Sie sind auf Seite 1von 7

Home Search Collections Journals About Contact us My IOPscience

A new model for heat conduction of nanofluids based on fractal distributions of nanoparticles

This article has been downloaded from IOPscience. Please scroll down to see the full text article.

2006 J. Phys. D: Appl. Phys. 39 4486

(http://iopscience.iop.org/0022-3727/39/20/028)

View the table of contents for this issue, or go to the journal homepage for more

Download details:
IP Address: 128.151.244.46
The article was downloaded on 10/08/2013 at 20:51

Please note that terms and conditions apply.


IOP PUBLISHING JOURNAL OF PHYSICS D: APPLIED PHYSICS
J. Phys. D: Appl. Phys. 41 (2008) 139801 (1pp) doi:10.1088/0022-3727/41/13/139801

Corrigendum
A new model for heat conduction of nanofluids
based on fractal distributions of nanoparticles
Jie Xu and Boming Yu 2006 J. Phys. D: Appl. Phys. 39
44864490

On page 4488 in the ninth line from the bottom of the right
column, the text and df = 1.7 1010 m for water should be
corrected to read as and df = 4.5 1010 m for water.

0022-3727/08/139801+01$30.00 1 2008 IOP Publishing Ltd Printed in the UK


INSTITUTE OF PHYSICS PUBLISHING JOURNAL OF PHYSICS D: APPLIED PHYSICS
J. Phys. D: Appl. Phys. 39 (2006) 44864490 doi:10.1088/0022-3727/39/20/028

A new model for heat conduction of


nanofluids based on fractal distributions
of nanoparticles
Jie Xu, Boming Yu1 , Mingqing Zou and Peng Xu
Department of Physics, Huazhong University of Science and Technology, 1037 Luoyu Road,
Wuhan, 430074, Peoples Republic of China
E-mail: yuboming2003@yahoo.com.cn

Received 21 June 2006, in final form 2 August 2006


Published 29 September 2006
Online at stacks.iop.org/JPhysD/39/4486

Abstract
In this paper we report a new model for predicting the thermal conductivity
of nanofluids by taking into account the fractal distribution of nanoparticle
sizes and heat convection between nanoparticles and liquids due to the
Brownian motion of nanoparticles in fluids. The proposed model is
expressed as a function of the average size of nanoparticles, fractal
dimension, concentration of nanoparticles, temperature and properties of
fluids. The model shows the reasonable dependences of the thermal
conductivity on the temperature of nanofluids, nanoparticle size and
concentration. The parameter c introduced in thermal boundary layer
depends on fluids, but is independent of nanoparticles added in the fluids.
The model predictions are in good agreement with the available experimental
data. The model also reveals that there is a critical concentration of 12.6%
of nanoparticles at which the contribution from heat convection due to the
Brownian movement of nanoparticles reaches the maximum value, below
which the contribution from heat convection decreases with the decrease in
concentration and above which the contribution from heat convection
decreases with the increase in concentration.

1. Introduction the traditional models for composites, such as the Hamilton


Crosser model (HC), completely fail to explain the
Fluids with suspended nanoparticles are called nanofluids. The enhancement in the effective thermal conductivity of
term, nanofluid, was named by Choi in 1995 [1]. Nanofluids nanofluids. Some researchers reported that the possible
are thought to be the next-generation heat transfer fluids mechanisms for enhancing thermal conductivity are mainly
because the thermal conductivity can be significantly increased related to the interaction between nanoparticles and liquids,
by adding small amount of nanoparticles. Many researchers the size of nanoparticles, nanoparticle clustering and the heat
have carried out their experiments [27] to verify the results convection due to Brownian motion of nanoparticles in the
by Choi. Recently Jow et al [8] designed a measurement liquids. Different authors proposed different renovated models
device for thermal conductivity of nanofluids. The thermal based on the former works to explain the mechanisms of heat
conductivity of nanofluids attracts more and more interest transfer in nanofluids [1422]. Kumar et al [21] considered
both experimentally and theoretically. However, no generally the heat convection due to Brownian motion of particles in
accepted theory for explaining this anomalous increase is the fluids and assumed that the effective thermal conductivity
available. of the particle is directly proportional to its mean velocity and
Since the pioneering work by Maxwell [9], several expressed the thermal conductivity of the particle as kp = c u p ,
researchers [1013] developed theories for the effective where u p is the average particle velocity and c is a constant.
However, this model results in an unreasonable result because
thermal conductivity (k) of multiphase systems. However,
when the mean velocity u p of nanoparticles is zero, the thermal
1 Author to whom any correspondence should be addressed. conductivity of the particle is kp = 0 according to the equation.

