International Journal of Heat and Mass Transfer 43 (2000) 37013707

www.elsevier.com/locate/ijhmt

Conceptions for heat transfer correlation of nanouids

Yimin Xuan a,*, Wilfried Roetzel b
a

School of Power Engineering, Nanjing University of Science and Technology, Nanjing 210094, People's Republic of China
b
Institut fuer Thermodynamik, Unversitaet der Bunderwehr Hamburg, D-22039 Hamburg, Germany

Abstract
The nanouid is a solidliquid mixture in which metallic or nonmetallic nanoparticles are suspended. The
suspended ultrane particles change transport properties and heat transfer performance of the nanouid, which
exhibits a great potential in enhancing heat transfer. The mechanism of heat transfer enhancement of the nanouid
is investigated. Based on the assumption that the nanouid behaves more like a uid rather than a conventional
soliduid mixture, this article proposes two dierent approaches for deriving heat transfer correlation of the
nanouid. The eects of transport properties of the nanouid and thermal dispersion are included. 7 2000 Elsevier
Science Ltd. All rights reserved.
Keywords: Enhancement; Heat transfer; Nanouid

1. Introduction
The term ``nanouid'' is envisioned to describe a
solidliquid mixture which consists of nanoparticles
and a base liquid, and which is one of the new challenges for thermo-science provided by the nano-technology. There are several approaches to prepare
nanouids. Some auxiliary activators or dispersants
are necessary to obtain the even distributed and stabilized suspensions. The ultrane particles may be either
metallic or nonmetallic. In general, the nanouids used
for the purpose of enhanced heat transfer are dilute
multicomponent uids and the volume fractions of
nanoparticles are below 510%. The nanoparticles suspended in the nanouids can be the nanostructured

* Corresponding author. Tel.: +86-25-43-15700; fax: +8625-44-31339.

E-mail addresses: [email protected] (Y. Xuan),
[email protected] (W. Roetzel).

materials below 100 nm in diameter. Heat transfer performance of the nanouid is superior to that of the
original pure uid because the suspended ultrane particles remarkably increase the thermal conductivity of
the mixture and improve its capability of energy
exchange. Several literature [13] reveal that with low
nanoparticles concentrations (15 vol %), the thermal
conductivity of the suspensions can increase by more
than 20%. Such enhancement mainly depends upon
factors such as the shape of particles, the dimensions
of particles, the volume fractions of particles in the
suspensions, and the thermal properties of particle materials.
The nanouids have a unique feature which is
quite dierent from those of the conventional solid
liquid mixtures in which millimeter and/or micrometer-sized particles are added. As we know, the
particles of millimeter or micrometer magnitudes
settle rapidly, clog ow channels, erode pipelines,
and cause severe pressure drops, etc. All these
shortages limits the application of the conventional

0017-9310/00/$ - see front matter 7 2000 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 1 7 - 9 3 1 0 ( 9 9 ) 0 0 3 6 9 - 5

3702

Y. Xuan, W. Roetzel / Int. J. Heat Mass Transfer 43 (2000) 37013707

Nomenclature
cp
dp
h
k
Nu
Pr
R
Re
q
T
u
V
r
x

specic heat capacity

diameter of particles
heat transfer coecient
thermal conductivity
Nusselt number
Prandtl number
radius of a tube
Reynolds number
heat ux
temperature
velocity
volume
radial variable
axial variable

solidliquid mixture. The concept of the nanouid

is an innovative idea [4], although the idea about
solidliquid suspensions appeared a long time ago.
Modern technology makes it possible to produce a
variety of stabilized nanouids. Because of their
excellent characteristics, the nanouids nd wide applications in enhancing heat transfer, even for
microscale heat transfer [5]. It is expected that the
nanouid will become a new type of heat transfer
uid for thermal engineering.
Compared with the existing techniques for enhancing heat transfer, the nanouids show a great
potential in increasing heat transfer rates in a variety of application cases, with incurring either little
or no penalty in pressure drop. Although the nanouids have great potential for enhancing heat transfer, research work on the concept, enhancement
mechanism, and application of the nanouids is still
in the primary stage. Several existing published
articles are mainly focused on prediction and
measurement techniques of thermal conductivity of
the nanouids. To our knowledge, there are only
few references involved in describing heat transfer
performance of the nanouids. A complete understanding about the heat transfer performance of the
nanouids is necessary for their practical application
to heat transfer enhancement. In nature, the nanouids are multicomponent uids, even the dispersed
particles are ultrane. Three fundamental motions of
multicomponent materials, sedimentation, shearing
ow, and wave motion, may coexist in ow of the
nanouids. It is dicult to establish any formulated
theory that could predict the ow of a dispersed
multicomponent material [6]. Such an approach may

