International Journal of Heat and Mass Transfer 68 (2014) 94–109

Contents lists available at ScienceDirect

International Journal of Heat and Mass Transfer

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

Effect of magnetic field on laminar convective heat transfer of magnetite

nanofluids
R. Azizian a,c, E. Doroodchi b, T. McKrell c, J. Buongiorno c, L.W. Hu d,⇑, B. Moghtaderi a,⇑
a
Center for Energy, Chemical Engineering Department, The University of Newcastle, Callaghan, NSW 2308, Australia
b
Center for Advanced Particle Processing, Chemical Engineering Department, The University of Newcastle, Callaghan, NSW 2308, Australia
c
Nuclear Science and Engineering Department, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139, USA
d
Nuclear Reactor Laboratory, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139, USA

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

Article history: The effect of an external magnetic field on the convective heat transfer and pressure drop of magnetite
Received 13 March 2013 nanofluids under laminar flow regime conditions (Re < 830) is investigated. Specifically, the influence
Received in revised form 24 July 2013 of magnetic field strength and uniformity on the convective heat transfer coefficient is examined through
Accepted 5 September 2013
experiments and supporting simulations of the magnetic flux density distribution and magnetic force act-
ing on nanoparticles. The data show that large enhancement in the local heat transfer coefficient can be
achieved by increasing the magnetic field strength and gradient. The convective heat transfer enhance-
Keywords:
ment becomes more pronounced at higher Reynolds numbers, with a four-fold enhancement (i.e., relative
Magnetite nanofluid
Thermal conductivity
to the case with no magnetic field) obtained at Re = 745 and magnetic field gradient of 32.5 mT/mm. The
Convective heat transfer coefficient effect of the magnetic field on the pressure drop is not as significant. The pressure drop increases only by
Laminar flow up to 7.5% when magnetic field intensity of 430 mT and gradients between 8.6 and 32.5 mT/mm are
Magnetic field applied. Based on the simulation results of magnetic field and magnetic force distribution, the mecha-
Aggregation nisms for heat transfer enhancement are postulated to be accumulation of particles near the magnets
(leading to higher thermal conductivity locally), and formation of aggregates acting enhancing momen-
tum and energy transfer in the flow.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction intensity, locally increases the heat transfer coefficient by up to

Magnetite nanofluids are engineered colloidal suspension of
magnetite nanoparticles in a basefluid, designed to be used in heat
transfer applications such as cooling systems of microdevices [1].
Recent studies have shown significant enhancement in the thermal
conductivity of magnetite nanofluids when magnetic fields are ap-
plied [2]. The main mechanism responsible for such enhancements
is believed to be particle alignment in the direction of the applied
magnetic field, parallel to the temperature gradient [2]. The align-
ment is assumed to reduce thermal resistance for heat transfer,
due to increasing the thermal conductivity of magnetite
nanofluids.
Remarkable heat transfer enhancements have been also re-
ported under forced convection conditions. Motozawa et al. [1]
examined the effect of uniform magnetic fields on heat transfer
coefficient of magnetite nanofluid in laminar flow regime experi-
mentally. They demonstrated that increasing the magnetic field
⇑ Corresponding authors. Tel.: +1 617 258 5860 (L.W. Hu). Tel.: +61 2 4985 4411;
fax: +61 249216893 (B. Moghtaderi).
E-mail addresses: Lwhu@mit.edu (L.W. Hu), Behdad.Moghtaderi@newcastle.
edu.au (B. Moghtaderi).
20%. Similarly, Lajevardi et al. [3] reported an enhancement in heat
transfer performance of magnetite nanofluid due to an increase in
the concentration of magnetite nanoparticles when a uniform
magnetic field was applied perpendicular to the flow. The observed
enhancement was attributed to increased aggregation at higher so-
lid volume fractions. Conversely, the investigation of Li and Xuan
[4] on the effect of external magnetic field strength and its orienta-
tion on heat transfer characteristics of magnetite nanofluid flow
around a fine wire showed enhancements only in presence of a
magnetic field gradient. The Kelvin forced-induced particle migra-
tion was considered to be responsible for the observed enhance-
ment. Li and Xuan [4] data however indicated that the heat
transfer coefficient can be reduced when a uniform magnetic field
perpendicular to the flow was applied. It is believed that the exper-
imental setup used by Li and Xuan [4] may have contributed to the
reported inconsistencies.
The aim of this study is to gain an insight into the effect of mag-
netic field on the laminar convective heat transfer of magnetite
nanofluids using a systematic experimental approach. Specifically
the effects of magnetic field uniformity, gradient, and strength as
well as magnets configuration, in terms of the number and
arrangement of magnets, on the convective heat transfer of

0017-9310/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijheatmasstransfer.2013.09.011
 R. Azizian et al. / International Journal of Heat and Mass Transfer 68 (2014) 94–109 95

Nomenclature

Variables z vertical distance from the entrance

B magnetic flux density
Br remnant magnetic flux density Greek symbols
cp specific heat capacity l dynamic viscosity
D diameter l0 permeability of free space
f friction factor lr relative permeability
F force t kinematic viscosity
h convective heat transfer coefficient q density
H magnetic field / particle volume fraction
k thermal conductivity vi magnetic susceptibility
kw temperature-dependent thermal conductivity of stain-
less steel Subscript and superscript
L length b bulk
Nu Nusselt number f fluid (liquid)
Pr Prandtl number i inside
Q thermal power M magnetic
q00 heat flux based on thermal power nf nanofluid
r radius o outside
Re Reynolds number p particle (solid)
T temperature w wall
V velocity

magnetite nanofluid in the laminar flow regime are examined. The Table 1
NdFeB, grade N42 permanent magnet specifications.
experimental findings combined with the simulation results for
magnetic flux density and magnetic force acting on nanoparticles, Maximum operating Temperature 80 °C
are then used to explain the potential driving mechanisms for the Surface field 333.2 mT
Residual flux density (Brmax) 1320 mT
observed enhancements under given conditions.
Maximum energy product (BHmax) 42 MGOe

2. Experimental method

2.1. Apparatus

A schematic diagram of the experimental setup used in this

study is shown in Fig. 1. The experimental setup was a closed loop

Fig. 1. A schematic diagram of the closed loop convective laminar flow system. Fig. 2. Particle size distribution of magnetite nanoparticles using (a) DLS, (b) TEM.
96 R. Azizian et al. / International Journal of Heat and Mass Transfer 68 (2014) 94–109

Table 2
Thermophysical properties of the magnetite nanoparticle, DI-water and magnetite nanofluid at 298 K.

