Você está na página 1de 16

Hydraulic and heat transfer study of SiO

2
/water nanouids in horizontal tubes
with imposed wall temperature boundary conditions
Sbastien Ferrouillat
a,
, Andr Bontemps
a
, Joo-Paulo Ribeiro
b
, Jean-Antoine Gruss
b
, Olivier Soriano
b
a
Universit Joseph Fourier, LEGI, BP 53X, 38041 Grenoble cedex, France
b
CEA/LITEN/DTS/LETH, 17, Avenue des martyrs, 38052 Grenoble cedex, France
a r t i c l e i n f o
Article history:
Received 6 October 2009
Received in revised form 25 August 2010
Accepted 18 January 2011
Available online 18 February 2011
Keywords:
Nanouid
Convective heat transfer
Imposed wall temperature
a b s t r a c t
The convective heat transfer of SiO
2
/water colloidal suspensions (534 wt.%) is investigated experimen-
tally in a ow loop with a horizontal tube test section whose wall temperature is imposed. Experiments
were performed at different inlet temperatures (20, 50, 70 C) in cooling and/or heating conditions at var-
ious ow rates (200 < Re < 10,000). The Reynolds and Nusselt numbers were deduced by using thermal
conductivity and viscosity values measured with the same temperature conditions as those in the tests.
Results indicate that the heat transfer coefcient values are increased from 10% to 60% compared to those
of pure water. They also show that the general trend of standard correlations is respected. The problem of
suspension stability at the highest temperatures is discussed. In order to evaluate the benets provided
by the enhanced properties of the nanouids studied, an energetic performance evaluation criterion (PEC)
is dened. This PEC decreases as the nanoparticle concentration is increased. This process is also dis-
cussed in this paper.
2011 Elsevier Inc. All rights reserved.
1. Introduction
The development of high-performance thermal systems has
been stimulated in many elds of new technologies. Conventional
heat transfer devices have to be substantially improved to answer
the needs of systems from the microscale to large power plants. In
this perspective, convective heat transfer can be enhanced in sev-
eral ways, by using either active or passive techniques. In the latter
case, it is made possible by changing the structure of the heat ex-
changer or the properties of the heat exchange surface. However
another possibility is to modify the uid itself by enhancing its
thermal conductivity. Various techniques have been used to in-
crease the thermal conductivity of base uids by introducing solid
particles whose conductivity is generally higher than that of liq-
uids. A new class of uids called nanouids have recently been
developed and tested. As a result, we are seeing an increasing
amount of published work on the subject. Nanouids are colloids
made of a base uid and nanoparticles (1100 nm). A substantial
increase in the thermal conductivities of nanouids containing a
small amount of metallic or non-metallic nanoparticles has been
reported. In addition, some authors have measured an increase of
heat transfer coefcients compared to pure liquids beyond the
mere thermal-conductivity effect. However, many results found
in published literature are not consistent with others or with stan-
dard correlations. Several reasons have been put forward to explain
these discrepancies. In an attempt to explain this type of
behaviour, a range of experiments has been dened: measure-
ments of thermophysical properties of nanouids, measurements
of pressure drop in channels, measurements of heat transfer coef-
cients with different boundary conditions, stability of colloidal
suspensions.
2. Selected bibliography
To characterise heat transfer in forced-convection, the heat
transfer coefcient is one of the parameters to be determined. It
takes into account the uid thermal conductivity either directly
or indirectly using the Nusselt number. Thus, a rst assessment
of the heat transfer potential of a nanouid involves considering
its thermal conductivity. Up to now, most of research has been
published in this area because thermal conductivity is probably
easier to measure than heat transfer coefcients. Consequently,
thermal conductivity results have been used extensively to esti-
mate nanouid heat transfer enhancement rates. Nevertheless,
while increases in effective thermal conductivity as well as
changes in density, specic heat, and viscosity are important indi-
cations of improved heat transfer behaviour of nanouids, the net
benet of nanouids as heat transfer uids is determined through
the heat transfer coefcient. Thus, it is essential to directly mea-
sure this coefcient under ow conditions typical of specic appli-
cations and, until now, there has been limited experimental work
0142-727X/$ - see front matter 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.ijheatuidow.2011.01.003

