, the nanoparticle and the base fluid are considered as
two different phases with different velocities and temperatures.
In this method, the interactions between the phases are taken into
account in the governing equations. There are a few studies that
used two-phase approach to study nanofluids. Behzadmehr et al.
(2007) used a two-phase mixture model to study the turbulent
nanofluid convection inside a circular tube. Comparing with an
experimental study they reported that the two-phase results are
more precise than the homogeneous modeling results. However,
they considered thermal equilibrium conditions (the same temper-
ature) for the phases. Mirmasoumi and Behzadmehr (2008a) used
the same method as in Behzadmehr et al. (2007) to study the
mixed convection of the nanofluid in a tube. Also, Mirmasoumi
and Behzadmehr (2008b) and Akbarinia and Laur (2009) investi-
gated the nanoparticles size effect on the mixed convective heat
transfer of a nanofluid using the two-phase mixture method. In
both studies an increase in heat transfer with a decrease in the
nanoparticles size was reported. Bianco et al. (2009) modeled the
nanofluid flow and heat transfer inside a tube. They used both sin-
gle-phase and two-phase methods. For the two-phase method,
they implemented Lagrangian approach to model the particle mo-
tion. They reported a maximum difference of 11% between the sin-
gle and two-phase results. Kurowski et al. (2009) used three
different homogeneous, Eulerian–Lagrangian and mixture methods
to simulate nanofluid flow inside a minichannel. Their results
showed almost the same behavior for all the methods. Fard et al.
(2010) studied the nanofluid heat transfer inside a tube consider-
ing both the single and two-phase methods. For a 0.2% copper–
water nanofluid, they reported that the average relative error be-
tween the experimental data and single-phase model was 16%
while for the two-phase method it was 8%. On the other hand, Lotfi
et al. (2010) used homogeneous, two-phase Eulerian and mixture
models for nanofluid flow inside a tube. They reported that among
these methods, the two-phase mixture method is more precise
than the others.
According to the literature, there is a non-uniform nanoparticle
volume concentration distribution in the entrance region (Wen
and Ding, 2004) and the homogeneous model under-estimates
the observed heat transfer enhancement in the experiments (Heris
et al., 2006; Behzadmehr et al., 2007; Bianco et al., 2009; Fard et al.,
2010; Lotfi et al., 2010). Thus, the two-phase modeling can be an
alternative method. On the other hand, the existing studies for
the two-phase method do not consider the temperature difference
between the phases (Behzadmehr et al., 2007; Mirmasoumi and
Behzadmehr, 2008a,b; Akbarinia and Laur, 2009) or do not present
detailed results on the relative velocity and temperature between
the phases and the volume concentration distribution (Bianco
et al., 2009; Fard et al., 2010). The amount of the relative velocity
and temperature between the phases along with the nanoparticle
concentration distribution can provide an estimation of the accu-
racy of the assuming nanofluid as a homogeneous solution. On
 M. Kalteh et al. / International Journal of Heat and Fluid Flow 32 (2011) 107–116 109

the other hand, all the above mentioned two-phase studies are for Also, subscripts l and p stand for liquid and particle phases,
macro-sized circular tubes and to the best of the knowledge of the respectively. According to the volume concentration definition we
authors there is no such study for microchannels. So, this paper have
aims to study the nanofluid laminar forced convection in a deep
ul þ up ¼ 1 ð3Þ
rectangular (parallel plate) microchannel with isothermally heated
walls, using the Eulerian–Eulerian two-phase model. To do this,
mass, momentum and energy conservation equations for both 2.2. Momentum equations
phases are solved with the iterative numerical methods. The
two-phase results are compared with the single-phase results from Momentum equations in the x-direction are
the literature and then the nanoparticle size, nanoparticle concen-  
@ðul ql ul ul Þ @ðul ql v l ul Þ @p @ @u
tration and Reynolds number effects on the nanofluid heat transfer þ ¼ ul þ ul ll l
behavior are studied. Also, the relative velocity and temperature @x @y @x @x @x
 
between the phases and the particle volume concentration distri- @ @ul
þ u l ll þ ðF d Þx þ ðF v m Þx ð4Þ
bution in the field are investigated. To the best knowledge of the @y @y
authors, this is the first paper reporting the detailed two-phase
and
nanofluid modeling in a microchannel that considers different
velocity and temperatures for the phases.  
@ðup qp up up Þ @ðup qp v p up Þ @p @ @u
þ ¼ up þ u p lp p
@x @y @x @x @x
 
2. Governing equations @ @u
þ up lp p þ ðF col Þx
@y @y
The geometry of the present problem is shown in Fig. 1. It con-
 ðF d Þx  ðF v m Þx ð5Þ
sists of a parallel plate microchannel with height 200 lm and the
length L is 100 times larger than the height (L/H = 100). The origin Here, p, l, (Fd)x, (Fvm)x and (Fcol)x are the pressure, viscosity, drag,
of the Cartesian coordinate system is considered to be at the plate virtual mass (added mass) and particle–particle interaction forces
symmetry axis and only the top half of the channel is used for in the x-direction, respectively. Due to the very small size of the
numerical simulation. In this problem laminar nanofluid flow that nanoparticles, the lift force between the phases is neglected in
is a mixture of water and copper nanoparticles enters the channel the present study.
with a uniform velocity and temperature and exchanges heat with Momentum equations in the y-direction can be written as
the isothermal microchannel walls. follows:
Considering the laminar, steady state and two-dimensional  
Eulerian–Eulerian two-phase model for the nanofluid, the govern- @ðul ql ul v l Þ @ðul ql v l v l Þ @p @ @v
þ ¼ ul þ ul ll l
ing mass, momentum and energy equations for the particle and @x @y @y @x @x
 
base liquid phases can be written as follows (Fluent user’s guide, @ @v l
2006; Hao and Tao, 2004). þ u l ll þ ðF d Þy þ ðF v m Þy ð6Þ
@y @y
 