Density (kg/m3) Specific heat (J/kg K) Thermal conductivity (W/mK) Viscosity (Ns/m2)
Magnetite nanoparticle (Fe3O4) 5180 670 80 –
DI-Water 997 4180 0.6 ± 0.012 0.0009 ± 0.0000078
Magnetite nanofluid
(0.86 Vol%) 1033 4029 0.6 ± 0.012 0.00127 ± 0.00002

Table 3
Simulation input data.

Input simulation parameters

Remnant magnetic flux density (Br) 1.32 T
Relative permeability of NdFeB permanent magnet (lr) 1.05
Magnetic permeability of free space (l0) 4p  107

Fig. 3. BH curve for 1018 low-carbon steel plate [11].

flow system equipped with a pump, flow meter, heat exchanger,

thermocouples and pressure transducer. The test section in the
flow system was fabricated from stainless steel tube with an inner
diameter of 5.54 mm, an outer diameter of 6.35 mm, and a length
of 1 m. Eleven K-type thermocouples were evenly distributed and
connected to the outer wall of the tubing along the test section.
A constant heat flux was provided across the test section using a
Lambda Genesys™ 20 V–500 A AC-to-DC power supply with a DC
output rating of 0–20 V and current of 500 A. The power was deliv-
ered to the test section through the copper electrode blocks provid-
ing a resistive heating to the test section. The heated section was
well insulated to minimise heat loss. A low-flow miniature gear
pump (McMaster-Carr 12 VDC) was used to circulate the fluid
through the loop flow system. The gear pump could provide a flow
rate of up to 2.31 L/min at the atmospheric pressure. The fluid vol-
umetric flow rate was measured using a FTB9504 Omega flow tur-
bine meter. A Polyscience heat exchanger was used to cool down
the fluid as it left the test section. This was achieved by recirculat-
ing chiller in the accumulator. Two RTD thermocouples were em-
ployed to measure the bulk fluid temperature at the inlet and Fig. 4. Nusselt number (Nu) vs. dimensionless distance (z+) for DI water at flow
outlet of the test section. The pressure loss in the closed loop sys- rates of (a) 0.13 L/min, (b) 0.16 L/min and (c) 0.26 L/min. Secondary x axis indicates
the thermocouple (TC) positions.
tem was measured using an Omega PX 154-001DI pressure trans-
ducer with 1% accuracy. A data acquisition system by Agilent
Technologies was used to record the output of all instruments. Be- the test section. The fluid was then returned to the accumulator
fore introducing the nanofluid to the loop, DI-water was used to where a heat exchanger maintained the fluid at a constant
ensure that the setup works properly. For each run, the fluid (either temperature.
DI-water or nanofluid) was introduced into the loop through an NdFeB, grade 42 block permanent magnets purchased from K&J
accumulator and was pumped through the system using the gear Magnetics were used to generate magnetic fields along the height
pump, passing the flow meter and entering the test section. The of the test section. These magnets had a length of 10.16 cm, a
fluid was heated up by a constant heat flux as it passed through width of 1.27 cm and a thickness of 0.635 cm. The specifications
 R. Azizian et al. / International Journal of Heat and Mass Transfer 68 (2014) 94–109
of the permanent magnets are summarised in Table 1. These mag-
nets were either directly positioned on the side of the test section
or were fitted over a 1018 low carbon steel plate. The plate was
used to generate a more uniform magnetic field along the test
section.
2.2. Magnetite nanofluid preparation, characterisation and properties
The polymer coated magnetite nanofluids used in this study
were prepared at the University of Newcastle using the one step
chemical precipitation method [5]. The procedure is described as
follows. First, the dissolved oxygen of DI-water was removed by
bubbling nitrogen through the water. Then a 1:2 molar ratio of fer-
rous chloride tetrahydrate (FeCl2  4H2O, Sigma Aldrich) and ferric
chloride hexahydrate (FeCl3  6H2O, Sigma Aldrich) was added to
the water. The mixture was stirred and heated to 80 °C. This was
Fig. 5. Nusselt number (Nu) vs. dimensionless distance (z+) for magnetite nanofluid
without a magnetic field at flow rates of (a) 0.13 L/min, (b) 0.16 L/min and (c)
0.26 L/min (T1–T11 present thermocouple positions).
97
followed by adding a coating polymer/surfactant, known as 4-sty-
renesulfonic acid-co-maleic acid (supplied by Sigma Aldrich), to
the mixture. The mixture then precipitated by adding 10 mL solu-
tion of ammonium hydroxide 30% (i.e., contains 30% ammonia by
weight supplied by Australian Scientific) to the mixture while stir-
ring vigorously. As ammonium hydroxide was added, the solution
turned black indicating the formation of iron oxide nanoparticles.
The solution was allowed to fully react at 80 °C for 30 min while
stirring. The fast adsorption of polymer on the particle surface pre-
vents the magnetite nanoparticles growth and hence the formation
of micro-sized particles. The overall stoichiometry of the reaction
is given as;
FeCl2  4H2 O þ 2FeCl3  6H2 O þ 8NH4 OH
! Fe3 O4 þ 8NH4 Cl þ 14H2 O ð1Þ
Nanoparticles were then rinsed with acetone and separated from
the liquid using an electromagnetic force. Finally, the excess ace-
tone was evaporated and the particles were dispersed in water.
The synthesised magnetite nanofluids were found to be stable for
more than six months.
The composition of the magnetite nanoparticles was verified to
be Fe3O4 using X-ray diffraction (XRD) method. The exact loading
of the magnetite nanoparticles measured with a Thermogravimet-
ric analyser (TGA). The TGA results confirmed that the weight con-
centration of the magnetite nanoparticles in water was 4.3%.
Dynamic light scattering (DLS) and transmission electron
microscopy (TEM) were used to measure the particle size in water
before and after the evaporation stage. Fig. 2(a) shows the nano-
particle size distribution in water using Zetasizer Nano ZS, Mal-
vern. As Fig. 2(a) shows, the mean diameter of the magnetite
particles are 60 nm with a very small fraction of the particles (less
than 0.065 wt%) having an average size of 150 nm.
A typical TEM result is shown in Fig. 2(b). In this figure, the
black dots represent magnetite nanoparticles. Similar to the DLS
results, the majority of particles were found to have a mean diam-
eter of 60 nm. Larger particles were also observed. However care
must be taken in interpreting these results since the drying stage
for conducting the TEM analysis may have contributed to the for-
mation of larger particles by promoting aggregate formation.
The thermophysical properties of magnetite nanoparticles,
DI-water and magnetite nanofluid are presented in Table 2.
The nanofluid bulk density, qnf, and specific heat, cp,nf, were esti-
mated using the following expressions [6];
qnf ¼ /qp þ ð1  /Þqf ð2Þ
Fig. 6. Pressure drop for DI-water and magnetite nanofluid (MNF) at different
Reynolds numbers. The solid lines represent the pressure drops calculated using
Darcy’s equation of pressure drop for fully developed laminar flows in a circular
pipe.
98 R. Azizian et al. / International Journal of Heat and Mass Transfer 68 (2014) 94–109