Corresponding author.
E-mail address: sebastien.ferrouillat@ujf-grenoble.fr (S. Ferrouillat).
International Journal of Heat and Fluid Flow 32 (2011) 424439
Contents lists available at ScienceDirect
International Journal of Heat and Fluid Flow
j our nal homepage: www. el sevi er. com/ l ocat e/ i j hf f
reported in published literature, as shown in Table 1. The nanouid
types, testing parameters and a summary of results are listed in
this table. Many research groups have found that heat transfer
enhancement exceeds thermal conductivity enhancement in lami-
nar ow (Chen et al., 2008; Faulkner et al., 2004; Hwang et al.,
2009; Jung et al., 2009; Lai et al., 2008; Lee and Choi, 1996; Lee
et al., 2005; Li and Xuan, 2002; Rea et al., 2009; Wen and Ding,
2004; Xuan and Li, 2003; Zeinali Heris et al., 2006a,b, 2007). This
nding indicates that the presence of nanoparticles in the ow
inuences the heat transfer beyond what would be expected from
increased thermal conductivity alone. Some authors have attrib-
uted this added effect to particle-uid interactions.
First, it must be remarked that due to the particle size, it has
generally been considered that the two-phase solidliquid ow
does not lead to specic ow patterns.
The effect of particle volume concentrations have been studied
by several research groups. In laminar ow, at particle volume con-
centrations below 2%, there is a limited Reynolds effect on heat
transfer enhancement. For particle volume concentrations above
2%, the heat transfer enhancement augments with the Reynolds
number. This trend is consistent with the increase of thermal con-
ductivity with increased particle volume concentration. However,
it has been noticed that heat transfer enhancement is more sub-
stantial than thermal conductivity enhancement.
Nevertheless, two groups found that the heat transfer enhance-
ment either is much lower than the effective thermal conductivity
enhancement (graphite nanouids, Yang et al., 2005) or is not sig-
nicant (nano-diamond nanouid or ethylene based titanium
nanouid, Ding et al. (2007)).
Amongst all nanouids tested, Carbon Nano Tube (CNT) solu-
tions seem to provide the highest heat transfer enhancement
(Faulkner et al., 2004; Ding et al., 2006) compared to other nano-
particles (Al
2
O
3
, CuO, TiO
2
, graphite, etc.). This enhancement can
be related to several potential reasons: improved thermal conduc-
tivity, shear-induced enhancement in ow, reduced boundary
layer, particle re-arrangement, and high aspect ratio of CNTs.
Indeed, particle shape or aspect ratio should be an important fac-
tor. Studies with nearly spherical nanoparticles (aspect ratio
around 1) (Li and Xuan, 2002; Wen and Ding, 2004; Xuan and Li,
2003) show an increase of the convective heat transfer coefcient
up to 60%. Results cited previously on CNT nanouids, which are
characterised by an aspect ratio above 100, show a heat transfer
enhancement of up to 350% at Re = 800 for 0.5 wt.% nanoparticle
concentration. However, graphite nanouids results cited previ-
ously (Yang et al., 2005), with an aspect ratio lower than 0.02,
showed a much lower increase of the convective heat transfer coef-
cient with respect to the effective thermal conductivity. Thus, the
available experimental data seem to show that the particle shape
and the aspect ratio are signicant parameters which affect the
thermal performance of nanouids. However, is yet to be examined
in-depth.
Heat transfer results in turbulent ow are available from few
groups (He et al., 2007; Kulkarni et al., 2008; Li and Xuan, 2002;
Nguyen et al., 2007; Pak and Cho, 1998; Sommers and Yerkes,
2009; Williams et al., 2008; Xuan and Li, 2003; Yu et al., 2009).
Some of them (He et al., 2007; Kulkarni et al., 2008; Li and Xuan,
2002; Nguyen et al., 2007; Xuan and Li, 2003) reported heat trans-
fer enhancement for turbulent ow higher than predicted by the
pure uid correlation (DittusBoelter), even when the measured
nanouid properties were used in dening the dimensionless
groups in the correlation. However, these same researchers show
that turbulent friction factors in their nanouids can be predicted
by the traditional friction factor correlations for pure uids if the
measured nanouid viscosity is used. Moreover, it has been shown
that the heat transfer enhancement increases with increased parti-
cle volume concentration. The heat transfer enhancement is the
highest for Cu particles, (Table 1), followed by Al
2
O
3
particles
and then TiO
2
particles at the same concentration levels. When
taking into account thermal conductivity, it is not surprising that
the Cuwater nanouid shows the highest heat transfer enhance-
ment. However, the thermal conductivity enhancements of
Al
2
O
3
and TiO
2
in water are similar although the heat transfer
enhancement of Al
2
O
3
in water is higher than that of TiO
2
in water.
Nevertheless, two groups found that the heat transfer coefcient of
Nomenclature
A
t
thermocouple cross section area
C
p
specic heat capacity
d tube diameter
d
h
hydraulic diameter
D thermocouple diameter
f fanning friction factor
h heat transfer coefcient
h
t
heat transfer coefcient near the thermocouple
k thermal conductivity
l thermocouple length
L tube length
_ m mass ow rate
Nu Nusselt number
P perimeter of the thermocouple
Pe Peclet number
Pr Prandtl number
_
Q heat ow rate
R
w
thermal resistance of the copper tube wall
Re Reynolds number
S heat exchange area
S
p
cross section area
U overall heat transfer coefcient
Greek symbols
b shape factor
DP pressure drop
DT
lm
log mean temperature difference
e absolute roughness
u nanouid volume fraction
u
w
nanouid mass fraction
l dynamic viscosity
K Darcy coefcient
q density
w particle sphericity
Subscripts
b bulk
e external
exp experimental
f base uid
i internal
in inlet
nf nanouid
out outlet
s nanoparticles
t thermocouple
w wall
S. Ferrouillat et al. / International Journal of Heat and Fluid Flow 32 (2011) 424439 425
nanouids (Al
2
O
3
and TiO
2
in water and SiC in water) was lower
than for pure water for constant average velocity in turbulent ow
(Pak and Cho, 1998; Yu et al., 2009).
Recently, Williams et al. (2008) and Rea et al. (2009) studied
turbulent convective heat transfer behaviour of Al
2
O
3
and ZrO
2
nanoparticle dispersions in water. They demonstrated that if the
measured temperatures dependent on the thermal conductiv-
ity and viscosity of the nanouids are used in calculating the
Reynolds, Prandlt, and Nusselt numbers, the existing correlations
(DittusBoelter and Blasius/MacAdams) accurately reproduce the
experimental convective heat transfer and viscous pressure loss
behaviour in tubes. This nding indicates that no abnormal heat
transfer enhancement was observed when nanouid properties
were accurately taken into account.
At present, there is too little data to establish the heat transfer
enhancement trend with laminar or turbulent ow as a function of
particle type and/or size. The thermal conductivity enhancement
seems to increase with particle size, but more experiments are re-
quired to establish this type of trend with regard to heat transfer
enhancement. Some authors show a heat transfer coefcient
enhancement with particle size (Kulkarni et al., 2008) and some
others show that for a given ow Reynolds number and particle
concentration, the convective heat transfer coefcient does not
seem to be sensitive to the average particle size (He et al., 2007 ).
Moreover, as nanouids are usually moderately concentrated
suspensions of anisotropic objects, they can be non-Newtonian
materials with a thermodependent rheology. Therefore, it is funda-
mental that a thermo-rheo-structural study of these materials be
conducted. As a conclusion of this scientic work, it seems clear
that there is a general lack of characterisation concerning the ther-
mal properties of nanouids.
Few studies have been conducted which carefully examine
nanouid stability (Sommers and Yerkes, 2009; Bontemps et al.,
2008a,b). Discoloration of the nanouid has been observed after
being cycled at high ow rates and increased temperatures for long
periods of time. This may be the result of nano-abrasion occurring
in the loop. The authors suppose that the optical change was due to
trace contaminants. Nanouid stability must be taken into account
in dening new nanouids for heat transfer purposes.
3. Nanouid characterisation
What is called nanouid is generally a dilute suspension of
nanoparticles (volume fraction 65%). To extend the eld of the
present study and to possibly relate our results to those obtained
with rheological uids the decision was made to vary the nanopar-
ticle concentration beyond the 5% value.
The nanouids used were colloidal suspensions of SiO
2
nano-
particles in water (Fig. 1). They were prepared from a commercial
solution (Ludox

TMA colloidal silica from SigmaAldrich) with a


mass fraction of 34%. Three mass fractions were used: 34%, and
after dilution in demineralised water, 16% and 5%. They correspond
to volume fractions of 18.93%, 7.95%, and 2.3% respectively. The
physical properties of SiO
2
particles are shown in Table 2.
Table 1
Bibliography on experimental forced convective heat transfer with nanouids.
Ref. Nanouid Re Nu
nf
/Nu
f
Lee and Choi (1996) Metallic nanoparticle suspension Laminar +100%
Pak and Cho (1998) Al
2
O
3
water TiO
2
water 3 vol.% Turbulent 3% to 12% for constant average velocity
Li and Xuan (2002) Cuwater 2 vol.% 80023,000 +60%
Xuan and Li (2003) Cuwater 0.32 vol.% Laminar and
turbulent
+30%
Wen and Ding (2004) Al
2
O
3
water 0.21.6% 6502050 Nu > Nu Shah especially near the entrance
Faulkner et al. (2004) MW CNT (aspect ratio > 100) 1.1
4.4 vol.%
217 +48% to +221% with high volume concentration
Yang et al. (2005) Graphite 22.5 wt.% 5110 Nu
nf
/Nu
f
< k
nf
/k
f
(aspect ratio l/d = 0.02)
Lee et al. (2005) Agwater 2.5 wt.% 10002000 +17% to +25%
Ding et al. (2006) CNTwater (aspect ratio > 100) 0.1
1 wt.%
8001200 +350%
Zeinali Heris et al.
(2006a)
CuOwater 0.23 vol.% 6502050 Enhancement of a with U and Pe
Zeinali Heris et al.
(2006b)
Al
2
O
3
water CuOwater 0.2 3 vol.% 6502500 Enhancement of a with u and Pe. Al
2
O
3
shows more enhancement than CuO
Zeinali Heris et al.
(2007)
Al
2
O
3
water 0.2-2.5 vol.% 7002050 Enhancement of a with u and Pe
Nguyen et al. (2007) Al
2
O
3
water 16.8 vol.% 300015,500 +40% enhancement of a with diameter decreases and U increases
Ding et al. (2007) Nano-diamond 0.1 wt.% ethylene-
based titanium 24 wt.%
135 No signicant enhancement
Chen et al. (2008) Titanate nanotubewater (aspect
ratio = 10) 0.52.5 wt.%
1700 a increases with aspect ratio (nanoparticle shape) increase
Williams et al. (2008) Al
2
O
3
water 0.93.6 vol.% ZrO
2

water 0.20.9 vol.%


900063,000 No abnormal heat transfer enhancement using measured properties of the
nanouid
Rea et al. (2009) Al
2
O
3
water 0.66.0 vol.% ZrO
2