a
f
m
r

thermal diusivity
volume fraction of the nanoparticles
viscosity
density

Subscripts
d
f
m
nf
r
s
x

dispersion eect
uid
mean value
nanouid
radial direction
solid
axial direction

also be impracticable for practical application of the

nanouids to heat transfer enhancement.
It is expected that heat transfer coecient (Nusselt
number) of the nanouid depends upon a number of
factors such as thermal conductivity and heat capacity
of both the base uid and the ultrane particles, the
ow pattern, the viscosity of the nanouid, the volume
fraction of the suspended particles, the dimensions and
the shape of these particles as well as on the ow
structure. Therefore, the general form of the Nusselt
number yields
Nunf f Re, Pr,

ks rcp s
,
, f,
kf rcp f

dimensions and shape of particles, flow structure

!
1

So far we have found no published literature on deriving the convective heat transfer correlation of the
nanouid from either theoretical or experimental
approach.
The purpose of this article is to analyze heat transfer
performance of the nanouids and to derive some fundamental correlations for predicting convective heat
transfer of the nanouids with two dierent
approaches. One is the conventional way to treat the
nanouids as the single-phase uids and another is to
account for the multiphase feature of the nanouid
and the dispersed nanoparticles.
2. The conventional approach
From various sources, one may nd a variety of
convective heat transfer correlations of single-phase

Y. Xuan, W. Roetzel / Int. J. Heat Mass Transfer 43 (2000) 37013707

ow for dierent application cases. Whether these relations are applicable for predicting heat transfer performance of the nanouids invokes a researchers'
interest. If the answer is positive, it will be convenient
to extend the available correlations of the conventional
single-phase uids to the corresponding applications of
the nanouids since there are no published correlations
for the nanouid application. On certain assumptions,
such extension may be feasible. The nanouids used
for the purpose of heat transfer enhancement are
usually dilute solidliquid mixtures. Since the solid
particles are ultrane (< 100 nm) and they are easily
uidized, these particles can be approximately considered to behave like a uid. Under the assumptions
that there exist no motion slip between the discontinuous phase of the dispersed ultrane particles and the
continuous liquid and the local thermal equilibrium
between the nanoparticles and the uid, the nanouid
can be treated as the common pure uid. All the
equations of continuity, motion, and energy for the
pure uid are directly extended to the nanouid.
Under the assumption of constant thermal properties,
for example, the energy equation for the incompressible ow of a pure uid without viscous dissipation
@T
r
uT r
af rT
@t

is also suitable to describe the heat transfer process of

the nanouid. It means that the solutions for the
single-phase uid is also valid for the nanouid in the
identical application cases. However, it must be
emphasized that the thermal properties appearing in
Eq. (2) refers to those of the nanouid.
Thus, the dimensionless correlations of heat transfer
of the pure uid are applicable for the nanouid. For
example, the following relations
Nu 3:66
fully developed laminar flow, tube, Tw const
Nu 0:023Re0:8 Pr1=3

turbulent flow, tube

3a

3b

can be used to calculate heat transfer rates of the

nanouid owing in a tube in the laminar region and
turbulent region, respectively. A similar approach was
introduced by Nield and Bejan [7] for comparing the
dimensionless expressions of fully developed heat
transfer in a channel without and with a porous
matrix. However, one must pay attention in selecting
the suitable thermal properties and transport properties. While applying the existing dimensionless relations
for pure uids to the nanouid, one needs to use the
properties corresponding to the nanouid. Three main
parameters involved in calculating heat transfer rate of

3703

the nanouid are heat capacity, viscosity, and thermal

conductivity, which may be quite dierent from those
of the original pure uid. For the synthesized nanoparticleliquid suspension, the parameter rcp nf of the
nanouid is expressed as
rcp nf 1 frcp f frcp s