2.1. Continuity equations @ðup qp up v p Þ @ðup qp v p v p Þ @p @ @v
þ ¼ up þ u p lp p
@x @y @y @x @x
 
@ðul ql ul Þ @ðul ql v l Þ @ @v
þ ¼0 ð1Þ þ up lp p þ ðF col Þy
@x @y @y @y
 ðF d Þy  ðF v m Þy ð7Þ
@ðup qp up Þ @ðup qp v p Þ
þ ¼0 ð2Þ where (Fd)y, (Fvm)y and (Fcol)y are the drag, virtual mass (added mass)
@x @y
and particle–particle interaction forces in the y-direction, respec-
where x, y, u, v, q and u are axial and vertical direction, axial and tively. Due to the small size of the channel, the gravitational force
vertical velocity, density and volume concentration, respectively. is neglected in the present study.

Fig. 1. Geometry of the isothermally-heated parallel plate microchannel with length L and height H. Nanofluid enters the channel with uniform velocity and temperature and
leaves the channel while it is fully developed.
110 M. Kalteh et al. / International Journal of Heat and Fluid Flow 32 (2011) 107–116

The drag force between the phases is calculated as 6ð1  ul Þ

  hv ¼ hp ð17Þ
dp
F d ¼ b ~
Vl  ~
Vp ð8Þ
where hp is the fluid–particle heat transfer coefficient that should
where ~V is the velocity vector. be calculated from empirical correlations. In the present study the
The friction coefficient b is calculated according to the particle fluid–particle heat transfer coefficient is calculated based on the
volume concentration range. For very dilute two-phase flows with Wakao and Kaguei (1982):
particle diameter dp, the friction coefficient is (Syamlal and Gidas-
hp dp 1
pow, 1985) Nup ¼ ¼ 2 þ 1:1Re0:6
p Pr
3 ð18Þ
k1
3 u ð1  ul Þ ~ ~ 2:65
b ¼ Cd l jV l  V p jul ð9Þ Here Pr is the base liquid Prandtl number.
4 dp
The effective thermal conductivities for liquid and particle
Eq. (9) is valid for two-phase flows with ul > 0.8 and Cd is the phases are estimated as (Kuipers et al., 1992)
drag coefficient and its magnitude depends on the particle Rey- kb;l
nolds number: keff ;1 ¼ ; ð19Þ
(
ul
24
Rep
ð1 þ 0:15Re0:697
p Þ Rep < 1000
Cd ¼ ð10Þ kb;p
0:44 Rep P 1000 keff ;p ¼ ð20Þ
up
where
where
u q j~
Vl  ~
V p jdp  qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Rep ¼ l l ð11Þ
ll kb;l ¼ 1 ð1  ul Þ kl ð21Þ

The virtual mass force is proportional to the relative accelera- qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

tion of the two phases and is written as (Drew and Lahey, 1993) kb;p ¼ ð1  ul ÞðxA þ ½1  xCÞkl ð22Þ
D
F v m ¼ 0:5up ql ð~Vl  ~
V pÞ ð12Þ and
Dt ( )
 
In Eq. (12) D is the material derivative, such that, for the steady 2 BðA  1Þ A ðB  1Þ B þ 1
C¼    ln     ð23Þ
state case, the convective terms of the material derivative are 1  AB A 1  B 2 B 1  AB 2
A
retained.
The particle–particle interaction force is calculated as (Bouillard with
et al., 1989)  10
½1  ul  9
~u B ¼ 1:25 ð24Þ
F col ¼ Gðul Þr l ð13Þ ul
Here, G is the particle–particle interaction modulus and it is cal- For spherical particles one can use
culated as
kp
A¼ and x ¼ 7:26  103 ð25Þ
G ¼ 1:0 expð600½ul  0:376Þ ð14Þ kl
Eqs. (18)–(25) are not correlations developed for nano-sized
Eqs. (8)–(14) are not correlations developed for nano-sized
particles. But, due to the lack of such correlations for nano-sized
particles. But, since there are no such correlations for nano-sized
particles, they are used in the present study.
particles, it is assumed that it is reasonable to use them for nano-
particles. Also, the importance of drag, virtual mass and particle–
particle interaction forces in nanofluid will be discussed in
2.4. Nusselt number definition
Section 6.1.
The Nusselt number is defined based on the temperature differ-
2.3. Energy equations
ence between the microchannel wall and the nanofluid mean
(bulk) temperature:
Considering the base fluid and the particle phase as incompress-
ible fluids, and neglecting the viscous dissipation and radiation, the ðhDh Þ
Nu ¼ ¼ q00 Dh =kl ðT w  T m Þ ð26Þ
energy equation is written as kl