Fig. 7. Schematic diagram for the configuration of (a) trails of three single-magnet assembly (Case I), (b) trails of three single-magnet fitted onto a plate assembly (Case II),
and (c) trails of three double-magnet fitted onto a plate assembly (Case III) placed over thermocouple number T6, T7 and T8. (d) two pair of single permanent magnets
between thermocouples T6, T7, and T8 on both side (Case IV), (e) four pair of single permanent magnets over thermocouples T6–T9 on both side (Case V), (f) eight pair of
single permanent magnets over thermocouples T3–T10 on both side (Case VI), (g) three single permanent magnets over thermocouples T6, T7, and T8 only on one side of the
pipe (Case VII), (h) three single permanent magnets opposite to the thermocouples T6, T7, and T8 only on one side of the pipe (Case VIII), (i) five pair of single permanent
magnets over thermocouples T2, T4, T6, T8 and T10 on both side of the pipe (Case IX). (The numbers represent thermocouples position over the test section).
 R. Azizian et al. / International Journal of Heat and Mass Transfer 68 (2014) 94–109 99

/qp cp;p þ ð1  /Þqf cp;f calculation of magnetite nanofluid thermal conductivity through
cp;nf ¼ ð3Þ
qnf Maxwell equation that gives the thermal conductivity of magnetite
nanofluid same as basefluid (i.e. 0.615 in comparison to 0.6). Tem-
perature dependence of the magnetite nanofluid thermal conduc-
where / is the volume concentration, qp, is the particle density, and
tivity was also found to be the same as that of the water.
qf, is the base fluid density. cp.p, and cp,f are specific heats of the par-
Moreover, the viscosity of the nanofluid was measured using a cap-
ticle and base fluid, respectively.
illary viscometer. The viscosity of the nanofluid was found to be
The thermal conductivity of the 4.3 wt% magnetite nanofluid
27% higher than the water viscosity.
was measured with a short transient hot wire apparatus with
±2% accuracy. The detailed explanation of the hot wire setup is re-
ported elsewhere [7]. The thermal conductivity of the 4.3 wt% 3. Experimental data analysis
magnetite nanofluid was found to be the same as that of the water
mainly due to a very low volume concentration of the magnetite The convective heat transfer coefficient, h, was calculated based
particles (0.86 vol%). This observation is further confirmed by on the calculated inner wall temperature of the test section as well
as the local bulk temperature using the following expressions;
q00
h¼ ð4Þ
T w;i  T b
where q00 is the heat flux based on thermal power, Tw,i is the inner
wall temperature and Tb is the bulk temperature at the axial loca-
tion of interest that calculated according to the energy balance.
The inner wall temperature (Eq. (5)) is calculated using the analyt-
ical solution of the conduction equation with the measured outer
wall temperature as the boundary condition and the temperature-
dependent thermal resistance of stainless steel.
2 3
 
Q 4 D2o Do
T w;i ¼ T w;o    log  0:5 5 ð5Þ
2pkw L D2  D2 Di
o i

In Eq. (5), Q represents the thermal power derived from bulk inlet
and outlet coolant temperature difference, L is the length of the test
section, kw is the temperature-dependent thermal conductivity of
stainless steel, Do and Di are outer and inner diameters of the test
section respectively. The detailed derivation of h from measured
outer wall temperature can be found elsewhere [6].
The calculated convective heat transfer coefficient was then
used to determine the Nusselt number;
Nu ¼ hDi=k ð6Þ
where Di is the test section inner diameter, and k is the thermal con-
ductivity of the fluid at the local bulk temperature.
The theoretical local Nusselt number was estimated by Shah
equation (Eq. (7)). This equation is a curve fitting to the complex
analytical solution of the local Nusselt number for laminar flow
under the constant-heat flux boundary condition [8];
8
< 1:302zþ 1=3  0:5 zþ  0:003
2
Nu ¼ ð7Þ
: 4:364 þ 0:263zþ 0:506 e41ðzþ =2Þ zþ > 0:003
2

where z+ is a dimensionless distance defined as;

2ðz=Di Þ
zþ ¼ ð8Þ
RePr
In Eq. (8) z is the vertical distance from the entrance of the tube
 to 
qVDi
the position of the thermocouples,
c l  Re is the Reynolds number l ,
and Pr is the Prandtl number pk .
The uncertainty associated with the reported value of the Nu
number (i.e. Nu number calculated based on the experimental
data) was determined considering the uncertainties associated
with the measurements of the individual variables in the definition
of the Nu number. Assuming that all the measured variables are
un-correlated, the uncertainty in the reported value of the Nu num-
ber can be expressed as;
Fig. 8. Nusselt number (Nu) vs. dimensionless distance (z+) for the two cases of Case sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2  2  2
I: trail of three single magnets, and Case II: trail of three single magnets fitted to the @Nu @Nu @Nu
1018 low carbon steel plate at three different flow rates of (a) 0.13 L/min, (b) 0.16 L/
dNu ¼ dh þ dDi þ dk ð9Þ
@h @Di @k
min, and (c) 0.26 L/min (open symbols represent experimental Nu number in the
absence of magnetic field).
100 R. Azizian et al. / International Journal of Heat and Mass Transfer 68 (2014) 94–109

Fig. 9. (a) Magnetic flux density distribution on zx plane, (b) magnetic flux density and magnetic force distribution along the center line of the pipe in z direction, (c) magnetic
flux density and magnetic force distribution in radial direction along the x-axis for Case (I).