The viscosity of the nanouid can be estimated with

the existing relations for the two-phase mixture. Drew
and Passman [6] introduced the well-known Einstein's
formula for evaluating the eective viscosity meff of a
linearly viscous uid of viscosity mf containing a dilute
suspension of small rigid spherical particles. The formula yields:
meff mf 1 2:5f

This relation is restricted for low volume concentration

f < 0:05: Einstein's equation was extended by Brinkman [8] as
meff mf

1
1 f 2:5

One may nd other relations of the eective viscosity

of the two-phase mixture in the literature [6,9]. Each
relation has its own application limitation. The direct
and reliable access to obtaining the apparent viscosity
of the nanouid is by experiment. Xuan and Li [10]
have experimentally measured the apparent viscosity of
the transform oilwater nanouid and of the water
copper nanouid in the temperature range of 20508C.
The experimental results reveal relatively good coincidence with Brinkman's theory.
The apparent thermal conductivity is the most important parameter to indicate the enhancement potential of the nanoparticleliquid suspension. Research
has shown that the thermal conductivity of the nanouid is a function of thermal conductivity of both the
base uid and the nanoparticle material, the volume
fraction, the surface area, and the shape of the nanoparticles suspended in the liquid, and distribution of
the dispersed particles. There are no theoretical formulas available yet for predicting the thermal conductivity of nanouids. Some existing theoretical models
for thermal conductivity were proposed for the solid
liquid mixtures with relatively large particles of the
order of micrometers or millimeters, in which the
apparent thermal conductivity of the suspensions
depends only upon the volume fraction and shape of
the suspended particles, not upon the size and distribution of the particles. Application of these models to
the nanouids is limited. One should experimentally
determine the thermal conductivity of the nanouids.
The transient hot-wire method can be adopted for this
purpose [13].

3704

Y. Xuan, W. Roetzel / Int. J. Heat Mass Transfer 43 (2000) 37013707

In the absence of experimental data and suitable theory for the thermal conductivity of the nanouid,
some existing formulas for predicting the thermal conductivity of solidliquid suspensions with relatively larger particles may be extended approximately to
estimate that of the nanouid. For solidliquid mixtures in which the ratio of conductivity of two phases
is larger than 100, Hamilton and Crosser [11] developed the following model:

kp n 1 k f n 1 f k f k p
keff

kf
kp n 1 kf f kf kp

where the empirical shape factor given by

n

3
c

where c is the sphericity, dened as the ratio of the

surface area of a sphere with a volume equal to that of
the particle to the surface area of the particle. Their
experimental results showed satisfactory coincidence
between the theoretical predictions and the experimental data for spherical particles in the range of volume
fraction up to 30%. For particles of other shapes, the
shape factor n can be allowed to vary from 0.5 to 6.0.
An alternative expression for calculating the eective
thermal conductivity of solidliquid mixtures was
introduced by Wasp [12]:

kp 2kf 2f kf kp
keff

kf
kp 2kf f kf kp

Comparison between these two expressions reveals that

the latter is a special case with the sphericity 1.0 of the
former.
The HamiltonCrosser model was used by Xuan
and Li [3] to obtain a rough estimation of the thermal
conductivity of the nanouids with dierent values of
c from 0.5 to 1.0 and the results from the model corresponding to c 0:7 are close to their experimental
data. Lee et al. [1] pointed out that the predicted thermal conductivity ratios for spherical particles c 1
from this model are in good agreement with their experimental results of Al2O3 nanouids. Thus, this
model may be used for approximation. However, it
must be emphasized that such a primitive estimation
gives no warranty and the accurate and reliable formulas as well as experimental data are needed for
determining the eective thermal conductivity of the
nanouid.
According to the above description, one learns that
the conventional approach of nding heat transfer
coecient of the nanouids is through a method in
which the existing heat transfer coecient correlations
for the pure uid are directly extended to the nano-

uids, with substituting the thermal properties and

transport parameters of the nanouids for those of the
pure uids. Here a supercial conclusion could be that
heat transfer enhancement of nanouids is realized by
increasing the eective thermal conductivity of nanouids. The augmentation eect of heat transfer can approximately be indicated by the ratio
 c
hnf nanofluid
knf
0
10
hf base fluid
kf
where the exponent c is a constant which depends on
the ow pattern. For example, c 2=3 for the turbulent ow. By means of this formula, one can approximately estimate heat transfer enhancement of the
nanouid by enlargement of its thermal conductivity.