@ @ where h, Dh, q00 are the convective heat transfer coefficient, micro-
ðu q ul cpl T l Þ þ ðul ql ml cpl T l Þ channel hydraulic diameter and the wall convective heat transfer
@x l l @y
    flux, respectively. Also, subscripts w and m stand for wall and mean,
@ @T @ @T
¼ ul keff ;l l þ ul keff ;l l  hv ðT l  T p Þ ð15Þ respectively. The mean temperature for a two-phase fluid can be
@x @x @y @y calculated from (Boulet and Moissette, 2002)
Pp RR
@ @ ð qi ui cpi T i dAÞ
ðu q up cpp T p Þ þ ðup qp mp cpp T p Þ T m ¼ Pi¼1
p RR ð27Þ
@x p p @y i¼1 ð qi ui c pi dAÞ
   
@ @T @ @T
¼ up keff ;p p þ up keff ;p p þ hv ðT l  T p Þ ð16Þ where the integration is performed on the channel cross section. For
@x @x @y @y two-phase flow according to Eqs. (15) and (16), wall convective
heat transfer flux can be calculated as
where cp, T, keff and hv are the heat capacity at constant pressure,
temperature, effective thermal conductivity and volumetric inter- @T l @T p
q00 ¼ ul keff ;l þ u k ð28Þ
@y w @y w
p eff ;p
phase heat transfer coefficient, respectively. For mono-dispersed
spherical particles hv can be calculated from
 M. Kalteh et al. / International Journal of Heat and Fluid Flow 32 (2011) 107–116 111

According to the local Nusselt number, the average Nusselt 4. Grid-independence study
number is
To check for the independency of the results from the number
Z L of grid points used, a grid independency study is done by consider-
1
Nu ¼ Nu dx ð29Þ ing the amount of the calculated average Nusselt number. To do
L 0
this, different numbers of grid points are used in the x- and y-direc-
tions. The results are shown in Table 1, where the flow Reynolds
2.5. Non-dimensionalization number is 1500. According to this study, the number of the grid
points in x- and y-directions are considered 500 and 30 respec-
Before the solution, all the governing equations are converted to tively in the present study.
non-dimensional form, using the following non-dimensional
parameters: 5. Code validation

x y ui vi p  pin Due to the lack of experimental data for nanofluid flow in a par-
X¼ ; Y¼ ; Ui ¼ ; Vi ¼ ; P¼ ; allel plate microchannel, the calculated average Nusselt numbers
Dh Dh uin uin ql u2in
for the special case of pure water flow (up = 0.0) at different Rey-
T i  T in
hi ¼ ð30Þ nolds numbers are compared to corresponding available data in
T w  T in
the literature in order to check the accuracy of the written com-
puter code. For a single-phase fluid flow in an isothermally-heated
where i = l, p stands for the liquid and the particle phases. parallel plate channel, the average Nusselt number is calculated as
(Ebadian and Dong, 1998)
2.6. Boundary conditions 0:024x1:14
Nu ¼ 7:55 þ ð31Þ
1 þ 0:0358Pr0:17 x0:64
Both phases enter the channel at the inlet with the same uni-
form axial velocity that is specified according to the flow Reynolds where x ¼ xD1 1
h Re Pr
1
is the non-dimensional axial length.
number. At the channel outlet, outflow velocity boundary condi- Table 2 shows the calculated average Nusselt number in the
tion is assumed for both phases. For liquid molecules the mean free numerical simulation, the corresponding results from Eq. (31)
path is 0.1–1 nm, so that the channel hydraulic diameter should be and the difference between the two results for a wide range of Rey-
smaller than 1 lm to be in the slip region (Morini, 2005). There- nolds number. The results show that the agreement between the
fore, for both of the phases the no-slip boundary condition at the present numerical solution results and the existing solution from
walls is appropriate for the present study. Eq. (31) is very good especially for smaller Reynolds numbers.
For thermal boundary conditions, it is assumed that the nano- According to Table 2 for lower Reynolds numbers the deviation is
fluid enters the channel with 293 K and the isothermal walls have less than 1% and it is less than 4% for Re = 2000. After checking
a temperature of 303 K. For the channel outlet, the outflow bound- for the accuracy of the computer code, in the following section heat
ary condition is considered for both phases. transfer and pressure drop results for different Reynolds numbers,
nanoparticle diameters and concentrations are presented.

3. Numerical method
6. Results and discussion
The non-dimensional form of the mass, momentum and energy
conservation equations for liquid and particle phases along with Since in the present study the particle phase is considered as a
the interphase correlations and boundary conditions are discret- continuum, its viscosity lp has to be obtained. In fact, due to the
ized using the finite volume method on the upper half of the
channel (Patankar, 1980; Versteeg and Malalasekera, 1995). A Table 1
non-uniform grid is employed in the computational domain. The Grid-independency study results for pure water flow (up = 0.0) and Re = 1500.
grids are finer close to the wall and also in the channel entrance No. of grid points in x- No. of grid points in y- Average Nusselt number
region using the cosine weighting function for control volume direction direction at the wall
length and height. The power-law scheme (Patankar, 1980; 150 10 12.252
Versteeg and Malalasekera, 1995) is used for the convection– 300 20 12.155
diffusion term discretization. The set of the discretized equations 600 40 12.122
250 15 12.182
is solved iteratively using the line by line method (Patankar, 1980;
500 30 12.130
Versteeg and Malalasekera, 1995) and for solving the pressure– 1000 60 12.116
velocity coupling the well-known SIMPLE (Semi-Implicit Method
for Pressure-Linked Equations) algorithm of Patankar (1980) is
used. To use the SIMPLE algorithm, the pressure-correction equa-
Table 2
tion is derived by combining the mass conservation equations for Comparison between average Nusselt number results from numerical simulations
particle and liquid phases. Also, for accelerating the convergence and Ebadian and Dong (1998) results for pure water flow at different Reynolds
of the SIMPLE algorithm, the under-relaxation for velocity and numbers.
pressure is used. For convergence criteria, the sum of the scaled Re Present study Ebadian and Dong (1998) Deviation (%)
absolute residual for every parameter (mass, velocity and temper-
20 7.66 7.621 0.51
ature) in all control volumes is calculated. The sum of the scaled 50 7.756 7.739 0.22
absolute residuals for every parameter is restricted to be smaller 100 7.924 7.934 0.13
than 106. For convergence criteria up to 109, the calculated 500 9.299 9.288 0.12
average Nusselt number remains unchanged up to the third digit 1000 10.843 10.638 1.93
1500 12.13 11.775 3.01
of the decimal. Thus, the convergence criterion is selected to be
2000 13.238 12.775 3.62
106 in the present study.
112 M. Kalteh et al. / International Journal of Heat and Fluid Flow 32 (2011) 107–116