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2  2  2ffi B ¼ l0 lr H þ Br ð10Þ
@h 00 @h @h
where dh ¼ dq þ dT w;i þ dT b
@q00 @T w;i @T b
  
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 4 3 vi B
 00 2  00 2  00 2 FM ¼ pr p rB ð11Þ
00 @q
_
@q @q 3 l0
dq ¼ dm þ dT out þ dT in
@m _ @T out @T in
where B is the magnetic flux density, l0 is the magnetic permeabil-
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ity of free space, lr is the relative permeability, H is the magnetic
 2  2  2  2  2
@T w;i @T w;i @T w;i @T w;i @T w;i field and Br is the remnant magnetic flux density, vi is the magnetic
dT w;i ¼ dT w;o þ dm _ þ dT out þ dT in þ dkw
@T w;o @m _ @T out @T in @kw susceptibility of magnetite nanoparticle and rp is the particle radius.
A summary of the simulation input data for these scenarios is pre-
Eq. (9) was evaluated using MATLAB software. sented in Table 3.
However, for the scenarios where the magnets were fitted on a
4. Simulation of magnetic flux density and magnetic force 1018 low carbon steel plate, the magnetic flux density was simu-
distribution lated as a function of the magnetic field using Fig. 3.
The model was validated against our experimental data on
The magnetic flux density and magnetic force distribution were magnetic flux density for (i) single magnet, (ii) a trail of three sin-
simulated using COMSOL Multiphysics software. For the magnets gle-magnets fitted on a steel plate, and (iii) a trail of three double-
positioned directly on the side of the test section, the magnetic flux magnets fitted on a steel plate. A good agreement between the
density distribution and the magnetic force distribution were sim- experimental data and the model predictions was obtained indi-
ulated based on the Eqs. (10) and (11). The magnetic flux density cating that the COMSOL Multiphysics based- model can be used
distribution (Eq. (10)) is a generalised form of the constitutive rela- with confidence to gain an insight into the magnetic flux density
tionship for the magnetic field of the nonlinear materials (i.e. Max- and magnetic force distribution for the magnet assemblies used
well equations subject to certain boundary conditions) [9]. Kelvin’s in this study. The comparison of the experimental data and model
body force has also been used for simulation of magnetic force dis- predictions for the abovementioned scenarios are reported in the
tribution (i.e. Eq. (11)) [10] Appendix A.
 R. Azizian et al. / International Journal of Heat and Mass Transfer 68 (2014) 94–109
Fig. 10. (a) Magnetic flux density distribution on zx plane, (b) magnetic flux density and magnetic force distribution along the center line of the pipe in z direction, (c)
magnetic flux density and magnetic force distribution in radial direction along the x-axis for Case (II).
5. Results and discussion
5.1. Convective heat transfer characteristics of DI-water
A series of experiments were performed using DI-water as the
working fluid at flow rates of 0.13, 0.16 and 0.26 L/min (corre-
sponding to Re numbers of 524, 639 and 1061 respectively).
The collected data was then compared with the Shah equation
(Eq. (7)) to verify the correct functioning of the experimental
apparatus. Fig. 4 shows plots of Nusselt number (Nu) vs. dimen-
sionless distance (z+) for the DI water when no magnetic field
was applied. The secondary x-axis shows the corresponding po-
sition of the thermocouples for each data point. The solid lines
represent the theoretical predictions of Shah equation (Eq. (7))
and the symbols represent the Nu number calculated using the
experimental data. The dashed lines show the 10% upper and
lower limits of the theoretical Nu number. It can be seen that
as the dimensionless distance increases, the Nusselt number re-
duces, as expected when it approaches fully-developed laminar
flow. Clearly, a good agreement between the theoretical predic-
tions and the experimental data was obtained indicating that
the experimental setup can provide reliable measurements of
the variables associated with the convective heat transfer prop-
erties of water. The experiments were repeated whilst applying
a magnetic field. The results (not shown here) were found to
be the same as those obtained in the absence of the magnetic
field, in turn, indicating that the permanent magnet did not
interfere with the thermocouple readings.
101
5.2. Convective heat transfer characteristics of magnetite nanofluid
without a magnetic field
Similar experiments were conducted using magnetite nanofluid
at flow rates of 0.13, 0.16 and 0.26 L/min (corresponding to Re
numbers of 451, 508 and 868 respectively). The theoretical predic-
tions for the magnetite nanofluid studies in the absence of mag-
netic field were obtained by substituting the thermophysical
properties of the nanofluid in Eq. (7). As Fig. 5 shows the predic-
tions are in good agreement with the experimental data. Similar
agreements between the predictions and the experimental data
were observed in earlier studies on alumina-water and zirconia–
water nanofluids at MIT [6,12], whereby Nu number predictions
were obtained using the thermophysical properties of the nano-
fluid in Shah equation instead of the water thermophysical
properties.
A comparison of Figs. 4 and 5 shows that the difference between
the convective heat transfer characteristics of the DI-water and the
magnetite nanofluid is marginal indicating that negligible
enhancement in convective heat transfer can be achieved when
no magnetic field is applied.
The use of suspension of particles as a heat transfer fluid how-
ever could increase the pressure drop across a flow system. To
investigate the influence of introducing nanoparticles on the pres-
sure loss, a series of experiments were performed whereby the vis-
cous pressure loss was measured in the isothermal section of the
loop over a length of 60 cm and compared with that of the DI
water. Fig. 6 shows pressure drop vs. Reynolds number for DI water
102 R. Azizian et al. / International Journal of Heat and Mass Transfer 68 (2014) 94–109
Fig. 11. Nusselt number (Nu) vs. dimensionless distance (z+) for three different
cases of Case I: trail of three single magnets, Case VII: three single magnets over
thermocouples T6–T8 only on one side, and Case VIII: three single magnets opposite
to the thermocouples T6–T8 only on one side at three different flow rates of (a)
0.13 L/min, (b) 0.16 L/min, and (c) 0.26 L/min (open symbols represent experimen-
tal Nu number in the absence of magnetic field).
and the magnetite nanofluid. As Fig. 6 demonstrates, the pressure
drop increases by up to 35% when nanoparticles are introduced. A
similar rise in pressure loss was observed theoretically using
Darcy’s equation of pressure drop for fully developed laminar flows
2
in a circular pipe (DP ¼ f DLi q2V ; where L is the length, V is the fluid
velocity, and f is the friction factor defined as f ¼ 64 [8]). The rise in
Re
the pressure drop is mainly attributed to an increase in viscosity
due to addition of nanoparticles.
5.3. Convective heat transfer characteristics of magnetite nanofluid