3. The modied conventional approach

Although the nanouid behaves more like a uid
than the conventional soliduid mixtures in which
relatively larger particles with micrometer or millimeter
orders are suspended, it is a two-phase uid in nature
and has some common features of the soliduid mixtures. In view of either microscale or macroscale, however, it may be questionable whether the theory of the
conventional two-phase ow can be applied in describing the ow characteristics of the nanouid. Because
of the eects of several factors such as gravity, Brownian force, and friction force between the uid and
ultrane solid particles, the phenomena of Brownian
diusion, sedimentation, dispersion may coexist in the
main ow of a nanouid. This means that the slip velocity between the uid and the particles may not be
zero, although the particles are ultrane. Irregular and
random movement of the particles increases the energy
exchange rates in the uid, i.e., thermal dispersion
takes place in the ow of the nanouid. The thermal
dispersion will atten the temperature distribution and
make the temperature gradient between the uid and
wall steeper, which augments heat transfer rate
between the uid and the wall. Therefore, the enhancement mechanism of heat transfer by the nanouid can
be explained from the following two aspects: one is
that the suspended particles increase the thermal conductivity of the two-phase mixture and another is that
the chaotic movement of the ultrane particles, the
thermal dispersion, accelerates the energy exchange
process in the uid. There is no question that the thermal dispersion plays an important role in heat transfer
enhancement. In their work on the dispersed solid
liquid two-phase ow, Sohn and Chen [13] treated the
soliduid mixture as a uid to investigate microconvective thermal conductivity in two-phase mixtures and
proposed a formula in which the total thermal conduc-

Y. Xuan, W. Roetzel / Int. J. Heat Mass Transfer 43 (2000) 37013707

tivity was treated as a function of the Peclet number.

The particles of 0.3 mm diameter were used in their experiment.
The aforementioned approach takes the rst factor
into account, but neglects the second. To account for
both the factors, the dispersion model can be adopted.
Assume that irregular movement of the ultrane particles induce small perturbations of both the temperature and velocity of the nanouid, i.e., T 0 and u 0 ,
respectively. Thus, the intrinsic phase averages are
given as
T hTif T

11a

u huif u 0

11b

where
one has hTif V1f Vf T dV, huif V1f Vf u dV,

1
0
Vf Vf T dV 0: In the light of the procedure described
by Kaviany [14] and by assuming that the boundary
surface between the uid and the particles is so small
that can be neglected, one obtains the following expression from Eq. (2)


@ hTif
huif
rhTif
rcp nf
@t

r
knf rhTif rcp nf rhu 0 T 0 if
12
The second term on the right-hand side of Eq. (12) indicates the eect of the thermal dispersion resulting
from chaotic movement of the nanoparticles in the
ow. By means of an analogy with the treatment of
turbulence, the heat ux induced by the thermal dispersion in nanouid ow can be expressed as
qd rcp nf hu 0 T 0 if kd
rhTif

13

where kd is the tensor of the dispersed thermal conductivity. Eq. (12) can be rewritten as
@ hTif
huif
rhTif
@t
"
r

kd
anf I
rc
p nf

#
f

rhTi

14

Then, the energy balance equation for the nanouid

can be given as

knf I
rcp nf

1 fDd

15

where D may be called as the total eective thermal

diusivity tensor which includes both the molecular
eect and the eect of the thermal dispersion. Here

16

It is expected that the thermal dispersion tensor is a

function of the ow pattern, properties of both the
base uid and the nanoparticles, the dimensions and
shape of the nanoparticles, and the volume fraction of
the nanoparticles suspended in the mixture.
For the ow of the nanouid inside a tube, the
energy equation may be simplied as
!
"
#
@T
@T
1 @
kd, r
@T
r
u

anf
17
@t
@x
r @r
@r
rcp nf
with accounting for the thermal dispersion in the radial
direction or
!
"
#
@T
@T 1 @
kd, r
@T
r
u

anf
@t
@x r @r
@r
rcp nf

@
@x

"
anf

kd, x
rcp nf

@T
@x

#
18

with the thermal dispersion in both the radial and

axial directions. Here the radial and axial dispersed
thermal conductivity may be dierent from each other.
With respect to the ow of salt-solution in a tube, Aris
[15] and Taylor [16] approximately derived the thermal
diusivity coecients for the laminar and turbulent
ow, respectively,
Dd
1
Pe 2