water 0.323.5 vol.%


Laminar No abnormal heat transfer enhancement using measured properties of the
nanouid
He et al. (2007) TiO
2
water 0.242 vol.% 8006000 Enhancement of a with u for a given Re and particle size but no abnormal heat
transfer enhancement with particle size increase
Lai et al. (2008) Al
2
O
3
water 0.51.0 vol.% Laminar Enhancement of a with u and volume ow rate
Kulkarni et al. (2008) SiO
2
ethylene glycol/water 2
10 vol.%
300012,000 +16% enhancement of a with 10 vol.%, 20nm particle diameter at Re = 10000
enhancement of a with particle size increase
Jung et al. (2009) Al
2
O
3
water 0.61.8 vol.% 5300 +32% enhancement of a with 1.8 vol.% without major friction loss
Sommers and Yerkes
(2009)
Al
2
O
3
propanol 0.53 wt.% 18002800 Small but signicant enhancement for 1 wt.%
Yu et al. (2009) Silicon carbidewater 3.7 vol.% 330010,000 +5060% for a given Re, but 7% for constant average velocity
Hwang et al. (2009) Al
2
O
3
water 0.010.3 vol.% 550800 +8% at 0.3 vol.%
426 S. Ferrouillat et al. / International Journal of Heat and Fluid Flow 32 (2011) 424439
It is essential to use the correct thermal and physical properties
of nanouids, since all correlations depend on these properties. The
thermal and physical properties of interest are discussed below.
3.1. Density
The density of the nanouid is evaluated according to the stan-
dard formula:
q 1 uq
f
uq
s
1
where u is the volume fraction of the nanouid, q
f
the density of
the base uid, and
s
is the density of the nanoparticles.
3.2. Specic heat
The formula for the specic heat of a mixture is given by:
C
p
1 u
w
C
pf
u
w
C
ps
2
where u
w
is the mass fraction of the nanouid, C
p
f
the specic heat
capacity of the base uid, and C
p
s
the specic heat capacity of the
nanoparticles.
3.3. Thermal conductivity
Currently, there are no reliable theories to determine the effec-
tive conductivity of a owing nanouid. However, there exist
numerous theoretical studies for particle-uid mixtures based on the
pioneering work of Maxwells effective medium theory (Maxwell,
1881). These studies are essentially concerned by relatively large
particles (down to micrometric sizes). The effective thermal con-
ductivity k for a mixture with spherical particles is given by
k k
f
k
s
2k
f
2 k
f
k
s
_ _
u
k
s
2k
f
k
f
k
s
_ _
u
3
u is the volume fraction of the nanouid, k
f
the thermal conductiv-
ity of the base uid, and k
s
is the thermal conductivity of the
nanoparticles.
Hamilton and Crosser (1962), proposed a model for non-spher-
ical particles by introducing a shape factor b given by b = 3/w,
where w is the particle sphericity, dened as the ratio of the sur-
face area of a sphere with the same volume as that of the particle
and the surface area of the particle. The conductivity is expressed
as follows:
k k
f
k
s
b 1k
f
b 1 k
f
k
s
_ _
u
k
s
b 1k
f
k
f
k
s
_ _
u
4
The Maxwell formula corresponds to sphericity equals one. Sev-
eral authors have proposed other models to take into account
either the effects of the interface between the nanoparticle and
the base uid or several micro-convection phenomena. They will
not be evoked here.
The available experimental data on conductivity from different
research groups vary widely, and the proposed theories to explain
such dispersion vary fromone author to the other. It seems that the
different preparation methods and the different measurement
techniques of each research group contribute to this dispersion.
Due to these problems, the thermal conductivity of our nanouids
was measured using two techniques: The rst by using the hot
wire classical transient method with an industrial instrument
(Kd2 Pro) and the second by using a steady-sate method in a coax-
ial cylinder cell (Glory et al., 2008). The results from this instru-
ment were validated by using demineralised water. The values
obtained (shown in Fig. 2) were found to be close to the Maxwell
theory in accordance with the molecular dynamics simulation of
heat ow in a well-dispersed nanouid (Evans et al., 2006). For
the analysis of our experiments we used the conductivity values
obtained from measurements.
3.4. Dynamic viscosity
The viscosity of nanouids was measured using a MCR300 An-
ton PAAR rheometer as a function of temperature for the three
mass fractions. The results obtained were compared with current
theories.
The limiting case for dilute suspensions of small, rigid, spherical
particles was treated by Einstein (1906), and extended to ellipsoi-
dal particles. The viscosity is given by:
l l
f
1 Bu 5
where B depends on the ratio of the revolution ellipsoid axes and is
equal to 2.5 for spherical particles.
Measurements revealed that all the tested uids have a dy-
namic viscosity nearly constant from shear rates varying from
100 to 1000 s
1
. However, Einsteins formula does not allow us
to predict the experimental values.
Results obtained as a function of temperature are given in Fig. 3
for a shear rate of 1000 s
1
which is a mean value in our
experiments.
4. Experimental set-up and data reduction
4.1. Experimental set-up
4.1.1. Test loop and test section
A test loop was constructed to measure pressure loss and con-
vective heat transfer coefcients with xed wall temperature
boundary conditions. The experimental apparatus is shown sche-
matically in Fig. 4.
A 1 l copper vessel equipped with drain valves was used as
the nanouid reservoir. After injection in the reservoir tank,
the nanouid, with specied concentration, was circulated using
a gear pump (Micropump, Ismatec, 0200 l h
1
). Assuming that
Fig. 1. SEM picture of SiO
2
nanoparticles used (34 wt%)
Table 2
Physical properties of SiO
2
particles.
Nanoparticles Mean
diameter
(nm)
Density
(kgm
3
)
Thermal conductivity
at 25 C (W m
1
K)
Specic
heat at
25 C
(J/kg K)
SiO
2
22 2200 1.38 740
S. Ferrouillat et al. / International Journal of Heat and Fluid Flow 32 (2011) 424439 427
0.40
0.45
0.50
0.55
0.60
0.65
0.70
0.75
0.80
20 30 40 50 60 70 80
Temperature (C)
T
h
e
r
m
a
l

c
o
n
d
u
c
t
i
v
i
t
y

(
W
/
(
m
.
K
)
SiO2 5%w SS
SiO2 16%w SS
SiO2 34%w SS
SiO2 5%w THW
SiO2 34%w THW
Water
Maxwell 5%w
Maxwell 16%w
Maxwell 34%w
Fig. 2. Thermal conductivity versus temperature(SS: Steady State Method, THW: Transient Hot Wire Method).
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
10 20 30 40 50 60 70
Temperature (C)
V
i
s
c
o
s
i
t
y

(
c
P
)
SiO2/Water 5%w
SiO2/Water 16%w
SiO2/Water 34%w before destabilizing
SiO2/Water 34%w after destabilizing
Fig. 3. Viscosity versus temperature for SiO
2
/water nanouids.
Pump
Gear
pump
Nanofluid
reservoir
Test section
P
T
P
Mixer
T
Pump
Acquisition
System
T
T
Sewer
heat
exchanger
Heated bath
T
Coriolis
flow
T T T
T T T T
pH meter
Heated bath
Tap water
Fig. 4. Schematic diagram of the experimental loop.
428 S. Ferrouillat et al. / International Journal of Heat and Fluid Flow 32 (2011) 424439
nanouids are considered as homogeneous uids, the ow rate
was measured by a Coriolis ow meter (Micro Motion ELITE,
CFM10) that was calibrated with 0.1% accuracy over the range
of 080 kg h
1
. The pressure drop was measured directly by
three differential strain-gauge pressure transducers operating
over a range of 01620 kPa with uncertainty within 0.075% f.s.,
as calibrated by the manufacturer (Rosemount). A pH meter (Eu-
tech Instruments) was inserted downstream of the test section
to follow the nanouid pH change with a maximum accuracy
of 0.01.
The test section (Fig. 5) consisted of a 0.5 m long tube-in-tube
heat exchanger, the tested nanouid owing into the 4 mm diam-
eter and 1 mm thick inner copper tube (CuA1) and heating or cool-
ing water owing into a 10 mm diameter and 1 mm thick stainless
steel annular tube. The test section was preceded by a 0.5 m (125
diameters) adiabatic section.
The nanouid was circulated inside the inner tube (primary
loop) with a temperature varying between 15 and 90 C. To ob-
serve the potential inuence of the transverse temperature gradi-
ent, the water temperature was varied within the same range
allowing us to change the temperature difference between the
uid and the wall. The uid could be heated or cooled thanks to
various valves in the experimental loop, and then the gradient
direction could be modied. After passing through the test section,
the nanouid entered a heat exchanger in which water was used as
a cooling or heating uid depending on nanouid heating or cool-
ing tests. For both primary and secondary loops, temperature was
controlled using two thermostatic baths (Polystat