Table 3 dp = 100 nm and different term conditions. According to Table 4

Sensitivity study of the average Nusselt num- virtual mass and particle–particle interaction forces do not have
ber on the particle phase viscosity for Re = 100,
up = 0.01 and dp = 100 nm. any effect on the average Nusselt number. But, the drag force af-
fects the average Nusselt number slightly. Also, the drag force ef-
Particle viscosity Average Nusselt fect on the average Nusselt number increases with an increase in
(Pa s) number
the nanoparticle volume concentration. So, it is possible to neglect
0.01 11.674 the virtual mass and particle–particle interaction forces for nano-
0.005 11.675
0.002 11.658
fluid in the mathematical modeling.
0.00138 11.659
0.001 11.648 6.2. Comparison with homogeneous model results
0.0008 11.66
0.0002 11.655
0.00005 11.651 In this section, the two-phase modeling results are compared to
0.00001 11.663 the homogeneous modeling results of Santra et al. (2009). They re-
ported average Nusselt numbers in different conditions for 100 nm
copper–water nanofluid flowing inside a parallel plate channel.
lack of experimental data, the solid viscosity for a liquid–solid two- They used the channel height as a characteristic length scale to de-
phase mixture is not available. So, for the first degree of approxi- fine Reynolds number (i.e., ReH) and Nusselt numbers. Also, they
mation the following method is adopted in the present study: used the temperature difference between the wall and the inlet
The corresponding pressure drop and the average Nusselt number fluid to define the heat transfer coefficient (in despite of Eq.
of a highly dilute nanofluid with volumetric concentration of (26)). So, to be able to compare the present two-phase results with
0.00001 (which is quite close to pure water), is compared with that the homogeneous modeling results of Santra et al. (2009), all the
of pure water, for Re = 1500. Using the trial and error method, the present results in this section are based to their definitions. Fig. 2
value of viscosity in the solid phase of the highly dilute nanofluid is depicts the two-phase modeling results in comparison with the
changed up to a point where the pressure drop and the average homogeneous modeling results for ReH = 100 and 100 nm particles.
Nusselt number of the highly dilute nanofluid and the pure water It can be seen from Fig. 2 that the two-phase modeling results
are matched. In doing so, the viscosity of the solid phase reaches a show higher heat transfer enhancement in comparison to homoge-
value of 1.38  103 Pa s. Under such conditions, the difference be- neous modeling results. Also, the heat transfer enhancement in-
tween the pressure and the average Nusselt number of the highly creases non-linearly with increase in nanoparticle volume
dilute nanofluid and that of pure water are 0% and 1.1%, respec-
tively. In order to investigate the effect of Reynolds number of 40
the flow, the same method of comparison between the pressure ReH=100, dp=100 nm
drop and the average Nusselt number of the highly dilute nanofluid Present Study
35
Average Nusselt number enhancement (%)

and the pure water is made, under the condition of very low Rey- Homogeneous Model of Santra et al. (2009)
nolds number (equal to 100). Under these circumstances, the dif-
ference between the pressure drop and the average Nusselt 30
number correspond to 0.04% and 1.4%, respectively. It can be con-
cluded that the viscosity of the solid phase is independent of the 25
Reynolds number. Also, Table 3 shows the sensitivity study of aver-
age Nusselt number on the solid viscosity magnitude for Re = 100, 20
up = 0.01 and dp = 100 nm. As can be seen from Table 3 changing
the solid viscosity, the amount of the average Nusselt number
15
changes slightly. i.e., when the solid viscosity changes three orders
of magnitude (from 0.01 to 0.00001) the average Nusselt number
remains unchanged up to the first digit of the decimal. So, it can 10
be concluded that the results are not very sensitive to the solid vis-
cosity and it is not critical to know the exact magnitude for particle 5
viscosity. With these discussions, the particle viscosity is consid-
ered to be 1.38  103 Pa s in the present study. 0
0 0.01 0.02 0.03 0.04 0.05
6.1. Importance of the terms in the momentum equation Nanoparticle volume concentration

Fig. 2. Comparison of percentage enhancement in average Nusselt number with

There are three interphase forces in the momentum equation respect to pure water flow for homogeneous and two-phase models at ReH = 100
(i.e. drag, virtual mass and particle–particle interaction forces). and different nanoparticle volume concentrations. Two-phase results show higher
Table 4 shows the average Nusselt number for Re = 300, heat transfer enhancement in comparison to homogeneous results.