with a magnetic field
Permanent magnets were used to apply magnetic field intensi-
ties ranging between 230 and 430 mT along the height of the test
section. The influence of the magnetic field on convective heat
transfer characteristics of the magnetite nanofluid was studied as
a function of the magnetic field uniformity, gradient, strength,
and magnets numbers and configurations. Schematic diagrams of
the assemblies and configurations used in this study are presented
in Fig. 7.
5.3.1. Magnetic field uniformity
The effect of magnetic field uniformity on convective heat
transfer characteristics of the magnetite nanofluid was examined
by comparing the Nusselt (Nu) number for trails of three single-
magnets (Case I) and trails of three single-magnets fitted onto a
1018 low carbon steel plate (Case II). The distance between the
magnets in each trail was 12 mm. The magnet assemblies were
then positioned over the thermocouple number T6, T7, and T8 on
both sides of the pipe. Fig. 8 shows the Nusselt (Nu) number vs.
dimensionless distance (z+) for Cases I and II at flow rates of (a)
0.13 L/min, (b) 0.16 L/min, and (c) 0.26 L/min. In these plots, the
convective heat transfer coefficients obtained experimentally un-
der the applied magnetic fields (i.e. closed symbols) are compared
with the convective heat transfer of magnetite nanofluid in the ab-
sence of a magnetic field (i.e. open symbol) and the theoretical pre-
dictions of Shah equation (i.e. solid line) at a given flow rate. As
Fig. 8 shows, applying the magnetic fields generally leads to an
enhancement in convective heat transfer coefficient. The convec-
tive heat transfer coefficient reached a maximum value at thermo-
couple numbers T7–T8 where the second and third magnets in the
trail of magnets were located. Beyond this point the convective
heat transfer began to weaken at almost the same rate at which
the maximum value for the convective heat transfer coefficient
was reached. Thermo-magnetic convection effect was proposed
earlier as one of the possible reasons for the observed enhance-
ment in convective heat transfer coefficient of magnetite nanofluid
under the effect of an external magnetic field [3]. However, in or-
der to have thermo-magnetic convection effect, large enough tem-
perature (to create enough magnetic susceptibility gradients) and
magnetic field gradients are required which is generally not the
case under forced convection situations. Thermo-magnetic convec-
tion is mostly effective under situations where natural convection
cannot provide adequate heat transfer (such as microelectronic
cooling). The observed increase in Nu number is assumed to be
associated with the aggregation of magnetite nanoparticles in the
direction of the external magnetic field. Aggregation of nanoparti-
cles leads to two distinct effects, i.e. enhancement of the local ther-
mal conductivity due to formation of low thermal resistance
pathways for the transport of heat, and enhancement in local con-
vective heat transfer coefficient due to interactions between the
aggregates and fluid flow. In the latter effect, the aggregates act
like local obstacles, which similarly to roughness in pipes, enhance
momentum and energy transfer; ultimately increasing the local
value of the convective heat transfer coefficient. It is difficult to
experimentally determine the relative importance of the thermal
conductivity enhancement effect and the local convective heat
transfer coefficient enhancement. As such, the Nu number in this
study is defined using the thermal conductivity of the magnetite
nanofluid measured in the absence of the external magnetic field,
and thus in the absence of aggregates. The drop in enhancement
of convective heat transfer coefficient in Fig. 8 corresponds to the
re-dispersion process of the aggregates as well as change in flow
pattern which happens in the absence of the magnetic field.
Fitting the magnets onto the plate however led to a reduction of
up to 38% in the value of the Nu number with respect to the cases
with the magnets but without the plate. To explain the observed
reduction in the convective heat transfer coefficient, the magnetic
flux density and magnetic force distribution for trails of three sin-
gle-magnets without and with a 1018 low carbon steel plate (i.e.
 R. Azizian et al. / International Journal of Heat and Mass Transfer 68 (2014) 94–109
Fig. 12. (a) Magnetic flux density distribution on zx plane, (b) magnetic flux density and magnetic force distribution along the center line of the pipe in z direction, (c)
magnetic flux density and magnetic force distribution in radial direction along x-axis for Case VII.
Fig. 13. Pressure drop at different Re numbers for two different cases presented in
Fig 7(a) and (g) in compared to the magnetite nanofluid (MNF) without any
magnetic field applied.
Cases I and II, respectively) were compared theoretically. In Figs. 9
and 10, part (a) shows the magnetic flux density distribution on zx
plane whilst parts (b) and (c) present the magnetic flux density and
magnetic force distribution along the center line of the pipe in z
and x directions, respectively. In the absence of the plate, a uniform
flux density of 430 mT was obtained away from the edges of the
magnets with a flux density gradient of 90 mT/mm exhibited at
the edges of the magnets (dashed line in Fig. 9(b)). Due to the di-
rect relation between magnetic force and magnetic flux density
103
gradient (Eq. (11)), large magnetic forces were expected at the
edges of the magnets (solid line in Fig. 9(b)). These large magnetic
forces cause trapping of magnetite nanoparticles at each magnet
position. At the same time, in the radial direction (x direction),
the magnetic flux density was found to be at its maximum of
459 mT close to the surface of magnets (dashed line in Fig. 9(c)).
The magnetic force was shown to have a gradient of 0.00023 pN/
mm with a zero magnetic force obtained at the centre line of the
pipe (solid line in Fig. 9(c)). This means the magnetic force drives
magnetite nanoparticles toward the walls on both sides.
Fitting the magnet assemblies onto a 1018 low carbon steel
plate on the other hand was found to remarkably improve the uni-
formity of the magnetic field profile in both z and x directions
(dashed lines in Fig. 10(b) and (c)) producing uniform magnetic
flux density along the test section in aforementioned directions.
A uniform magnetic field in turn produces almost zero magnetic
force acting on the magnetite nanoparticles (solid lines in
Fig. 10(b) and (c)). In this situation the only force acting on the
magnetite nanoparticles is the magnetic torque which tries to align
the magnetic moment of the particles with the external magnetic
field (magnetisation relaxation). This phenomenon may lead to
the formation of suspended aggregates. The drawback however
was that lower magnetic field strengths were achieved when the
magnets were fitted onto the plate. A comparison of Figs. 9 and
10 show that for the case of single-magnet assemblies, the mag-
netic strength was almost halved when the plates were used. The