0022-3727/06/204486+05$30.00 2006 IOP Publishing Ltd Printed in the UK 4486


A new model for heat conduction of nanofluids

This is inconsistent with the practical situation [23] because respectively. If T = 343 K, the values of Re and P r are
when u p = 0, even though all the particles are stationary, the 1.8 102 and 2.75, respectively. It is seen that the Reynolds
thermal conductivity of the particle should be equal to that of number Re  1. According to Tomotika et al [24], the
the particle itself and should be non-zero. Jang et al [22] Nusselt number for liquid flowing around a spherical particle
proposed a model for the effective thermal conductivity of can be expressed as Nu = 2.0 + 0.5ReP r + O(Re2 P r 2 ), so
nanofluids, but the model introduced an empirical constant Nu = 2.0 + 0.5ReP r + O(Re2 P r 2 ) 2.0 is applied for
which has no physical meaning. Ren et al [20] built a model a single particle in this work. We have noted that Jang and
which includes the effects of nanolayer and heat convection of Choi [22] used Nu Re2 P r 2 for the Nusselt number by
nanoparticles in the fluids. While most models are based on assuming that the Reynolds number and the Prandtl numbers
an assumption that suspended nanoparticles in nanofluids have are on the order of 1 and 10, respectively, for typical nanofluids.
uniform diameters, recent experimental study [16] showed However, no details of calculation of the Nusselt number are
the nanoparticles might differ in size and follows the fractal provided in Jang and Chois work.
distribution. To date there is no model that takes into account The heat transfer coefficient is defined by h =
the size distribution of nanoparticles. (N u kf )/. The heat transferred by convection for a single
In this work, we derive a model for heat convection due to nanoparticle moving in liquids is given by
the Brownian motion of nanoparticles by applying the fractal
geometry theory and explain the mechanisms of the enhanced q = hA (Tp Tf ), (3)
heat transfer in nanofluids.
where Tp and Tf are the temperatures of particle and liquid,
respectively, A is the surface area of a particle whose diameter
2. Description of the thermal conductivity model is and A = 2 .
It has been shown that the size distribution of nanoparticles
We divide the heat transfer in nanofluids into two portions:
in nanofluids follows the fractal power law [16]. Analogous to
the heat transferred by stationary nanoparticles and the other
pores in fractal porous media, the number of particles whose
by heat convection caused by moving nanoparticles. The total
sizes are within the infinitesimal range +d can be obtained
thermal conductivity of nanofluids is thus expressed as
as [26, 27]
(Df +1)
knf = ks + kc , (1) dN = Df D max
f
d. (4)
In equation (4), dN > 0, which implies the number of
where ks is the thermal conductivity by stationary nanoparticles particles decreases with the increase in particle size. The
in the liquid, which is simulated by the HC model and fractal dimension Df is determined by
kc is the thermal conductivity by heat convection caused
by Brownian motion of nanoparticles. The HC model for ln
Df = d , (5)
spherical nanoparticles suspended in nanofluids is expressed ln(min /max )
as [10]
ks kp + 2kf 2(kf kp ) where d = 2 in two dimensions, is the concentration
= , (2) of nanoparticles and min and max are the minimum and
kf kp + 2kf + (kf kp )
maximum diameters of nanoparticles, respectively. Wang
where kp and kf are the thermal conductivities of particles and et al [16] measured the SiO2 ethanol nanofluid and obtained
fluid, respectively, and is the concentration of nanoparticles the fractal dimension 1.57 for nanoparticles when is about
in nanofluids. Here we focus our attention on the second 6.5%. From equation (5), it is estimated min /max 103 .
term kc in equation (1). According to the theory by Tomotika Feng et al [28] compared the fractal dimension by equation
et al [24], the Nusselt number for liquid flowing around a (5) with the experimental data by several investigators on the
sphere can be expressed as Nu = 2.0+0.5ReP r +O(Re2 P r 2 ), fractal dimensions versus the fractions of pores and found
where P r(=/ = /f = cp /kf ) is the Prandtl number that the ratio of min /max = 103 best fits the experimental
for fluids and , , , f and cp are kinematical viscosity, data for different porous media. We consider that the pores
thermal diffusion coefficient, absolute viscosity, density and in porous media are analogous to the particles in nanofluids,
specific heat of fluid at constant pressure, respectively. Re(= then equation (5) can also be used to describe the particles in
up /) is the Reynolds number, where is the diameter of nanofluids and the ratio of min /max = 103 is thus applied
nanoparticles and up is the velocity of particle, and up = for nanoparticles in this work.
D0 / lf [22], where lf is the mean free path of liquid molecule, So the heat transferred by convection of all particles in the
and we assume that lf is the same as the diameter of liquid liquid can be found from
molecule. D0 is the nanoparticle diffusion coefficient, which  max  max
is determined by D0 = kb T /3, where T is temperature qc = q dN = hA (Tp Tf ) dN
and kb = 1.38 1023 J K1 is the Boltzmann constant. min min
The viscosity is given by = A 10B/(T C) [25], where  max
T
A = 2.414 105 , B = 247.8 and C = 140 for water. At = hA T dN , (6)
room temperature, say T = 300 K, we can calculate the value min T
of (=2.414 103.45 Pa s1 ). The values of f and lf are [7] where T is the thickness of the thermal boundary layer of
1.0103 Kg m3 and 1.71010 m, respectively. The specific heat convection caused by the nanoparticles Brownian motion,
heat of water at constant pressure is 4.2103 J (Kg K)1 . Then T /P r [22] and T = Tp Tf . In equation (6), we have
the values of Re and P r are found to be 4.0 103 and 6.0, assumed the local thermal equilibrium. The laminar boundary