37, Fischer Sci-


entic) and a second heat exchanger.
The entire test section was insulated with polyurethane foam
(Armaex) in order to minimize heat losses.
A simplied test section of identical dimensions in which
only inlet and outlet uid temperatures were measured, was also
used in order to estimate the inuence of thermocouple
insertions on heat transfer (Section 4.1.2) and on pressure losses
(Section 5.1).
4.1.2. Temperature measurements in the test section
Two (K-type) thermocouples were inserted into the ow at the
inlet and outlet of the test section for measuring bulk tempera-
tures of nanouid. In order to increase the outlet temperature
accuracy for laminar ow, a static mixer was inserted down-
stream of the test section. To record the temperature at the
outer surface of the copper tube and the bulk temperature, four
(K-type) thermocouples were brazed on the inner tube wall and
four (K-type) thermocouples were inserted into the inner tube
at equally spaced 10 cm distances. The thermocouples were
calibrated before tests and had a maximum accuracy of 0.1 C.
All the data were recorded by an Agilent 34970 A data acquisition
unit.
To determine inner wall temperature, the thermal resistance
due to conduction through the tube (Fig. 5) was taken into account
(Eq. (8)). To determine inner ow bulk temperature we added a
corrective term by writing an energy balance between forced con-
vective ow perpendicular to the thermocouple and conduction in
the thermocouple between its extremity and the wall. The thermo-
couple is considered as a n of constant area, and if the heat loss
from its end is negligible, the temperature at the n tip is given
by the equation (Kaka et al., 1985):
T T
b
T
w
T
b

1
cosh dl
6
4 mm 6 mm 10 mm 12 mm
Copper
Copper
Stainless steel
Stainless steel
T
Wall
(K-type thermocouple)
T
Bulk
(K-type thermocouple)
Fig. 5. Schematic diagram of the test section with thermocouple locations.
S. Ferrouillat et al. / International Journal of Heat and Fluid Flow 32 (2011) 424439 429
where d is given by:
d

h
t
P
k
t
A
t

7
P is the perimeter of the thermocouple, k
t
is the average conductiv-
ity of the thermocouple (15 W m
1
K
1
) and A
t
, its cross section
area.
The heat transfer coefcient h
t
is calculated from the following
correlation developed by Churchill and Bernstein (1977):
Nu
D
0:3
0:62Re
1=2
D
Pr
1=3
1 0:4=Pr
2=3
_ _
1=4
1
Re
D
282000
_ _
5=8
_ _
4=5
8
where Nu
D
= h
t
D/k
t
, with D being the thermocouple diameter. This
formula holds for all values of Re
D
and Pr, provided the Peclet num-
ber Pe
D
= Re
D
Pr is greater than 0.2. As thermocouple temperatures
are close to bulk temperatures, physical properties in dimensionless
numbers are evaluated at T
b
.
4.2. Data reduction
The heat ow rate
_
Q was determined from the mass ow rate _ m
and the inlet and outlet temperatures of the uid:
_
Q _ mC
p
T
in
T
out
9
The internal heat transfer coefcient h
i
between the nanouid
and the wall was derived from the following expression of the heat
ow rate:
_
Q
1
1
h
i
R
w
_ _
S T
we
T
bi
10
where S is the heat exchange area (m
2
), T
we
the average external
wall temperature of the four K-type thermocouples brazed on the
inner tube (K), T
bi
the average internal bulk temperature of the four
K-type thermocouples inserted into the inner tube (K), and R
w
is the
thermal resistance of the copper tube wall (m K W
1
).
This thermal resistance R
w
is given by:
R
w

d
i
2k
w
ln
d
e
d
i
_ _
11
where d
i
and d
e
are respectively the inner and outer diameters of
the inner tube (m),
k
w
is the thermal conductivity of the inner tube (W m
1
K
1
).
The internal heat transfer coefcient h
i
(W m
2
K
1
) can thus be
calculated from
h
i

S T
we
T
bi

_
Q
R
w
_ _
1
12
Once the experimental heat transfer coefcient h
i
is deter-
mined, the experimental Nusselt number must be compared with
the value obtained experimentally with pure water, which is the
base uid. This comparison is done by plotting the ratio of the Nus-
selt number measured with the nanouid Nu
nf
and the Nusselt
number measured with pure water Nu
f
. In each case, the Reynolds
number was deduced from the mass ow rate measurement by:
Re
4 _ m
pd
i
l
13
where l is the measured uid dynamic viscosity taken at average
bulk temperature. Knowing the exact value of viscosity is crucial
because incorrect determination of the Reynolds number can cause
a shift in the curves and lead to misinterpretation of the Nu
nf
/Nu
f
ratio.
Using three differential strain-gauge pressure transducers, the
pressure drop measurement enables the Darcy coefcient to be de-
duced with the following expression:
K
exp
2DP
d
h
L
qS
2
pi
_ m
2
14
where the Darcy coefcient is 4 times larger than the fanning
friction factor f. The maximum relative uncertainties of the Rey-
nolds number and Darcy coefcient were estimated and are respec-
tively less than 1.9% and 4.3%. The maximum relative uncertainty of
the Nusselt number is highly dependent on the Reynolds number.
For example, this uncertainty is between 78% and 5% for a Reynolds
number between 100 and 12,000. Details of uncertainty calcula-
tions are given in Appendix A.
5. Results and discussion
5.1. Pressure drop
5.1.1. Preliminary tests
To be condent in the experimental loop and its instrumenta-
tion, the pressure drop of pure demineralised water owing
through the entire length of the copper tube was measured. Several
measurement conditions were studied as shown in Table 3. Fig. 6
shows experimental results in isothermal, heating and cooling
conditions at several temperature levels (20 C, 50 C, 70 C). In
the heating and cooling conditions, the temperature of one uid
was 20 C, the other being at 50 or 70 C. These results were
compared with classical relationships.
In laminar ow regime (Re < 2300), the following Poiseuille
equation is used in the calculations:
K
64
Re
15
In turbulent ow regime, the Blasius equation is used:
K 0:316Re
0:25
16
In heat transfer conditions, the Poiseuille and the Blasius laws
are followed provided that the experimental Darcy coefcient is
modied by using a corrective factor as indicated by Petukhov
(1970):
K K
exp
l
w
=l
_ _
m
17
where l
w
is the viscosity of the uid near the wall and l is the vis-
cosity of the bulk temperature.
The m exponent was experimentally found to be equal to the
following:
for heating conditions, m = 0.58 for laminar ow and
m = 0.25 in turbulent ow;
for cooling conditions, m = 0.50 for laminar ow and
m = 0.25 for turbulent ow.
Table 3
Measurement conditions.
Reference Measurement
conditions
Inlet internal
temperature (C)
Inlet external
temperature (C)
1 Isothermal 20 20
2 Isothermal 50 50
3 Isothermal 70 70
4 Heating 20 50
5 Heating 20 70
6 Cooling 50 20
7 Cooling 70 20
430 S. Ferrouillat et al. / International Journal of Heat and Fluid Flow 32 (2011) 424439
In laminar ow regime, the results followed Poiseuille classical
laws with demineralised water at Re < 1000. However, for
1000 < Re < 2300, the Poiseuille law underpredicts the experimen-
tal values. This deviation may be associated with the presence of
four (K-type) thermocouples inserted into the inner tube. These
thermocouples may generate turbulence for a Reynolds number
lower than 2300 and induce an increased pressure drop and thus
an increase of the Darcy coefcient. This was conrmed by com-
paring results with those obtained with the test section without
thermocouple insertions. Fig. 7 shows that transition between lam-
inar and turbulent regime takes place for Re $ 2000 as usual for
smooth tubes.
For Re > 2300, the Blasius law also slightly underpredicts the
Darcy coefcient measurement by approximately 10%. It should
be noted that the Colebrook correlation (Colebrook and White,
1965) with e = 20 lm absolute roughness seems to correspond bet-
ter with the results. Nevertheless, the above explanation requires
further experimental validation, through copper tube roughness
measurements for instance.
The Darcy coefcient with the Colebrook correlation is calcu-
lated with the following expression:
1