observed reduction in the convective heat transfer enhancement
therefore is the combined effect of a more uniform magnetic field
profile and a lower magnetic field strength. To better understand
104 R. Azizian et al. / International Journal of Heat and Mass Transfer 68 (2014) 94–109
Fig. 14. Nusselt number (Nu) vs. dimensionless distance (z+) for three different
cases of Case I: trail of three single magnets, Case II: trail of three single magnets
fitted to the 1018 low carbon steel plate, and Case III: trail of three double magnets
fitted to the 1018 low carbon steel plate at three different flow rates of (a) 0.13 L/
min, (b) 0.16 L/min, and (c) 0.26 L/min (open symbols represent experimental Nu
number in the absence of magnetic field).
the individual effect of each parameter on the convective heat
transfer coefficient, a series of experiments were performed focus-
ing on the magnetic field gradient and the magnetic field strength
independently.
5.3.2. Magnetic field gradient
A magnetic field gradient was enforced by positioning the per-
manent magnets on one side of the tube. Fig. 11 shows a compar-
ison between the cases where a trail of three permanent magnets
was placed on both sides of the pipe (Case I), only one side of
the pipe over the thermocouple number T6–T8 (Case VII) and only
one side of the pipe opposite to the thermocouple number T6–T8
(Case VIII). Fig. 11shows the Nu number vs. dimensionless distance
for the abovementioned cases at flow rates of (a) 0.13 L/min, (b)
0.16 L/min, and (c) 0.26 L/min. The convective heat transfer coeffi-
cient enhancement was much higher when the magnets were
located on only one side over the thermocouple number T6–T8
with the maximum value of Nu number reaching four times the
no magnet case at the flow rate of 0.26 L/min.
This behaviour is considered to be a result of the large magnetic
field gradient (rB) encountered when the magnets were placed on
only one side of the pipe. Simulation of magnetic flux density and
magnetic force for Case (VII) (dashed line and solid line in
Fig. 12(b) and (c)) confirms the presence of such a large magnetic
flux gradient and negative magnetic force in the x direction (solid
line in Fig. 12(c)). Generally, the force acting on the particle leads
to the translational motion of particles driving them over the mag-
nets (solid line in Fig. 12(b)) and at the same time to the surface of
the pipe near the magnets (solid line in Fig. 12(c)). The migration of
particles to the wall leads to an increase in local particles concen-
tration and consecutively enhances the local thermal conductivity.
A rise in the thermal conductivity coefficient in turn leads to an in-
crease in the convective heat transfer coefficient considering that
the convective heat transfer coefficient has a direct relationship
with the thermal conductivity coefficient.
The above argument was further supported by the results of the
experiments performed using a trail of three single magnets on
only one side of the pipe located opposite to the thermocouples
(Fig. 7(h)). It can be seen from Fig. 11 that the enhancement in con-
vective heat transfer coefficient for Case VIII is much lower than
when the magnets were positioned over the thermocouples (Case
VII). That is because in Case VIII as the particles migrate towards
the magnets, the particle concentration next to the thermocouples
reduces. The lower particle concentration next to the thermocou-
ples leads to lower enhancement effects.
In Case I, Fig. 9(c), the near-uniform magnetic flux density at the
centre of the pipe (dashed line in Fig. 9(c)) causes the particles to
rotate and elongate in the direction of the magnetic field (magnet-
isation relaxation); simultaneously small magnetic force gradient
in x direction (i.e. solid line in Fig. 9(c)) moves the particles as well
as formed aggregates toward the walls. As the particles accumulate
on the wall, chain-like aggregates are likely to form with their ori-
gin next to the pipe wall. Alternatively the aggregates may create a
low resistance heat transfer path resulting in an increased local
thermal conductivity.
Moreover, the findings of these studies show an enhancement
effect downstream of the test section where no magnetic field is
applied. This enhancement can be attributed to possible changes
in the thermal boundary layer thickness and flow pattern as the
particles migrate to the pipe wall under the applied magnetic field
(momentum exchange between the flow and aggregates) and/or
the kinetics of aggregate dispersion when no magnetic field is ap-
plied. Generally speaking, the change in the thermal boundary
layer thickness inversely affects the convective heat transfer coef-
ficient. Therefore a reduction in the boundary layer thickness due
to presence of the particles at the wall may have contributed (par-
ticles at the wall act as pipe roughness) to the observed enhance-
ment downstream of the test section. Alternatively, during
aggregate dispersion stage, the aggregate size reduces in size at a
given rate. The dispersion rate is governed by the particle surface
forces and the flow conditions. The enhancement effect therefore
continues whilst the aggregate size reduces to or close to its pri-
mary nanoparticle size. This argument is supported by the earlier
studies which have shown an increase in the particle size could
lead to an enhancement in conductive heat transfer coefficient of
nanofluids [13].
Furthermore, it can be seen that in all of the reported results,
the enhancement is more pronounced at higher flow rates
 R. Azizian et al. / International Journal of Heat and Mass Transfer 68 (2014) 94–109
Fig. 15. (a) Magnetic flux density distribution on zx plane, (b) magnetic flux density and magnetic force distribution along the center line of the pipe in z direction, (c)
magnetic flux density and magnetic force distribution in radial direction along the x-axis for Case III.
Fig. 16. Pressure drop at different Re numbers for Cases II and III presented in
Fig. 7(b) and (c) in compared to the magnetite nanofluid (MNF) without any
magnetic field applied.
(Fig. 11 for instance). This may be due to the enhanced mixing
caused by the aggregates, which act as roughness obstacles, at
higher flow rates.
It was mentioned earlier that as the magnetic field becomes
more uniform, for instance in Case II (dashed line in Fig. 10(c)),
the magnitude of the magnetic force (solid line in Fig. 10(c)) ap-
proaches the zero value. When the force is zero, the particles trans-
lational motion is no longer affected by the magnetic force. A
uniform magnetic flux then gives rise to a torque, aligning the par-
ticles in the direction of the magnetic field [14]. That in turn may
105
lead to the formation of suspended chain-like aggregates increas-
ing the apparent diameter of the suspended particles. Therefore
the increase in the apparent diameter of particles due to aggrega-
tion can be potentially responsible for the observed enhancement
in a uniform magnetic field. Also it is shown that, in a capillary
tube, the presence of suspended aggregates could alter the fluid