4487
J Xu et al

layer can be approximated by = 3df [29], where df is the 1.45


diameter of liquid molecule. Al2O3-water(13nm,experiment)[2]
1.40
The equivalent thermal conductivity contributed by heat CuO-water(28.6nm,experiment)[3]
convection is approximated by 1.35 Al2O3-water(present model)
 max  max 1.30 CuO-water(present model)
qc min hA T dN hA T dN
kc = = = minmax , (7) 1.25
A T
T A

knf / kf
min A dN
1.20
where A is the total area of nanoparticles.
The total dimensionless effective thermal conductivity of 1.15
nanofluids is then 1.10
knf kp + 2kf 2(kf kp )
= ks+ + kc+ = 1.05
kf kp + 2kf + (kf kp )
1.00
 max 0.00 0.01 0.02 0.03 0.04 0.05
1 min hA T dN
+  max , (8) Particle volume fraction
kf
min A dN
 max  max Figure 1. The dimensionless effective thermal conductivity versus
where ks+ = kkfs and kc+ = kkcf = (1/kf )( min hA T dN/ min the concentration of nanoparticles.
A dN).
The dimensionless thermal conductivity kc+ can be Equation (11) is a function of average nanoparticle size,
obtained by combining equations (3)(4) and (7)(8) as minimum and maximum nanoparticle sizes, fluid property,
temperature, concentration and fractal dimension. It is
1Df
max min
Nu df 2 Df 1D f evident that no new empirical constant is introduced and
kc+ = c 2D , (9) every parameter has clear physical meaning in equation (11).
Pr 1 Df max f 2D
min
f