K
p 2log
10
2:51
Re

K
p 0:270e
_ _
18
5.1.2. Results with nanouids
As previously done for demineralised pure water, experiments
were carried out for different temperature conditions (isothermal;
heating and cooling conditions) with several nanouids for differ-
ent concentrations (5 wt.%, 16 wt.%, 34 wt.%).
0.01
0.1
1
10
10000 1000 100 10
Reynolds number
D
a
r
c
y

c
o
e
f
f
i
c
i
e
n
t
Poiseuille
Blasius
Colebrook 20 m
1
2
3
4
5
6
7
Fig. 6. Darcy coefcient as a function of the Reynolds number for several measurement conditions with water.
0.01
0.1
1
0 0 0 0 1 0 0 0 1 0 0 1
Reynolds number
D
a
r
c
y

c
o
e
f
f
i
c
i
e
n
t
Poiseuille
Blasius
Test section with thermocouples, 1
Test section without thermocouple, 1
Fig. 7. Comparison of Darcy coefcients for water owing in a smooth tube with those obtained with water owing in a tube with thermocouple insertions.
S. Ferrouillat et al. / International Journal of Heat and Fluid Flow 32 (2011) 424439 431
Fig. 8 shows the Darcy coefcient for water and several nano-
uids versus the Reynolds number. As for demineralised pure
water, using the measured viscosities, nanouid results correlate
quite well with Poiseuilles law for Re < 1000. Then, Poiseuilles
law underpredicts the measurements for 1000 < Re < 2300 be-
cause of inserted thermocouples. Finally, for Re > 2300, the Cole-
brook correlation seems to concord with experimental results
better than the Blasius law. It should be noted that the greatest
difference between experimental results and the Blasius law or
Colebrook correlation is observed for some results with the
34 wt.% nanouid. These differences with the Blasius law or
Colebrook correlation are respectively slightly higher than 25%
and 10%.
5.2. Heat transfer coefcient
5.2.1. Preliminary tests
As for pressure drop analyses, in order to be condent in the
experimental loop and its instrumentation, the heat transfer
coefcient of pure demineralised water was measured. Several
measurement conditions were studied, as shown in Table 3.
Fig. 9 shows the measured Nusselt number versus the predicted
Nusselt number calculated with the classical correlation of
Gnielinski valid for Re > 2300 in transition and turbulent regime
in heating and cooling conditions (measured conditions 4, 5, 6
and 7) (Gnielinski, 1976):
Nu
K=8Re 1000Pr
1 12:7

K=8
_
Pr
2=3
1
_ _
Pr
Pr
w
_ _
0:11
1
d
h
L
_ _
2=3
_ _
19
In this formula, the Darcy coefcient is given by K= (1.82log
10-
Re 1.64)
2
, where Re is the Reynolds number, Pr and Pr
w
are the
Prandtl numbers calculated at the water bulk temperature and at
the inner wall temperature respectively, L is the tube length and
d
h
the hydraulic diameter. The bulk temperature is an average be-
tween the inlet and outlet uid temperatures.
It can be seen that experimental data correspond well with the
predictions of the correlation to within 20%. It can be noted that
20% is also the Gnielinski range of validity.
0.01
0.1
1
10
10000 1000 100 10
Reynolds number
D
a
r
c
y

c
o
e
f
f
i
c
i
e
n
t
Poiseuille
Blasius
Colebrook 20 m
SiO2/Water 5%w (heating and cooling)
SiO2/Water 16%w (heating and cooling)
SiO2/Water 34%w (heating and cooling)
Water (heating and cooling)
Fig. 8. Darcy coefcient as a function of the Reynolds number for several measurement conditions with water and nanouids.
0
10
20
30
40
50
60
70
80
90
0 10 20 30 40 50 60 70 80 90
Predicted Nu
f
M
e
a
s
u
r
e
d

N
u
f
20%
-20%
Water heating (4)
Water heating (5)
Water cooling (6)
Water cooling (7)
Fig. 9. Tube averaged Nusselt number for water tests.
432 S. Ferrouillat et al. / International Journal of Heat and Fluid Flow 32 (2011) 424439
5.2.2. Results with nanouids
Figs. 1012 present the Nusselt number versus the Reynolds
number for the three mass fraction nanouids and pure deminer-
alised water for both heating and cooling conditions.
Signicant enhancement of the nanouid Nusselt number com-
pared to the base uid in the turbulent regime with nanouid con-
centration can be observed. There is a strong particle concentration
inuence in that the larger the mass fraction, the higher the
enhancement is.
Considering Fig. 10, these results can be divided into three
parts. A rst part, for Re > 1000, shows that the heat transfer is con-
trolled by turbulence regime ow. As previously observed on pres-
sure drop results, heat transfer results seemto showturbulent ow
regime development below the classical value (Re = 2300) due to
thermocouples inserted in the test section.
A second part, for 200 < Re < 1000, characterises the heat trans-
fer controlled by laminar regime ow.
A third part for Re < 200 shows a probable longitudinal con-
duction effect which implies a Nusselt number decrease (Bontemps,
2005). Indeed, the elementary theory of heat exchangers assumes
that heat is transferred locally from the hot to the cold uid
through an interposed solid wall which only acts as a thermal
resistance, while conduction along the wall is neglected. In real
equipment, however, heat transfer is actually a multi-dimensional
conjugate problem, in which heat conduction may play a role
not only in the direction orthogonal to the walls (transverse
conduction), but also in that parallel to them (longitudinal
conduction). As it can be observed in Fig. 10, a strong scattering
of data occurs for Re < 1000. The uncertainty on measured values
was calculated (see Appendix A) and relative errors vary from
78% to 5% from low to high Reynolds numbers. In Fig. 10, an
example of uncertainty for low Reynolds numbers is given. For
high Reynolds numbers, uncertainty is of the order of magnitude
of point sizes.
1
10
100
100000 10000 1000 100 10
Reynolds number
N
u
s
s
e
l
t

n
u
m
b
e
r
SiO2/Water 5%w (heating and cooling)
SiO2/Water 16%w (heating and cooling)
SiO2/Water 34%w (heatind and cooling)
Water (heating and cooling)
200
Part 3 Part 2 Part 1
Fig. 10. Nusselt number versus Reynolds number.
0
10
20
30
40
50
60
70
80
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
Reynolds number
N
u
s
s
e
l
t