flow profile from a parabolic profile to a flat profile [15]. The fluid
velocity profile in macrosize tube could also undergo similar
changes as the aggregates gather in the centre of the pipe increas-
ing the fluid velocity at the wall.
Fig. 13 shows a plot of pressure drop vs. Re number for the cases
when magnets are placed on both sides of the pipe or only one side
of the pipe. The pressure drop experiments were carried out sepa-
rately and in isothermal conditions, for the same magnet configu-
rations used in the heat transfer study, but now applied to the
pressure drop test section (Fig. 1). The results are compared with
those of no magnet case. The pressure drop for the one side of
the pipe only was found to be 7.2% higher than the case with mag-
nets located on both sides of the pipe. The rise in pressure drop
could be associated with an increase in particle concentration
and aggregations which act as roughness along the test section in
conjunction with magneto-viscous effect in the results of an exter-
nal magnetic field which is applied perpendicular to the fluid flow
[16]. The relation between Nu number and Re number (for instance
Fig. 8a–c) as well as the relationship between pressure drop and Re
number (for instance Fig. 13) is presented for every scenario. To
draw a conclusion in a more explicit manner, the relation between
average Nu number (i.e. Nu number between T5–T9) and pressure
drop is also clarified (not shown here). Clearly as the pressure drop
increases, the Nu number increases as well.
106 R. Azizian et al. / International Journal of Heat and Mass Transfer 68 (2014) 94–109
Fig. 17. Nusselt number (Nu) vs. dimensionless distance (z+) for two different cases
of Case I: trail of three single magnets, Case IV: two pair of single permanent
magnet between thermocouples T6, T7 and T8 at three different flow rates of (a)
0.13 L/min, (b) 0.16 L/min, and (c) 0.26 L/min (open symbols represent experimen-
tal Nu number in the absence of magnetic field).
It is worth noting that in the absence of direct measurements of
particle distribution in the channels, the assumptions regarding
particle aggregation and its effect on heat transfer and pressure
drop are rather speculative, but can be deemed to be plausible.
5.3.3. Magnetic field strength
The magnetic field strength was altered by assembling three
double magnets fitted onto a 1018 low carbon steel plate. The re-
sults were then compared with the three single magnets fitted
onto a 1018 low carbon steel plate assembly. Fig. 14 shows the
Nusselt (Nu) number vs. dimensionless distance (z+) for single-
magnet assembly (Case II) and double magnet assembly (Case III)
fitted onto a plate at flow rates of (a) 0.13 L/min, (b) 0.16 L/min,
Fig. 18. Nusselt number (Nu) vs. dimensionless distance (z+) for two different cases
of Case I: trail of three single magnets, Case V: four pair of single permanent magnet
over thermocouples T6–T9 on both side of the pipe at three different flow rates of
(a) 0.13 L/min, (b) 0.16 L/min, and (c) 0.26 L/min (open symbols represent exper-
imental Nu number in the absence of magnetic field).
and (c) 0.26 L/min. It can be seen that greater enhancements were
achieved when a stack of double magnets were used indicating
higher magnetic strengths generated under these conditions lead
to greater connective heat transfer coefficients.
To determine the magnitude of rise in the magnetic field
strength, the magnetic flux density and magnetic force distribu-
tion for Cases II and III were simulated and compared (Figs. 10
and 15). As Fig. 15 shows the magnetic field strength for the
double-magnet assembly is almost 37% greater than that of
the single-magnet assembly with no magnetic field gradient ob-
served in the radial direction (Fig. 15(c)). The greater magnetic
field intensities could increase the likelihood of forming larger
aggregates with longer chain-like structure. As mentioned in
 R. Azizian et al. / International Journal of Heat and Mass Transfer 68 (2014) 94–109
Fig. 19. Nusselt number (Nu) vs. dimensionless distance (z+) for three different
cases of Case VI: trail of eight pair of single magnets over thermocouples T3–T10,
and Case IX: five pair of single magnets over thermocouples T2, T4, T6, T8 and T10
at three different flow rates of (a) 0.13 L/min, (b) 0.16 L/min, and (c) 0.26 L/min
(open symbols represent experimental Nu number in the absence of magnetic field).
the previous section, the larger aggregates in turn can provide
higher heat transfer coefficient.
A comparison of the pressure drop measurements for Cases II
and III was also performed (Fig. 16). The result shows a slight in-
crease in pressure drop as the magnetic field strength increased
from 210 to 345 mT. Clearly this is very encouraging since a signif-
icant enhancement in convective heat transfer coefficient (up to
35%) can be achieved with only a slight increase in pressure drop.
5.3.4. Permanent magnets configuration
In this section the magnetic field was altered by changing the
magnets configuration in terms of the number of trailing perma-
nent magnets and their arrangement. The experiments were per-
formed using single permanent magnets on both side of the pipe.
107
5.3.4.1. Number of permanent magnets. Fig. 7(d), (a) and (e) show
the schematic diagrams of the magnets arrangements along the
test section for Cases IV, I and V. For the first comparison, trails
of two and three permanent magnets were placed on both sides
of the pipe covering different lengths of the test section. In the case
of two single permanent magnets, the magnets were placed be-
tween thermocouples T6, T7, and T8 on both sides of the test sec-
tion (Case IV) whilst in the case of three single magnets (Case I), the
magnets were positioned over the thermocouple number T6–T8
respectively.
Fig. 17 shows the Nu number vs. dimensionless distance (z+) for
the case of two single magnets (Case IV) and three single magnets
(Case I) at flow rates of (a) 0.13 L/min, (b) 0.16 L/min, and (c) 26 L/
min. the experimental results show an increase in the convective
heat transfer coefficient even at the locations between two trailing
magnets where the magnetic flux is relatively small (i.e. T7 and T8
for Case IV). This behaviour indicates that the aggregates structures
remain unchanged (negligible dispersion) over the distance be-
tween the two trailing magnets. As discussed earlier, another pos-
sible reason for the enhancement effect exhibited between the
magnets could be also explained by possible change in the flow
pattern.
A comparison between Nu number as a function of dimension-
less distance (z+) for the case of three and four trailing magnets (i.e.
Cases I and V respectively) at three different flow rates are pre-
sented in Fig. 18. As Fig. 18 shows, the maximum value of Nu num-
ber was found to be independent of the number of trailing magnets
indicating that the maximum enhancement is governed by the
magnetic field strength only. Also it can be observed that the max-
imum enhancement generally achieved where the last magnet was
positioned.