48
af

for the laminar flow

Dd
7:14Peg1=2
af

for the turbulent flow

19a

19b

where the Peclet number Pe is dened as Pe Rum =af

and gRe is the resistance coecient. The Peclet number Pe comprehends the eects of the macroscale convective and microscale molecular diusion.
So far there is neither theoretical nor experimental
work published on the thermal diusivity coecients
of the nanouid. By reference to some publications
[17,18] about the thermal dispersion in porous media,
it is assumed that the dispersed thermal conductivity
of the nanouid may take the following form:
kd Crcp nf udp Rf

@ hTi
rcp nf hui
rhTi
rcp nf
@t
rcp nf r
D
rhTi

3705

or

kd C  rcp nf uR

20

where C or C  is an unknown constant which should

be determined by matching experimental data. Other
forms of the dispersed thermal conductivity are also
possible, which may vary from case to case. To take
the eect of thermal dispersion into account, the surface heat ux can be expressed as

3706

Y. Xuan, W. Roetzel / Int. J. Heat Mass Transfer 43 (2000) 37013707



@T
@T
kd, r
j
q knf
@r
@ r rR

21

This expression is for the ow in a tube. Similar to the

procedure introduced by Nield and Bejan [7] in their
book, the dimensionless analysis of Eq. (21) formally
results in the heat transfer correlation of the nanouid
as


Nux 1 C  Pe f 0 0 y 0 0 Rem
22a
and with regard to the ow pattern, it may be modied
as
Nux 1 C  Pen f 0 0y 0 0Rem

22b

where Pe Rum =anf , the exponent m and n depend

upon the ow pattern, and the dimensionless velocity f
and dimensionless temperature y may be dened
according to the case of application. f 0 and y 0 are the
derivatives of the dimensionless velocity and the
dimensionless temperature, respectively. Thus, the thermal dispersion increases heat transfer. What needs to
be emphasized is that experiment is necessary to determine the unknown coecient C  : The case C  0
corresponds to zero thermal dispersion. This expression clearly indicates that heat transfer enhancement of the nanouid increases with the Peclet
number. For the nanouid ow over a at plate or
other forms, the Nusselt correlations similar to Eq.
(22) can be derived.

4. Discussions
The nanoparticles enhance heat transfer rate by
increasing the thermal conductivity of the nanouid
and incurring thermal dispersion in the ow, which is
an innovative way of augmenting heat transfer process.
Although there are some sophisticated theories as well
as correlations for the conventional soliduid ows
[6,1921], it is questionable and doubtful that these
theories and correlations are applicable to the nanouid. As a new type of heat transfer medium, the
nanouid behaves more like a single-phase uid
because the discontinuous phase consists of ultrane
particles, so that the heat transfer correlation for the
single-phase uid, rather than those for the conventional solidliquid two-phase ow, has been taken as
the start point. By starting from the existing theory
and correlations for the heat transfer process of pure
uids, this paper has proposed two approaches of deriving heat transfer correlation for the nanouid. To
handle the thermal dispersion resulting from irregular
movement of the nanoparticles, the dispersion model
has been used.
The nanouid appears with development of

nanoscience and nanotechnology. The great potential

of nanouids in enhancing heat transfer means chances
and challenges in thermal science and engineering.
Since the concept of the nanouid is newly proposed,
there are number of questions which remain unclear
and need to be solved. Besides the above-mentioned
movement modes of the ultrane particles, for
example, aggregation may take place and clusters may
be formed in the ow of the nanouid. All these
chaotic movement modes of the nanoparticles will
aect the distribution of the particles, transport properties and heat transfer performance of the nanouid.
Research on microscale is necessary to learn the microstructure of the nanouid, which would help in understanding the ow and heat transfer process of the
nanouid. Theoretical and experimental research is
needed in order to apply the nanouid for the
enhancement of heat transfer and to assess the eects
by its use.