n
u
m
b
e
r
Water (heating 4)
SiO2/Water 5%w (heating 4)
SiO2/Water 16%w (heating 4)
SiO2/Water 34%w (heating 4)
Water (heating 5)
SiO2/Water 5%w (heating 5)
SiO2/Water 16%w (heating 5)
SiO2/Water 16%w (heating 5)
Fig. 11. Nusselt number versus Reynolds number for heating condition.
S. Ferrouillat et al. / International Journal of Heat and Fluid Flow 32 (2011) 424439 433
Taking into account the measurement accuracy, it is difcult
to know whether or not some heat transfer intensication/dete-
rioration occurs in laminar regime. For the following analyses,
results below Re < 1000 will not be considered. In Figs. 11 and
12, results in linear scales for heating and cooling conditions
are indicated respectively. It is clearly seen that in turbulent re-
gime and compared to pure water, heat transfer enhancement
occurs when using nanouids. In order to quantify the possible
intensication of the heat transfer coefcient due to nanouids
compared to that of the base uid (pure demineralised water),
the ratio Nu
nf
/Nu
f
versus the Reynolds number is given in
Fig. 13 (heating and cooling conditions 4, 5, 6 and 7). For
Re > 1000, whatever the measurement conditions, signicant
heat transfer enhancement with nanouid concentration up to
+50% with 34 wt.% nanouid can be observed.
The temperature gradient direction effect (comparison of heat-
ing and cooling condition) must be analysed with caution. Heating
and cooling conditions were not realized at the same bulk temper-
ature. Nevertheless, in most cases, the cooling condition seems to
lead to equal or better heat transfer performance characteristics
than the heating condition.
To compare theses results with the prediction of classical corre-
lation, Fig. 14 shows the experimental ratio Nu
nf
/Nu
f
versus the
predicted ratio Nu
nf
/Nu
f
calculated with the classical correlation
of Gnielinski valid for Re > 2300 in transition and turbulent regime.
It can be observed that heat transfer performance characteristics
for all nanouids concentrations are predicted by the Gnielinski
correlation within 20% if the nanouid mixture properties are ta-
ken into account.
5.3. Nanouid stability
We have seen in the selected bibliography that heat transfer
enhancement in various nanouids has been attributed to different
0
10
20
30
40
50
60
70
80
90
0 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000
Reynolds number
N
u
s
s
e
l
t

n
u
m
b
e
r
Water (cooling 6)
SiO2/Water 5%w (cooling 6)
SiO2/Water 16%w (cooling 6)
SiO2/Water 34%w (cooling 6)
Water (cooling 7)
SiO2/Water 5%w (cooling 7)
SiO2/Water 16%w (cooling 7)
SiO2/Water 34%w (cooling 7)
Fig. 12. Nusselt number versus Reynolds number for cooling condition.
0%
20%
40%
60%
80%
100%
120%
140%
160%
0 2000 4000 6000 8000 10000 12000 14000
Reynolds number
N
u
n
f

/

N
u
f
5%w heating (4) 5%w heating (5) 5%w cooling (6) 5%w cooling (7)
16%w heating (4) 16%w heating (5) 16%w cooling (6) 16%w cooling (7)
34%w heating (4) 34%w heating (5) 34%w cooling (6) 34%w cooling (7)
Fig. 13. Nu
nf
/Nu
f
versus Reynolds number.
434 S. Ferrouillat et al. / International Journal of Heat and Fluid Flow 32 (2011) 424439
mechanisms. There have recently been further discussions that
point to particle coatings on heat transfer surfaces as being impor-
tant (Yu et al., 2009) and nanouid discoloration (Sommers and
Yerkes, 2009).
Studying the thermal performance of the nanouid at the Rey-
nolds number was difcult due to the increase in uid viscosity
and the limitations imposed by the pump. However, in order to
collect results at higher Reynolds number for 34 wt.% nanouids,
some tests were carried out by increasing the nanouid inlet tem-
perature to higher than 80 C to reduce the apparent viscosity of
the nanouid. These measurement conditions show the 34 wt.%
nanouid to be destabilized, leading to a modication of hydraulic
performance. Fig. 15 shows the Darcy coefcient increase after the
nanouid destabilization, by 1.5 in laminar ow and 3.0 in turbu-
lent ow. According to Ludox product information, most Ludox
applications involve the use of sols at room temperature, thereby
minimizing concentration by evaporation. Higher temperatures
not only increase the loss of water by evaporation but also the
movement of the colloidal particles in suspension and the dissoci-
ation of electrolytes present in the system and surfactants avoiding
particle agglomeration. Each of these factors contributes to gela-
tion or formation of silica aggregates.
To validate the hypothesis of the formation of silica aggregates,
after emptying the test section, other tests with pure deminera-
lised water were conducted. Fig. 16 also shows an increase of the
Darcy coefcient after nanouid destabilizing. These results seem
to indicate that 34 wt.% nanouid destabilization leads to a coating
of the test surface.
Another test for assessing uid stability involved analysing uid
samples after different periods of time and at different tempera-
ture conditions. The 34 wt.% nanouid was stored in two different
sealed glass beakers for two weeks. The rst one was heated to
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
Predicted Nu
nf
/Nu
f
M
e
a
s
u
r
e
d

N
u
n
f
/
N
u
f

5%w heating (4 and 5)
5%w cooling (6 and 7)
16%w heating (4 and 5)
16%w cooling (6 and 7)
34%w heating (4 and 5)
34%w cooling (6 and 7)
20%
-20%
Fig. 14. Measured (Nu
nf
/Nu
f
) versus predicted (Nu
nf
/Nu
f
).
0.01
0.1
1
10
10000 1000 100 10
Reynolds number
D
a
r
c
y

c
o
e
f
f
i
c
i
e
n
t
Poiseuille
Blasius
Colebrook 20 m
Before destabilizing
After destabilizing
Fig. 15. Darcy coefcient evolution after 34%w nanouid destabilizing.
S. Ferrouillat et al. / International Journal of Heat and Fluid Flow 32 (2011) 424439 435
88 C for 24 h, the second was left at room temperature. It was
observed that the heated nanouid showed some particle
aggregates with sediment on the beaker bottom conrming the
temperature effect. This type of effect was already observed with
Al
2
O
3
nanoparticles in water by Lee and Mudawar (2007), who
observed sedimentation in long-term use and by Nguyen et al.
(2008), who found a critical temperature beyond which the
particle suspension properties seem to be drastically altered.
Moreover, viscosity measurements of the 34 wt.% nanouid
after destabilization show a decrease which conrms a formation
of silica aggregates which lead to sedimentation and thus to
coating. By interpolation, we deduce from Fig. 3 a new concen-
tration at roughly 31 wt.%. With this new mass concentration,
we deduce the other new thermal and physical properties of
nanouids.
From results after nanouid destabilization presented in Fig. 16,
we have estimated a new hydraulic diameter of the test section in
laminar ow regime thanks the following equation deduced from
Eqs. (13) and (14):
d
h

128L
p
:
lm
:
qDP
exp
_ _1
4
20
The new hydraulic diameter is roughly 3.5 mm instead of
4.0 mm.
The results after destabilization presented in the Fig. 15 were
modied to take into account the new thermal and physical
properties and hydraulic diameter. The new results are presented
in Fig. 17. It can be observed that in laminar ow regime, the re-
sults follow Poiseuille classical laws up toRe < 1000. But for
1000 < Re < 2300, the Poiseuille law underpredicts the measure-
ments. This deviation may also be associated with the presence
of the four (K-type) thermocouples inserted into the inner tube.
For Re > 2300, the Blasius law underpredicts the Darcy
coefcient measurement. It should be noted that the Colebrook
0.01
0.1
1
10
10000 1000 100 10
Reynolds number
D
a
r
c
y

c
o
e
f
f
i
c
i
e
n
t
Poiseuille
Blasius
Colebrook 20 m
1 before destabilizing
1 after destabilizing
4 before destabilizing
4 after destabilizing
7 before destabilizing
7 after destabilizing
Fig. 16. Darcy coefcient evolution after 34%w nanouid destabilizing with pure demineralised water.
0.01
0.1
1
10
10000 1000 100 10
Reynolds number
D
a
r
c
y

c
o
e
f
f
i
c
i
e
n
t
Poiseuille
Blasius
Colebrook 20 m
Colebrook 250 m
7 before destabilizing
7 after destabilizing
7 after destabilizing with new properties and hydraulic diameter
Fig. 17. Darcy coefcient evolution after 34%w nanouid destabilizing with new physical properties and hydraulic diameter.
436 S. Ferrouillat et al. / International Journal of Heat and Fluid Flow 32 (2011) 424439
correlation with e = 250 lm absolute roughness seems to concord
better with the results. This absolute roughness corresponds well
with the estimation of the new hydraulic diameter. Nevertheless,
the above explanation requires further experimental validation
through copper tube roughness measurements, for instance.
5.4. Energetic performance evaluation criterion (PEC)
In general, the results on SiO
2
/water nanouids are apparently
attractive in terms of their thermal performance. Nevertheless,
an important point must be discussed. Indeed, the strong increase
of dynamic (and kinematic) viscosity of nanouids inevitably in-
volves an increase of the pressure losses inside the system. Conse-
quently, even if a heat transfer enhancement is observed, the
required power for the pumping is increased compared to the base
uid. This is the reason why a signicant increase of viscosity may
lead to an unfavourable energetic balance. There are several ways
to characterise the energetic or thermal performance of a uid
owing in a specic device (Colburn, AP, 1933) (Sahiti et al.,
2006). We can use the PEC (performance evaluation criterion) de-
ned below and based on an energetic global approach.
It is dened as the ratio of heat transferred to the required
pumping power in the test section:
PEC
_ m C
p
T
out
T
in