5.3.4.2. Permanent magnet arrangements. Two other configurations
were also examined in which the magnets were arranged on every
second thermocouples on both sides of the pipe (i.e. 125.6 mm
apart) (Fig. 7(i)). The results were then compared with the case
in which the magnets were placed next to each other (i.e. 12 mm
apart) on both sides of the pipe (Fig. 7(f)).
Fig. 19 shows the Nu number at different positions for the last
two scenarios at different flow rates. A more uniform enhancement
profile was achieved when the magnets were staggered. Con-
versely the convective heat transfer coefficient was found to fluc-
tuate along the length of the test section when the magnets were
positioned next to each other. The observed behaviours could be
explained by the dynamics of the aggregate growth, breakage
and dispersion assuming the convective heat transfer coefficient
enhancement is directly related to the aggregate size. The fluctua-
tion possibly corresponds with the aggregate growth followed by a
breakage. The breakage happens when the aggregates reach their
maximum size under a given conditions. This is followed by a
growth process where the primary particles in the suspension join
the broken aggregates.
6. Conclusions
Convective heat transfer coefficient of magnetite nanofluid in
laminar flow regime was investigated under applied external mag-
netic field. It was demonstrated that the local heat transfer coeffi-
cient of magnetite nanofluids can be increased significantly (up to
300%) when a magnetic field was applied. The observed enhance-
ments were found to be a function of flow rate (Re number), mag-
netic field strength and gradient. The largest local heat transfer
coefficient enhancement occurs near the fully-developed regime
and when a large magnetic flux gradient is applied in the radial
direction. The main mechanism responsible for the convective heat
108 R. Azizian et al. / International Journal of Heat and Mass Transfer 68 (2014) 94–109
transfer enhancement was assumed to be particle migration driven
by the magnetic field gradient. Other factors potentially affecting
the convective heat transfer enhancement were aggregation kinet-
ics and fluid flow profiles.
Acknowledgments
The authors wish to acknowledge the financial support of the
University of Newcastle (Australia), Granite Power Pty Ltd and
the Australian Research Council through the ARC-Linkage grant
LP100200871 (2010–2012). The authors also wish to thank Mr. Eric
Forest of MIT for his suggestions regarding nanoparticle production
and his advice on the experimental setup. The support and assis-
Fig. I-1. Simulation results of magnetic flux density for a single permanent magnet
vs. experimental data.
Fig. I-2. Simulation results vs. experimental data (a) for a trail of three single magnets fitted onto the plate at the surface, (b) for a trail of three single magnets fitted onto the
plate at 0.25 cm above the surface, (c) for a trail of three double magnets fitted onto the plate at the surface, (d) for a trail of three double magnets fitted onto the plate at
0.25 cm above the surface.
tant provided to the team by Mr. Tat Nghia Nguyen of MIT is also
much appreciated. Similarly the authors thank Dr. Shahriar
Khushrushahi of MIT for his valuable advice and constructive sug-
gestions. The heat transfer experiments conducted at MIT were
supported by King Saud University, Saudi Arabia.
Appendix A
Three dimensional simulations of magnetic flux density using
COMSOL Multiphysics software for (i) single magnet, (ii) a trail of
three single-magnets fitted on a steel plate and (iii) a trail of three
double-magnets fitted on a steel plate are presented in Figs. I-1 and
I-2, respectively. In these figures the solid line is the model predic-
tions and the symbols are the flux density data obtained experi-
mentally using a digital Gaussmeter (DGM-202). The
experimental measurements were performed at the surface of
the magnet and at a distance away from the surface of the magnet.
Generally a satisfactory agreement between the experimental re-
sults and the model predictions were obtained which verifies the
validity of the model.
References
[1] M. Motozawa, J. Chang, T. Sawada, Y. Kawaguchi, Effect of magnetic field on
heat transfer in rectangular duct flow of a magnetic fluid, Phys. Proc. 9 (2010)
190–193.
[2] J. Philip, P.D. Shima, B. Raj, Enhancement of thermal conductivity in magnetite
based nanofluid due to chainlike structure, Appl. Phys. Lett. 91 (2007) 203108.
[3] M. Lajvardi, J.M. Rad, I. Hadi, A. Gavili, T.D. Isfahani, F. Zabihi, J. Sabbaghzadeh,
Experimental investigation for enhanced ferrofluid heat transfer under
magnetic field effect, J. Mag. Mag. Mater. 322 (2010) 3508.
[4] Q. Li, Y. Xuan, Experimental investigation on heat transfer characteristics of
magnetic fluid flow around a fine wire under the influence of an external
magnetic field, Exp. Therm. Fluid Sci. 33 (2009) 591–596.
 R. Azizian et al. / International Journal of Heat and Mass Transfer 68 (2014) 94–109
[5] W.C. Williams, E. Forrest, L.W. Hu, J. Buongiorno, Preparation and
characterization of water-based nanofluids for nuclear applications, in:
Proceeding of ICAPP 06, Reno, NV, USA, June, 2006.
[6] U. Rea, T. McKrell, L.W. Hu, J. Buongiorno, Laminar convective heat transfer and
viscous pressure loss of alumina–water and zirconia–water nanofluids, Int. J.
Heat Mass Transfer 52 (2009) 2042–2048.
[7] R. Rusconi, W.C. Williams, J. Buongiorno, R. Piazza, L.W. Hu, Numerical
analysis of convective instabilities in a transient short-hot-wire setup for
measurement of liquid thermal conductivity, Int. J. Thermophys. 28 (2007)
1131–1146.
[8] J.H. Lienhard IV, J.H. Lienhard V, A Heat Transfer Textbook, second ed.,
Phlogiston Press, 2002.
[9] COMSOL Multiphysics Modeling Guide (www.comsol.com).
[10] Q.A. Pankhurst, J. Connolly, S.K. Jones, J. Dobson, Applications of magnetic
nanoparticles in biomedicine, J. Phys. D: Appl. Phys. 36 (2003) 167–181.
[11] J. Ziegenbein, Magnetic clamping structures for the consolidation of composite
laminates, MSc. Thesis, Georgia Institute of Technology, 2011.
109
[12] 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 (nanofluids) in horizontal tubes, J. Heat Transfer
130 (2008) 042412.
[13] M.P. Beck, Y. Yuan, P. Warrier, A.S. Teja, The effect of particle size on the
thermal conductivity of alumina nanofluids, J. Nanoparticle Res. 11 (2009)
1129–1136.
[14] A.-L. Gassner, M. Abonnenc, H.-X. Chen, J. Morandini, J. Josserand, J.S. Rossier, J-
M. Busnel, H.H. Girault, Magnetic forces produced by rectangular permanent
magnets in static microsytems, Lab on a Chip 9 (2009) 2356–2363.
[15] H. Kikura, J. Mastsushita, N. Kakuta, Y. Kobayashi, M. Aritomi, Flow
visualization of ferromagnetic nanoparticles on microchannel flow using
dark field microscopy, in: 16th international Symposium on Transport
Phenomena, Prague, 2005.
[16] N. Andhariya, B. Chudasama, R. Patel, R.V. Upadhyay, R.V. Mehta, Field induced
rotational viscosity of ferrofluid: effect of capillary size and magnetic field
direction, J. Colloid Interface Sci. 323 (2008) 153–157.