Acknowledgements
The rst author would like to express his thanks to
the National Science Foundation of China (Grant No.
59976012) and Universitaet der Bundeswehr Hamburg
for the nancial aid.

References
[1] S. Lee, S.U.-S. Choi, S. Li, J.A. Eastman, Measuring
thermal conductivity of uids containing oxide nanoparticles, J. of Heat Transfer 121 (1999) 280289.
[2] H. Masuda, A. Ebata, K. Teramae, N. Hishinuma,
Alternation of thermal conductivity and viscosity of
liquid by dispersing ultra-ne particles (dispersion of gAl2O3, SiO2 and TiO2 ultra-ne particles), Netsu Bussei
(Japan) 4 (1993) 227233.
[3] Y. Xuan, Q. Li, Heat transfer enhancement of nanouids, Int. J. Heat Fluid Flow 21 (2000), 15864.
[4] S.U.-S. Choi, Enhancing thermal conductivity of uids
with nanoparticles, Developments and applications of
non-Newtonian ows, ASME FED 231/MD 66 (1995)
99103.
[5] S. Lee, S.U.-S. Choi, Application of metallic nanoparticle suspensions in advanced cooling systems, in:
Recent Advances in Solids/Structures and Application
of Metallic Materials, ASME PVP 342/MD 72 (1996)
227234.
[6] D.A. Drew, S.L. Passman, Theory of Multicomponent
Fluids, Springer, Berlin, 1999.
[7] D.A. Nield, A. Bejan, Convection in Porous Media, 2nd
ed., Springer, Berlin, 1952.
[8] H.C. Brinkman, The Viscosity of concentrated suspensions and solutions, J. Chemistry Physics 20 (1952) 571
581.
[9] N. Zuber, On the dispersed two-phase ow in the lami-

Y. Xuan, W. Roetzel / Int. J. Heat Mass Transfer 43 (2000) 37013707

[10]
[11]
[12]
[13]

[14]
[15]
[16]

nar ow regime, Chemical Engineering Science 19

(1964) 897917.
Y. Xuan, Q. Li, Experimental research on the viscosity
of nanouids, Report of Nanjing University of Science
and Technology (in Chinese), 1999.
R.L. Hamilton, O.K. Crosser, Thermal conductivity of
heterogeneous
two-component
systems,
I&EC
Fundamentals 1 (1962) 182191.
F.J. Wasp, SolidLiquid Slurry Pipeline Transportation,
Trans. Tech, Berlin, 1977.
C.W. Sohn, M.M. Chen, Microconvective thermal conductivity in disperse two-phase mixture as observed in a
low velocity couette ow experiment, J. of Heat
Transfer 103 (1981) 4751.
M. Kaviany, Principles of Heat Transfer in Porous
Media, Springer, Berlin, 1995.
R. Aris, On the dispersion of a solute in a uid owing
through a tube, Proc. Roy. Soc. (London) A235 (1956)
6777.
G.I. Taylor, The dispersion of matter in turbulent ow

[17]
[18]

[19]
[20]

[21]

3707

in a pipe, Proc. Roy. Soc. (London) A223 (1954) 446

468.
M.L. Hunt, C.L. Tien, Eects of thermal dispersion on
forced convection in brous media, Int. J. Heat Mass
Transfer 31 (1988) 301309.
O.A. Plumb, The eect of thermal dispersion on heat
transfer in packed bed boundary layers, Proc. ASME
JSME Thermal Engineering Joint Conference 2 (1983)
1722.
D. Gidaspow, Multiphase Flow and Fluidization
Continuum and Kinetic Theory Descriptions, Academic
Press, San Diego, 1994.
M. Jamialahmadi, M.R. Malayeri, H. MuellerSteinhagen, Prediction of heat transfer to liquidsolid
uidized beds, Canadian J. of Chemical Engineering 73
(1995) 444455.
Y. Kato, K. Uchida, T. Kago, S. Morooka, Liquid
holdup and heat transfer coecient between bed and
wall in liquidsolid and gasliquidsolid uidized beds,
Powder Technology 28 (1981) 173179.