_ v DP
21
where _ m is the mass ow rate (kg/s), _ v the volumic ow rate
(m
3
/s), T
in
and T
out
the tube inlet and outlet temperatures and DP
the pressure drop (Pa).
Fig. 18 shows the evolution of this energetic criterion with the
Reynolds number in the case of water and of the SiO
2
/water nano-
uids. We can notice that all measurements lead to PEC values be-
neath those corresponding to the case of water, which means that
the energy budget is unfavourable.
For specic applications in which the energetic cost is not
important, the use of such nanouids could be relevant.
6. Conclusions
As seen in the literature survey, heat transfer coefcient
enhancement was generally observed in using nanouids. However,
in a recent experiment it was shown that by using measured
thermophysical properties, the heat transfer in forced convection
experiments with nanouids was quite similar to that of conven-
tional uids (Williams et al., 2008). To nd possible effects in con-
vection to interpret the literature data we performed experiments
with several conditions: mass fraction varying from a small value
corresponding to that of a nanouid to a high value, different wall
boundary conditions, imposed ux and, as described in this paper,
imposed temperature.
In this paper, the convective heat transfer of colloidal suspen-
sion of SiO
2
nanoparticles in water was studied experimentally.
The owregime was varied fromlaminar to turbulent and constant
wall temperature was considered as a thermal boundary condition.
Both ow cooling and ow heating were studied. Results have
shown that with the presence of nanoparticles, heat transfer of
the resulting nanouid signicantly increases compared to the
base uid (water) in turbulent regime. Such an enhancement has
been found more pronounced with the increase of particle concen-
tration. The Nusselt number increases from 10% to about 50% when
volume concentration varies from 2.3% to 18.93%.
It is shown that, if the measured thermal and physical proper-
ties of the nanouid were taken into account to calculate the
dimensionless numbers, the existing correlations reproduce the
convective heat transfer and pressure loss behaviour in tubes with-
in the correlations range of validity. Therefore, the merits of nano-
uids for heat transfer enhancement depend on the compromise
between thermal conductivity increase and viscosity increase. In
this objective, a Performance Evaluation Criterion (PEC) was de-
ned, which indicates that the global energy budget is not favour-
able to the studied nanouid.
Finally, this paper examines nanouid destabilizing with high
temperature. A temperature inuence on uid stability was
observed.
Based on these results, the use of nanouids seems to remain
suitable for applications in which an increase in pumping power
is not of great concern. Nevertheless, nanouid stability must be
carefully studied for an industrial application.
Acknowledgments
This work was partially supported by the Programme Interdis-
ciplinaire Energie Microcond of the CNRS (National Scientic
Research Centre) and the Environment and Energy Management
Agency (ADEME) under grant No. 0566C00.
1000
10000
100000
1000000
10000000
100000 10000 1000 100
Reynolds number
P
E
C
Water
SiO2/Water 5%w
SiO2/Water 16%w
SiO2/Water 34%w
Fig. 18. Evolution of the PEC versus Reynolds number in cooling condition (6).
S. Ferrouillat et al. / International Journal of Heat and Fluid Flow 32 (2011) 424439 437
The authors thank Olivier PONCELET from CEA L2T for his work
preparing nanouids and Marco BONETTI from CEA IRAMIS for
thermal conductivity measurements.
Appendix A
The internal heat transfer coefcient is given by:
h
i

1
S TweT
bi

_
Q
R
w
If U
i

S TweT
bi

_
Q

S TweT
bi

_ mCpT
in
Tout
; the uncertainty of the internal heat
transfer coefcient h
i
was determined using Moffats method:
dh
i

@h
i
@U
i
dU
i
_ _
2

@h
i
@R
w
dR
w
_ _
2

where
@h
i
@U
i

1
U
2
i
1
U
i
R
w
_ _
2

h
i
U
i
_ _
2
and
@h
i
@Rw

1
U
i
R
w
_ _
2
h
2
i
In the same way, the uncertainty of U
i
is written as:
where
@U
i
@ _ m

Cp T
in
Tout j j
TweT
bi
j j S

U
i
_ m
@U
i
@S

_ mC
p
T
in
T
out
j j
T
we
T
bi
j j S
2

U
i
S
@U
i
@C
p

_ m T
in
T
out
j j
T
we
T
bi
j j S
2

U
i
C
p
@U
i
@ T
we
T
bi


_ mC
p
T
in
T
out
j j
T
we
T
bi
j j
2
S

U
i
T
we
T
bi
j j
@U
i
@ T
in
T
out

_ mC
p
T
we
T
bi
j j S

U
i
T
in
T
out
j j
The internal heat transfer area S is given by: S = pd
i
L
Thus, the uncertainty of S is written as:
dS

@S
@d
i
dd
i
_ _
2

@S
@L
dL
_ _
2

where
@S
@d
i
pL and
@S
@L
pd
i
The thermal resistance of the copper tube wall R
w
is given by:
R
w

d
i
2k
Cu
ln
d
e
d
i
_ _
In the same way, the uncertainty of the thermal resistance R
c
is
written as:
dR
w