Equation (11) explicitly denotes that the dimensionless thermal


where c is an empirical constant. Equation (9) is the fractal conductivity of nanofluids due to heat convection by Brownian
model for the dimensionless thermal conductivity caused motion of nanoparticles is proportional to ( )1 . Note that
by heat convection due to Brownian motion. There is no the fractal dimension is also related to the concentration of
new empirical constant introduced in equation (9) except the nanoparticles (see equation (5)) and Prandtl number P r is
parameter c, which is relevant to the thickness of thermal related to temperature.
boundary layer, i.e. T /P r. The Nusselt number is taken The total dimensionless effective thermal conductivity of
to be 2.0 in equation (9). nanofluids based on equations (1), (2) and (11) can be written as
The average diameter of nanoparticles can be found from knf kp + 2kf 2(kf kp )
the statistical property of fractal media, and it is = ks+ + kc+ =
kf kp + 2kf + (kf kp )
    2
Df min 
= f ()d = max , (10a) max
1Df
1
Df 1 max Nu df (2 Df )Df min 1
+c  2Df
. (12)
where f () = Df D (Df +1) Pr (1 Df )2
f
min is the density distribution max
min
1
function [26]. Equation (10a) can be also rewritten as
Equation (12) indicates that the total dimensionless effective
 1
Df 1 min thermal conductivity knf /kf varies with the Prandtl number P r,
max = , (10b) the diameter df of molecule of fluids, the concentration, fractal
Df max
dimension and sizes of nanoparticles. Since min /max =
Df 1 103 is used in this work, once the concentration is
min = . (10c)
Df given/measured, the fractal dimension Df can be found from
equation (5) and the thermal conductivity can be found from
From equations (10a)-(10c) it is seen that due to min /max =
equation (12) as long as the parameters P r, df , kf and kp are
103 , once the fraction/concentration and mean diameter of
known.
nanoparticles are given/measured, the fractal dimension Df
can be determined by equation (5), and the maximum or Figures 1 and 2 compare the model (equation (12)) predic-
minimum diameter of nanoparticles can be determined by tions with the available experimental data (experimental infor-
equations (10b) or (10c). mation is at room temperature, P r = 6.0, kf = 0.610W mK1
With the help of equation (10a)(10c) the dimensionless and df = 1.7 1010 m for water; P r = 20.6, kf =
thermal conductivity caused by heat convection can be 0.258W mK1 and df 1.7 1010 m for ethylene glycol; kp
obtained from equation (9) as are 46.0 W mK1 for Al2 O3 nanoparticles and 69.0 W mK1
 2 for CuO nanoparticles, respectively.), and good agreement is
1Df
found between them. Figure 1 also clearly indicates that the
max
1
Nu df (2 Df ) Df min 1 thermal conductivity of the nanofluids increases with the in-
kc+ = c  2Df .
Pr (1 Df ) 2
crease in nanoparticles concentration and with the decrease
max
min
1 in nanoparticle sizes. It should be noted that there is an em-
(11) pirical parameter c in equation (12) because of uncertainty of

4488
A new model for heat conduction of nanofluids

1.40 expremental data(CuO-water,28.4nm)[3]


Al2O3-water(13nm)[2]
expremental data(Al2O3-water,11nm)[7]
1.35 Al2O3-water(60.4nm)[4] 1.50
expremental data(Al2O3-water,47nm)[7]
CuO-water(28.6nm)[3] 1.45
1.30 Al2O3-EG(35nm)[5] present model(CuO-water,28.4nm)
Experiment data

1.40 present model(Al2O3-water,11nm)


CuO-EG(35nm)[5]
1.25 present model(Al2O3-water,47nm)
CuO-EG(30nm)[6] 1.35
1.20 line of slope =1
1.30

knf / kf
1.15 1.25

1.10 1.20
1.15
1.05
1.10
1.00 1.05
1.00 1.05 1.10 1.15 1.20 1.25 1.30 1.35 1.40
Present model prediction 1.00
290 295 300 305 310 315 320 325 330 335 340
Figure 2. A comparison of the present model predictions with the Temperature (K)
experimental data.
Figure 4. The effective thermal conductivity of nonofluids versus
temperature at = 0.01.
2.0
The effective thermal conductivity