Table 4
Effect of the different terms in the momentum equation on the average Nusselt number for Re = 300, dp = 100 nm and different volume concentrations.

Volume concentration Considering all Neglecting the drag Neglecting the particle–particle interaction Neglecting the virtual mass
(%) terms term term term
1 12.343 12.385 12.343 12.343
2 14.068 14.151 14.068 14.068
3 15.507 15.637 15.507 15.507
4 16.818 16.998 16.818 16.818
5 18.051 18.292 18.051 18.051
 M. Kalteh et al. / International Journal of Heat and Fluid Flow 32 (2011) 107–116 113

Table 5
Percentage increase in average Nusselt number with respect to pure water flow for homogeneous model simulations of Santra et al. (2009) and two-phase models at different
Reynolds numbers and nanoparticle volume concentrations. Two-phase results show higher heat transfer enhancement in comparison to the homogeneous modeling results.

up (%) ReH = 500 ReH = 1000 ReH = 1500

Present study Santra et al. (2009) Present study Santra et al. (2009) Present study Santra et al. (2009)
0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.5 20.75917 3.21628 20.90864 3.26732 20.94575 3.30307
1.0 29.31589 6.40564 29.5805 6.51014 29.62227 6.5825
1.5 35.96397 9.56952 36.36213 9.72955 36.42431 9.83894
2.0 41.64701 12.70968 42.21522 12.9274 42.30332 13.07415
2.5 46.75102 15.82767 47.52114 16.10541 47.62851 16.28976
3.0 51.49046 18.92487 52.42912 19.26517 52.58449 19.48727
3.5 55.84388 22.00248 57.05521 22.40813 57.25646 22.66808
4.0 60.02573 25.06157 61.49892 25.53565 61.70122 25.83347
4.5 64.07892 28.10305 65.7105 28.64899 66.01818 28.98467
5.0 67.76753 31.12774 69.88891 31.74931 70.19313 32.12279

concentration for two-phase modeling results, while it is almost
linear for homogeneous modeling results. For instance, for 0.05
nanoparticle volume concentration, two-phase modeling shows
32.6% heat transfer enhancement, while it is 19% for homogeneous
modeling results. For quantitative comparisons between the two-
phase and homogeneous modeling results for different Reynolds
numbers the data are collected in Table 5. The same behavior as
Fig. 2 can be seen for other Reynolds numbers in Table 5. This
observation also has been reported by previous two-phase model-
ing studies (Behzadmehr et al., 2007; Bianco et al., 2009; Fard et al.,
2010; Lotfi et al., 2010). One reason for this behavior can be related
to the accuracy of the nanofluid thermophysical property models
that are used in the homogeneous modeling.
6.3. Effect of the nanoparticles on the velocity and temperature field
Two-phase simulation results show that the relative velocity
and temperature between the phases is very small and negligible.
For instance, for Re = 100 and 0.01 nanoparticle volume concentra-
tion, the relative velocity in almost the full computation field is of
the order of 106 or smaller and in a very small region close to the
entrance it is of the order of 104. For the temperature difference,
in almost the entire flow field it is of order 104 and in the very
small region close to the channel entrance it is of order 102. Also,
the results show that the nanoparticle volume concentration dis-
tribution is uniform in the entire flow field. Such a result for nano-
particle distribution also has been reported by Akbarinia and Laur
(2009) for a curved tube. Thus, considering the nanofluid as a
homogeneous solution seems to be reasonable. Fig. 3 shows the ef-
fect of nanoparticles on the velocity field for the Re = 1500 and
up = 0.01 case. In Fig. 3 the bottom contour lines show the velocity
field for pure water flow in the computational domain, while the
middle and the top ones show the velocity field for base liquid
and nanoparticle phases in the nanofluid. From Fig. 3 it can be seen
that the effect of nanoparticle on the velocity profile is very small
and it slightly increases the hydrodynamic entrance length. This
behavior is expectable due to very small size and concentration
of the nanoparticles. This is the reason for a small increase in pres-
sure drop for a nanofluid in comparison to the pure water. Accord-
ing to the velocity contour lines for base liquid and nanoparticle
phases in the nanofluid in Fig. 3, it is clear that the relative velocity
between the phases is very small and negligible.
Fig. 4 shows the effect of nanoparticles on the temperature field
for the Re = 1500 and up = 0.01 case. In this figure the temperature
contour line for pure water (the bottom plot) is compared to tem-
perature contour lines for base liquid (middle plot) and nanoparti-
cle (top plot) phases in the nanofluid. According to Fig. 4
nanoparticles increase the thermal boundary layer development
considerably. This can be interpreted as increase in the thermal
conductivity of the fluid due to the presence of the nanoparticles
and as a result an increase in the heat transfer rate between
the nanofluid and the microchannel wall. Also, the very small

nanoparticle phase
0.25 0 0.5
.9
Y

1.3
1.2

0 .4
0 10 20 30 40 50
X
base liquid phase
0.25
0.5
0.9
Y

1. 2
1.3
0
0 10 20 30 40 50
X
pure water
0.25 0.5
0.9
Y

1.2
1.2

.3

0
0 10 20 30 40 50
X

Fig. 3. Velocity contour lines for pure water (bottom plot) and nanofluid (base liquid phase in middle plot and nanoparticle phase in top plot) at Re = 1500, up = 0.01.
Nanoparticles affect the velocity profile slightly and the relative velocity between the nanoparticles and base liquid is very small and negligible.
114 M. Kalteh et al. / International Journal of Heat and Fluid Flow 32 (2011) 107–116

particle phase
0.25 0.3
0.5
0.1

Y
0.01

0
0 10 20 30 40 50
X

base liquid phase

0.25 0.3
1
0. 0
1
Y

0.1
0.01
0
0 10 20 30 40 50
X

pure water
0.25
0.5
0.1 0.3
Y

0.01

0
0 10 20 30 40 50
X

Fig. 4. Temperature contour lines for pure water (bottom plot) and nanofluid (base liquid phase in middle plot and nanoparticle phase in top plot) at Re = 1500, up = 0.01.
Nanoparticles cause the temperature boundary layer to develop faster while the relative temperature between the nanoparticle and the base liquid phases is very small and
negligible.