@R
w
@d
i
dd
i
_ _
2

@R
w
@d
e
dd
e
_ _
2

where
@Rw
@d
i

1
2k
Cu
ln
de
d
i
1
_ _
and
@Rw
@de

1
2k
Cu
d
i
de
References
Bontemps, A., 2005. Measurements of single-phase pressure drop and heat transfer
coefcient in micro and minichannels in Microscale He at transfer:
fundamentals and applications, NATO science series, II: mathematics. Physics
and Chemistry 193, 2548.
Bontemps, A., Ribeiro, J.P., Ferrouillat, S., Gruss, J.A., Soriano, O., Wang, B., 2008a.
Experimental study of convective heat transfer and pressure loss of SiO
2
/water
nanouids. Part 1: nanouid characterization imposed wall temperature.
Thermal Issues in Emerging Technologies, ThETA 2, Cairo, Egypt, 1720th
December, pp. 275284.
Bontemps, A., Ribeiro, J.P., Ferrouillat, S., Gruss, J.A., Soriano, O., Wang, B., 2008b.
Experimental study of convective heat transfer and pressure loss of SiO2/water
nanouids. Part 2: imposed uniform heat ux, energetic performance criterion.
Thermal Issues in Emerging Technologies, ThETA 2, Cairo, Egypt, 1720th
December, 2008, pp. 285292.
Chen, H., Yang, W., Ding, Y., Zhang, L., Tan, C., Lapkin, A.A., Bavykin, D.V., 2008. Heat
transfer and ow behaviour of suspensions of titanate nanotubes. Powder
Technology 183, 6372.
Churchill, S.W., Bernstein, M., 1977. A correlating equation for forced convection
from gases and liquids to a circular cylinder in cross-ow. Journal of Heat
Transfer 99, 300306.
Colebrook, C.F., White, C.M., 1965. Experiments with uid friction-factor equations.
Chemical Engineering 29, 8687.
Ding, Y., Alias, H., Wen, D., Williams, R.A., 2006. Heat transfer of aqueous
suspensions of carbon nanotubes (CNT nanouids). International Journal of
Heat and Mass Transfer 49, 240250.
Ding, Y., Chen, H., He, Y., Lapkin, A., Yeganeh, M., Siller, L., Butenko, Y.V., 2007.
Forced convective heat transfer of nanouids. Advanced Powder Technology 18
(6), 813824.
Einstein, A., 1906. Eine neue Bestimmung der Molekldimensionen (a new
determination of molecular dimensions). Annalen der Physik 19, 289306.
Evans, W., Fish, J., Keblinski, P., 2006. Role of Brownian motion hydrodynamics on
nanouid thermal conductivity. Applied Physics Letters 88, 093116-1093116-
3.
Faulkner, D.J., Rector, D.R., Davidson, J., Shekarriz, R., 2004, Enhanced heat transfer
through the use of nanouids in forced convection. In: Proceedings of IMECE
2004. Anaheim, California, USA.
Glory, J., Bonetti, M., Helezen, M., Mayne-LHermite, M., Reynaud, C., 2008. Thermal
and electrical conductivity of water-based nanouids prepared with long multi-
walled carbon nanotubes. Journal of Applied Physics 103, 094309.
Gnielinski, V., 1976. New equations for heat and mass transfer in turbulent pipe and
channel ow (translated from German). International Chemical Engineering 16
(2), 359368.
Hamilton, R.L., Crosser, O.K., 1962. Thermal conductivity of heterogeneous two-
component systems. Industrial & Engineering Chemistry Fundamentals 1 (3),
187191.
He, Y., Jin, Y., Chen, H., Ding, Y., Cang, D., Lu, H., 2007. Heat transfer and ow
behaviour of aqueous suspensions of TiO
2
nanoparticles (nanouids) owing
upward through a vertical pipe. International Journal of Heat and Mass Transfer
50, 22722281.
Hwang, K.S., Jang, S.P., Choi, S.U.S., 2009. Flow and convective heat transfer
characteristics of water-based Al
2
O
3
nanouids in fully developed laminar ow
regime. International Journal of Heat and Mass Transfer 52, 193199.
Jung, J.Y., Oh, H.S., Kwak, H.Y., 2009. Forced convective heat transfer of nanouids in
microchannels. International Journal of Heat and Mass Transfer 52, 466472.
Kaka, S., Yener, Y., 1985. Heat Conduction. Hemisphere Publishing Corporation,
Washington.
Kulkarni, D.P., Namburu, P.K., Bargar, H.E., Das, D.K., 2008. Convective heat transfer
and uid dynamic characteristics of SiO
2
ethylene-glycol/water nanouid. Heat
Transfer Engineering 29 (12), 10271035.
Lai, W.Y., Phelan, P.E., Vinod, S., 2008, Convective heat transfer for water-based
alumina nanouids in single 1.02 mm tube. In: 11th IEEE Intersociety
Conference on Thermal and Thermomechanical Phenomena in Electronic
Systems, vols. 13, Intersociety Conference on Thermal and
Thermomechanical Phenomena in Electronic Systems, pp. 970978.
Lee, C.H., Kang, S.-W., Kim, S.H., 2005. Effects of nano-sized Ag particles on heat
transfer of nanouids. Journal of Industrial Engineering Chemistry 11 (1), 152
158.
Lee, S., Choi, S.U.S., 1996, Application of metallic nanoparticle suspensions in
advanced cooling systems. In: International Mechanical Engineering Congress
and Exhibition. Atlanta, USA.
Lee, J., Mudawar, I., 2007. Assessment of the effectiveness of nanouids for single-
phase and two-phase heat transfer in micro-channels. International Journal of
Heat and Mass Transfer 50, 452463.
Li, Q., Xuan, Y., 2002. Convective heat transfer and ow characteristics of Cuwater
nanouid. Science in China Series E 45, 408416.
Maxwell, J.C., 1881, 2nd ed. A Treatise on Electricity and Magnetism, vol. 1
Clarendon press, Oxford.
Nguyen, C.T., Roy, G., Gauthier, C., Galanis, N., 2007. Heat transfer enhancement
using Al
2
O
3
water nanouid for an electronic liquid cooling system. Applied
Thermal Engineering 27, 15011506.
dU
i

@U
i
@ _ m
d _ m
_ _
2

@U
i
@S
dS
_ _
2

@U
i
@C
p
dC
p
_ _
2

@U
i
@ T
we
T
bi

d T
we
T
bi

_ _
2

@U
i
@ T
in
T
out

d T
in
T
out

_ _
2

438 S. Ferrouillat et al. / International Journal of Heat and Fluid Flow 32 (2011) 424439
Nguyen, C.T., Desgranges, F., Galanis, C.N., Roy, G., Mar, T., Boucher, S., Angue
Mintsa, H., 2008. Viscosity data for Al
2
O
3
water nanouid hysteresis: is heat
transfer enhancement using nanouid reliable? International Journal of
Thermal Science 47, 103111.
Pak, B.C., Cho, Y. Lee, 1998. Hydrodynamic and heat transfer study of dispersed
uids with submicronic metallic oxide particles. Experimental Heat Transfer 11,
151170.
Petukhov, B.S., 1970, Heat transfer and friction in turbulent pipe ow with
variable physical properties. In: Irvine, T.F., Hartnett, J.P. (Eds.), Advances in
Heat Transfer Vol. 6. New York, pp. 503564.
Rea, U., McKrell, T., Hu, L.W., Buongiorno, J., 2009. Laminar convective heat transfer
and viscous pressure loss of aluminawater and circoniawater nanouids.
International Journal of Heat and Mass Transfer 52, 20422048.
Sahiti, N., Lemouedda, A., Stojkovic, D., Durst, F., Franz, E., 2006. Performance
comparison of pin n in-duct ow arrays with various pin cross-sections.
Applied Thermal Engineering 26 (1112), 11761192.
Sommers, A.D., Yerkes, K.L., 2009, Experimental investigation into the convective
heat transfer and system-level effects of Al
2
O
3
propanol nanouid. Journal of
Nanoparticle Research (Published on line).
Wen, D., Ding, Y., 2004. Experimental investigation into convective heat transfer of
nanouids at the entrance region under laminar ow conditions. International
Journal of Heat and Mass Transfer 47 (24), 51815188.
Williams, W., Buongiorno, J., Hu, L.-W., 2008. Experimental Investigation of
turbulent convective heat transfer and pressure loss of alumina/water and
zirconia/water nanoparticle colloids (nanouids) in horizontal tubes.
Transactions on ASME, Journal of Heat Transfer 130, 042412-1042412-7.
Xuan, Y., Li, Q., 2003. Investigation on convective heat transfer and ow
features of nanouids. Transactions on ASME, Journal of Heat Transfer 125,
151155.
Yang, Y., Zhang, Z.G., Grulke, E.A., Anderson, W.B., Wu, G., 2005. Heat transfer
properties of nanoparticles-in-uid dispersions (nanouids) in laminar ow.
International Journal of Heat and Mass Transfer 48 (6), 11071116.
Yu, W., France, D.M., Smith, D.S., Singh, D., Timofeeva, E.V., Routbort, J.L., 2009. Heat
transfer to a silicon carbide/water nanouid. International Journal of He at and
Mass Transfer 52, 36063612.
Zeinali Heris, S., Nasr Esfahany, M., Eternad, S.G., 2006a. Investigation of CuO/water
nanouid laminar convective heat transfer through a circular tube. Journal
Enhanced Heat transfer 13 (4), 279289.
Zeinali Heris, S., Nasr Esfahany, M., Eternad, S.G., 2006b. Experimental investigation
of oxide nanouids laminar ow convective heat transfer. International
Communications in Heat and Mass Transfer 33 (4), 529535.
Zeinali Heris, S., Nasr Esfahany, M., Eternad, S.G., 2007. Experimental investigation
of convective heat transfer of Al
2
O
3
/water nanouid in circular tube.
nternational Journal of Heat and Fluid Flow 28, 203210.
S. Ferrouillat et al. / International Journal of Heat and Fluid Flow 32 (2011) 424439 439

Você também pode gostar