1.9 +
H-C model,ks (Al2O3-water) effective thermal conductivity (knf /kf ) decreases with the
1.8
present model increase in nanoparticles sizes. This phenomenon can be
1.7 explained by the theory of Brownian motion: the smaller the
average size of nanoparticles in the fluid, the higher the velocity
1.6
of nanoparticles Brownian motion. Thus, the heat transferred
1.5 by heat convection is enhanced. It is also interestingly seen
1.4 from figure 3 that when the particle size is less than about
1.3 15 nm, the effective thermal conductivity increases drastically.
When the diameter of nanoparticles is 5 nm, the thermal
1.2
conductivity contributed from the heat convection takes about
1.1 20% in total thermal conductivity, whereas the HC model
1.0 gives 1.10. This exactly shows that the present model is
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 more suitable than the HC model to describe the thermal
Nanoparticle average diameter (nm) conductivity of nanofluids and the thermal conductivity of
Figure 3. The effective thermal conductivity (knf /kf ) versus the nanofluids is enhanced greatly. In addition, equation (12) also
nanoparticles sizes at = 0.03. shows that since the Prandtl number varies with the change in
temperature, it is expected that the value of knf /kf might change
the value of T , T /P r, which is primarily related to the with the temperature of nanofluids. Therefore, equation (12)
character of liquids. We determine the values of c in our model provides the reasonable result. Figure 4 clearly shows that
to predict the thermal conductivities of nanofluids by matching the value of knf /kf increases with temperature. But the HC
with the experimental data. We find that the values of c are model is not relevant to temperature. Figure 4 also compares
related to fluids. When different nanoparticles are added in the the present model predictions with the available experimental
same fluid, the values of c are same; for example, the value data [3, 7] on the deionized water with different nanoparticles
of c is 85.0 both for the Al2 O3 nanoparticles and for the CuO for the thermal conductivity knf /kf versus temperature. It is
nanoparticles added in the deionized water. However, when seen that the model predictions present good agreement with
the same nanoparticles are added in different fluids, the values the experimental data for nanofluids suspended with Al2 O3
of c are different, e.g. c = 85.0 for the deionized water and particles and fair agreement with the experimental data for
c = 280.0 for the ethylene glycol. This means that the value nanofluids suspended with CuO particles, and overall good
of c is independent of the particles added in the fluids and is agreement between the model predictions and the experimental
only relevant to fluids, which is consistent with the other ob- data is found.
servations [18]. The order of magnitude of c is about 102 . This The contributions from convection (i.e. from the second
is primarily because we take Nu 2.0 (for a single particle) term in equation (12)) to the total thermal conductivity are
in calculation. Whereas Brodkey et al [30] pointed out that shown in figure 5, in which we take the nanofluids of
the Nusselt number Nu for whole nanofluid should be 20100 Al2 O3 water with the average diameters of 5.0 nm, 13.0 nm,
times of a single nanoparticle. 30.0 nm and 60.0 nm, respectively. It is again found from
the figure that the contributions from the heat convection
3. Results and discussion due to nanoparticles Brownian motion increase with the
decrease in nanoparticle sizes. The maximum values of the
Taking the nanofluid of the deionized water with the Al2 O3 contributions from the heat convection are about 21% and
nanoparticles as an example, figure 3 shows that the total 2.1%, respectively, for the nanoparticles with the average

4489
J Xu et al

nanoparticles is higher than the critical concentration of


0.20 5nm 12.6%. The model predictions are compared with the available
13nm experimental data, and good agreement between the model
30nm
60nm
predictions and experimental data is found.
0.15