difference between the temperature profiles for base fluid and the 0%
24 1%
nanoparticle phases is clear in Fig. 4.
2%
3%
22 4%
6.4. Effect of particle volume concentration on heat and fluid flow 5%

20
Fig. 5 depicts the effect of varying the particle volume concen-
Average Nusselt number

tration on the pressure drop for 100 nm particles. As it can be seen,

18
the pressure drop increases slightly with increase of the nanopar-
ticle volume concentration for all Reynolds numbers. This corre-
sponds to the results reported by previous studies (e.g. Jung 16
et al., 2009; Wu et al., 2009). This behavior is expectable since
the nanoparticles have a very small effect on the velocity field as 14
presented in the previous section. For instance, for Re = 1000 and
up = 0.01, the percentage increase in pressure drop in comparison 12

30 10
25
0% 8
20 200 400 600 800 1000 1200 1400 1600
2%
Re
3%
15 5%
Non-dimensional pressure drop

Fig. 6. Average Nusselt number ðNuÞ versus Reynolds number for different
nanoparticle volume concentrations (dp = 100 nm). The average Nusselt number
increases with an increase of the Reynolds number and the nanoparticle volume
10
concentration.

to pure water is 1.99%. This is one of the advantages of using nano-

5
particles rather than millimeter or micrometer sized particles.
Fig. 6 shows the particle volume concentration and flow Rey-
nolds number effect on the average Nusselt number for 100 nm
particles. It can be seen that the average Nusselt number increases
with an increase in the nanoparticle volume concentration as well
as an increase in the flow Reynolds number. By comparing Figs. 5
and 6 it is obvious that the increase in the Nusselt number is much
400 800 1200 1600 higher than the increase in the pressure drop. So, this shows that
Re nanofluids can be used as efficient working fluids in cooling sys-
Fig. 5. Non-dimensional pressure drop versus Reynolds number for different
tems. According to Fig. 6, increasing the Reynolds number and
nanoparticle volume concentrations (dp = 100 nm). The pressure drop increases the nanoparticle volume concentration increases the heat transfer
slightly with an increase of the nanoparticle volume concentration. amount due to higher convection effects and higher nanoparticle
 M. Kalteh et al. / International Journal of Heat and Fluid Flow 32 (2011) 107–116 115

2.2 specified amount of nanoparticles (constant nanoparticle volume

1%
2% concentration) on enhancing the heat transfer ratio is larger for
2.1 3% the lower Reynolds numbers. For instance, at up = 0.01, the average
4% Nusselt number ratio is 1.45 and 1.34 for Re = 200 and Re = 1600,
2 5% respectively.
1.9
6.5. Effect of the nanoparticle diameter on heat transfer
1.8
Nunf / Nupw

Fig. 8 depicts the nanoparticle size effect on the amount of the

1.7
1.6
1.5
1.4
1.3
1.2
200 400 600 800
Re
Fig. 7. Nanofluid average Nusselt number normalized by the pure water average
Nusselt number versus Reynolds number for different particle concentrations and
dp = 100 nm. The normalized average Nusselt number decreases slightly with an
increase in the Reynolds number.
participation in increasing the effective thermal conductivity of the
nanofluids, respectively. This behavior is in agreement with the
experimental study of Wu et al. (2009) and the numerical study
of Li and Kleinstreuer (2008) for microchannels.
Fig. 7 shows the nanofluid average Nusselt number normalized
by the corresponding Nusselt number for pure water flow. From
the figure it can be seen that for every Reynolds number, an in-
crease in nanoparticle volume concentration increases the average
Nusselt number ratio. This behavior is expected since the higher
nanoparticle concentration increases the thermal conductivity of
the nanofluid and thus causes an increase in the heat transfer rate.
Also, Fig. 7 shows that for the same nanoparticle volume concen-
tration, with an increase in Reynolds number, the average Nusselt
number ratio decreases slightly. In other words, the effect of a
heat transfer enhancement in comparison to pure water flow, at
different volume concentrations and for Re = 500. It can be seen
that an increase of the volume concentration and a decrease of
the nanoparticle diameter increases the heat transfer enhance-
ment. According to Fig. 8, the increase in heat transfer enhance-
ment with a decrease in the nanoparticle size is not very
pronounced especially for lower volume concentrations, since the
particle sizes considered here are at the same order of magnitude.
For instance, for a nanofluid with 0.01 nanoparticle volume con-
centration, for 100 nm and 30 nm particle sizes, the enhancement
1000 1200 1400 1600
in heat transfer is 40.43% and 42.47%, respectively. The increase in
heat transfer rate with a decrease in particle size was also reported
by other researchers (Heris et al., 2007; Mirmasoumi and Behzad-
mehr, 2008b; Anoop et al., 2009; Akbarinia and Laur, 2009). Anoop
et al. (2009) reported an increase in heat transfer enhancement
with a decrease in particle size for experimental study of the alu-
mina-water nanofluid flow in the developing region of a tube with
nanoparticle diameters 45 nm and 150 nm. But, He et al. (2007)
experimentally studied the TiO2 nanoparticle size effect on the
heat transfer rate inside a tube. They performed the experiments
for 95, 145 and 210 nm particles with 0.006 particle volume con-
centrations. Their heat transfer rate results did not show sensitivity
to the nanoparticle size. They suggested a particle migration effect
causing a depletion layer that is created due to migration of the lar-
ger particles to the channel center to explain this behavior. How-
ever, the continuum method used in the present study is not
able to resolve a very small depletion layer of the order of the par-
ticle diameter close to the wall and the results show a uniform
nanoparticle volume concentration distribution in the computa-
tional field.

120 7. Conclusions

Pressure drop and heat transfer due to copper–water nanofluid

110 100 nm
Average Nusselt number enhancement (%)

flow inside an isothermally-heated parallel plate microchannel is

70 nm
50 nm studied numerically for a wide range of Reynolds numbers, nano-
100 30 nm particle volume concentrations and nanoparticle diameters. To do
this, the nanofluid flow is modeled using the Eulerian two-fluid
90 model. In this method, the difference between the velocity and
temperature for liquid and nanoparticle phases are considered
80
and the governing equations for both phases are solved numeri-
cally using the finite volume method. It is observed that the rela-
tive velocity and temperature for base liquid and nanoparticle
70 phases are very small and negligible. Thus, the liquid and the nano-
particles have almost the same velocity and temperature. Also, the
60 nanoparticle volume concentration distribution is uniform in the
computational domain. Therefore we conclude that considering
50 the nanofluid as a homogeneous solution is reasonable.
The heat transfer enhancement results for two-phase modeling
show higher magnitudes in comparison to the homogeneous mod-
40
0.01 0.02 0.03 0.04 0.05 eling results. Such an observation also reported by other researches
Nanoparticle volume concentration (Behzadmehr et al., 2007; Bianco et al., 2009; Fard et al., 2010; Lotfi
et al., 2010). Thus, under-estimation of the heat transfer enhance-
Fig. 8. Average Nusselt number enhancement in comparison to pure water flow
versus nanoparticle volume concentration for different nanoparticle diameters and
ment by homogeneous modeling seems to be related to the insuf-
Re = 500. The effect of the size of the nanoparticles is more pronounced for higher ficient accuracy of the nanofluid thermophysical property models
nanoparticle volume concentrations. that are used in the homogeneous modeling. Also, the pressure
116 M. Kalteh et al. / International Journal of Heat and Fluid Flow 32 (2011) 107–116
drop for nanofluids is slightly higher than the pressure drop for the
pure water flow, while the average Nusselt number increases with
increase in the Reynolds number and particle volume concentra-
tion. For the same nanoparticle volume concentration, the average
Nusselt number ratio is higher for lower Reynolds numbers. Keep-
ing all the other parameters constant, heat transfer enhancement is
higher for the nanofluids with smaller nanoparticle sizes. However,
this effect is not very pronounced at low nanoparticle volume con-
centrations. Also, the most important advantage of this method in
comparison to homogenous modeling is that there is no need for
effective thermophysical models for the nanofluid.
Acknowledgements
The first author would like to acknowledge the Ministry of Sci-
ence, Research and Technology of the Islamic Republic of Iran for
financial support to perform the research at the Eindhoven Univer-
sity of Technology. The authors also like to thank Anton Darhuber
for fruitful discussions.
References
Akbarinia, A., Laur, R., 2009. Investigating the diameter of solid particles effects on a
laminar nanofluid flow in a curved tube using a two phase approach. Int. J. Heat
Fluid Flow 30, 706–714.
Anoop, K.B., Sundararajan, T., Das, S.K., 2009. Effect of particle size on the convective
heat transfer in nanofluid in the developing region. Int. J. Heat Mass Transfer 52,
2189–2195.
Behzadmehr, A., Saffar-Avval, M., Galanis, N., 2007. Prediction of turbulent forced
convection of a nanofluid in a tube with uniform heat flux using a two phase
approach. Int. J. Heat Fluid Flow 28, 211–219.
Bianco, V., Chiacchio, F., Manca, O., Nardini, S., 2009. Numerical investigation of
nanofluids forced convection in circular tubes. Appl. Thermal Eng. 29, 3632–
3642.
Bouillard, J.X., Lyczkowski, R.W., Gidaspow, D., 1989. Porosity distributions in a
fluidized bed with an immersed obstacle. AIChE J. 35, 908–922.
Boulet, P., Moissette, S., 2002. Influence of the particle-turbulence modulation
modeling in the simulation of a non-isothermal gas–solid flow. Int. J. Heat Mass
Transfer 45, 4201–4216.
Chakraborty, S., Padhy, S., 2008. Anomalous electrical conductivity of nanoscale
colloidal suspensions. ACS NANO 2, 2029–2036.
Chakraborty, S., Roy, S., 2008. Thermally developing electroosmotic transport of
nanofluids in microchannels. Microfluid Nanofluid 4, 501–511.