Acknowledgments
kc / knf

0.10
This work was supported by the National Natural Science
Foundation of China through Grant number 10572052.
0.05

References
0.00
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 [1] Choi S U S 1995 Enhancing thermal conductivity of fluids
Concentration() with nanoparticles Development and Applications of
Non-Nemtonian Flows ed D A Siginer and H P Wang,
Figure 5. The contributions from the second term in equation (12) FED-vol 231/MD-vol 66 (New York: ASME) p 99
versus the concentration of nanoparticles. [2] Masuda H, Ebata A, Teramae K and Hishinuma N 1993 Netsu
Bussei 4 227
[3] Das S K, Putra N, Thiesen P and Roetzel W 2003 J. Heat
diameters of 5.0 and 60.0 nm. It is also interestingly found that Transfer 125 567
there is a critical concentration of nanoparticles, at which the [4] Xie H Q, Wang J C, Xi T J, Liu Y and Ai F 2002 J. Appl. Phys.
contribution from the convection reaches the maximum value, 91 4586
below which the contribution from the convection increases [5] Eastman J A, Choi S U S, Li S, Yu W and Thompson L 2001 J.
with the concentration and above which the contribution from Appl. Phys. Lett. 78 718
[6] Lee S, Choi S U S, Li S and Eastman J A 1999 J. Heat
the convection decreases with the increase in the concentration. Transfer 121 28
It is found that the critical concentration is 12.6% for all [7] Chon C H and Kihm K D 2005 Appl. Phys. Lett. 87 153107
nanoparticles, i.e. the critical concentration is independent [8] Jow C S, Teng T P, Hung C J and Guo Y T 2005 J. Phys. Conf.
of nanoparticle sizes. It can be also seen from figure 5 Ser. 13 55
that when the concentration of nanoparticles continuously [9] Maxwell J C 1881 A Treatise on Electricity and Magnetism
2nd edn, vol 1 (Oxford: Clarendon) p 435
increases, the contribution from the convection tends to zero [10] Hamilton R L and Crosser O K 1962 I & EC Fundamentals
because at higher concentration the nanoparticles may be 1187
easily agglomerated to form clusters, and the strength of the [11] Bionnecaze R R and Brady J F 1991 Proc. R. Soc. Lond. A 432
Brownian motion is weakened, leading to the decrease in heat 445
convection and even to zero (if = 1, the fluid becomes a [12] Adler P M 1992 Porous Media: Geometry and Transports
(Boston: Butterworth-Heinemann) p 434
solidsolid medium and thermal conductivity is completely [13] Auriault J L and Ene H I 1994 Int. J. Heat Mass Transfer
determined by the HC model, with no convection at all). 37 2885
[14] Yu W and Choi S U S 2003 J. Nanopart. 5 167
[15] Xie H Q 2003 Phys. Lett. A 307 313
4. Conclusions [16] Wang B X, Zhou L P and Peng X F 2003 Int. J. Heat Mass
Transfer 46 2665
Based on the fractal distribution of nanoparticles and heat [17] Keblinski P, Phillpot S R, Choi S U S and Eastman J A 2002
convection between the nanoparticles and fluid due to the Int. J. Heat Mass Transfer 45 855
Brownian motion of nanoparticles in fluids, a fractal model [18] Prasher R 2005 Phys. Rev. Lett. 94 025901
for the effective thermal conductivities of nanofluids has been [19] Gao L and Zhou X F 2006 Phys. Lett. A 348 355
[20] Ren Y, Xie H and Cai A 2005 J. Phys. D: Appl. Phys. 38 3958
proposed. The proposed model is expressed as a function of the [21] Kumar D H, Patel H E, Kumar V R R, Sundararajan T,
average size of nanoparticles, fractal dimension, concentration Pradeep T and Das S K 2004 Phys. Rev. Lett. 93 144301
of nanoparticles and temperature and properties of fluids. No [22] Jang S P and Choi S U S 2004 Appl. Phys. Lett. 84 4316
new empirical constant is introduced in the proposed model and [23] Bastea S 2005 Phys. Rev. Lett. 95 019401
every parameter in this model has a clear physical meaning. [24] Tomitika S, Aoi T and Yosinabu H 1953 Proc. R. Soc. Lond.
Ser. A 129 233
It is found that the parameter c introduced later in the thermal [25] Fox R W, McDonald A T and Pritchard P J 2004 Introduction
boundary is related to fluids, but independent of nanoparticles to Fluid Mechanics 6th edn (New York: Wiley), p 724
added in the fluids. It is also found that the thermal [26] Yu B M and Li J H 2001 Fractals 9 365
conductivity contributed by heat convection due to Brownian [27] Yu B M and Cheng P 2002 Int. J. Heat Mass Transfer 45 2983
movement of nanoparticles increases as the concentration [28] Feng Y, Yu B, Zou1 M and Zhang D 2004 J. Phys. D: Appl.
Phys. 37 3030
of nanoparticles is less than the critical concentration of [29] Yu C J, Richter A G, Datta A, Durbin M K and Dutta P 1999
12.6%, at which the contribution from heat convection reaches Phys. Rev. Lett. 82 2326
the maximum value, and decreases as the concentration of [30] Brodkey R S 1991 Int. J. Heat Mass Transfer 34 2327

4490

Das könnte Ihnen auch gefallen