Choi, S.U.S., 1995. Enhancing thermal conductivity of fluids with nanoparticles. In:
Siginer, D.A., Wang, H.P. (Eds.), Developments and Applications of Non-
Newtonian Flows, FED 231/MD 66. ASME, New York, pp. 99–105.
Drew, D.A., Lahey, R.T., 1993. Analytical modeling of multiphase flow. In: Roco, M.C.
(Ed.), Particulate Two-Phase Flow. Butterworth–Heinemann, Boston, pp. 509–
566.
Ebadian, M.A., Dong, Z.F., 1998. Forced convection, internal flow in ducts. In:
Rohsenow, W.M., Hartnett, J.P., Cho, Y.I. (Eds.), Handbook of Heat Transfer.
McGraw-Hill, New York, pp. 5.1–5.137.
Fard, M.H., Esfahany, M.N., Talaie, M.R., 2010. Numerical study of convective heat
transfer of nanofluids in a circular tube two-phase model versus single-phase
model. Int. Commun. Heat Mass Transfer 37, 91–97.
Feng, Y., Yu, B., Xu, P., Zou, M., 2007. The effective thermal conductivity of
nanofluids based on the nanolayer and the aggregation of nanoparticles. J. Phys.
D: Appl. Phys. 40, 3164–3171.
Fluent 6.3 User’s Guide. Fluent Inc., 2006.
Hao, Y.L., Tao, Y.X., 2004. A numerical model for phase-change suspension flow in
microchannels. Numer. Heat Transfer, Part A 46, 55–77.
He, Y., Jin, Y., Chen, H., Ding, Y., Cang, D., Lu, H., 2007. Heat transfer and flow
behavior of aqueous suspensions of TiO2 nanoparticles (nanofluids) flowing
upward through a vertical pipe. Int. J. Heat Mass Transfer 50, 2272–2281.
Heris, S.Z., Etemad, S.Gh., Esfahany, M.N., 2006. Experimental investigation of oxide
nanofluids laminar flow convective heat transfer. Int. Commun. Heat Mass
Transfer 33, 529–535.
Heris, Z.S., Esfahany, M.N., Etemad, G., 2007. Numerical investigation of nanofluid
laminar convective heat transfer through a circular tube. Numer. Heat Transfer,
Part A 52, 1043–1058.
Jung, J.-Y., Oh, H.-S., Kwak, H.-Y., 2009. Forced convective heat transfer of nanofluids
in microchannels. Int. J. Heat Mass Transfer 52, 466–472.
Koo, J., Kleinstreuer, C., 2004. A new thermal conductivity model for nanofluids. J.
Nanoparticles Res. 6, 577–588.
Koo, J., Kleinstreuer, C., 2005. Laminar nanofluid flow in microheat-sinks. Int. J. Heat
Mass Transfer 48, 2652–2661.
Kuipers, J.A.M., Prins, W., van Swaaij, W.P.M., 1992. Numerical calculation of wall-
to-bed heat-transfer coefficients in gas-fluidized beds. AIChE J. 38, 1079–1091.
Kurowski, L., Chmiel-Kurowska, K., Thullie, J., 2009. Numerical simulation of heat
transfer in nanofluids. In: Jezowski, J., Thullie, K. (Eds.), Computer Aided
Chemical Engineering, vol. 26. Elsevier, pp. 967–972.
Li, J., Kleinstreuer, C., 2008. Thermal performance of nanofluid flow in
microchannels. Int. J. Heat Fluid Flow 29, 1221–1232.
Lotfi, R., Saboohi, Y., Rashidi, A.M., 2010. Numerical study of forced convective heat
transfer of nanofluids: comparison of different approaches. Int. Commun. Heat
Mass Transfer 37, 74–78.
Mirmasoumi, S., Behzadmehr, A., 2008a. Numerical study of laminar mixed
convection of a nanofluid in a horizontal tube using two-phase mixture
model. Appl. Thermal Eng. 28, 717–727.
Mirmasoumi, S., Behzadmehr, A., 2008b. Effect of nanoparticles mean diameter on
mixed convection heat transfer of a nanofluid in a horizontal tube. Int. J. Heat
Fluid Flow 29, 557–566.
Morini, G.L., 2005. Viscous heating in liquid flows in micro-channel. Int. J. Heat Mass
Transfer 48, 3637–3647.
Patankar, S.V., 1980. Numerical Heat Transfer and Fluid Flow. Hemisphere,
Washington, DC.
Santra, A.K., Sen, S., Chakraborty, N., 2009. Study of heat transfer due to laminar flow
of copper–water nanofluid through two isothermally heated parallel plates. Int.
J. Thermal Sci. 48, 391–400.
Syamlal, M., Gidaspow, D., 1985. Hydrodynamics of fluidization: prediction of wall
to bed heat transfer coefficients. AIChE J. 31, 127–134.
Versteeg, H.K., Malalasekera, W., 1995. An introduction to computational fluid
dynamics the finite volume method. Longman Scientific and Technical, England.
Wakao, N., Kaguei, S., 1982. Heat and Mass Transfer in Packed Beds. Gordon and
Breach, New York.
Wen, D., Ding, Y., 2004. Experimental investigation into convective heat transfer of
nanofluids at the entrance region under laminar flow conditions. Int. J. Heat
Mass Transfer 47, 5181–5188.
Wu, X., Wu, H., Cheng, P., 2009. Pressure drop and heat transfer of Al2O3–H2O
nanofluids through silicon microchannels. J. Micromech. Microeng. 19, 105020.
11pp.
Xuan, Y., Li, Q., Hu, W., 2003. Aggregation structure and thermal conductivity of
nanofluids. AIChE J. 49, 1038–1043.