Escolar Documentos
Profissional Documentos
Cultura Documentos
2
_
(5)
Table 1
Uncertainties estimation for primary measurements and dependent quantities.
Primary measurements
Diameter 0.05 mm
Length 0.2 mm
Temperature 0.1 K
Inner tube ow rate, range: 30e540 L h
1
2.0% at the lowest ow rate
Annulus ow rate, range: 30e300 L h
1
2.0% at the lowest ow rate
Pressure drop across the inner tube 1.5% at the lowest ow rate
Dependent quantities
Mass ow rate m, kg s
1
2.0%
Heat ux q, W m
2
2.8%
LMTD, K 1.5%
Apparent Darcy friction factor f
app
3.3%
Heat transfer coefcient h, W m
2
K
1
3.2%
Table 2
Several existing equations for effective dynamic viscosity and thermal conductivity
of nanouids.
Authors Equations
Einstein [19] Theoretical model for dilute non-interacting
suspensions of small, rigid, spherical particles,
F < 2%: m
n
m
f
1 2:5F
Maxwell [20] Effective medium theory, for dilute non-contact
suspensions of rigid spherical particles,
F < 2%: kn k
f
kp2kf 2Fkpkf
kp2kf Fkpkf
Maiga et al. [21] Least-square curve tting of three data sets
for g-Al
2
O
3
/water nanouid:
m
n
m
f
1 7:3F 123F
2
,
kn k
f
1 2:72F 4:97F
2
ln T
hi
T
co
=T
ho
T
ci
(6)
Assuming no fouling resistance and ignoring the wall thermal
resistance due to the thin wall, large tube length and high thermal
conductivity of copper, the inner tube heat transfer coefcient h
was determined by
h
1
A
i
_
LMTD
q
1
haAo
_ (7)
The annulus thermal resistance in Eq. (7) was also neglected
because of the following reasons: (1) the annulus heat transfer
coefcient h
a
is relatively large due to the intensive turbulence
induced by the ns on the outer surface of the inner tube; (2) A
o
/
A
i
4.83; (3) the volumetric ow rate on the annulus side was kept
relatively large during the experiments; (4) a small change in h was
noticed for a 20% change of the annulus ow rate during the ex-
periments. Thus, Eq. (7) can be simplied as
h
q
A
i
$LMTD
(8)
Only the inner tube heat transfer coefcient was investigated
and evaluated in this study. Uncertainties of the dependent quan-
tities were listed in Table 1.
Before and after the nanouid tests, water experiments were
conducted in the same double-pipe helically coiled heat exchanger
to verify the nanouid stability, as shown in Fig. 3. The water
experimental data points before and after the nanouid tests show
very similar thermal behavior, indicating very small and negligible
deposition of nanoparticles during the nanouid tests.
3. Results and discussion
3.1. Pressure drop
The relationship between the apparent Darcy friction factor f
app
calculated fromEq. (4) and the Reynolds number Re for tap water is
illustrated in Fig. 4. The apparent friction factor decreases with Re
when Re < 6000, while it increases slowly when Re > 6000. In this
study, a critical Reynolds number of approximately 6000 was
assumed, which agrees with the transition value of 6494 calculated
by the transition criterion in Ito [26]. The Ito equation [26] and the
Seban and McLaughlin equation [27] can predict the experimental
value relatively well for both laminar ow and turbulent ow,
respectively. The apparent Darcy friction factors for the ve nano-
uids are presented in Fig. 5. Data of tap water is also included for
comparison. The transition from laminar ow to turbulent ow for
all the tested uids occurs almost at the same Reynolds number.
Therefore, the transitional velocity of the nanouids will be larger
than that of the base uid due to the larger viscosity of the former
compared to the latter. The nanoparticles may stabilize the ow in
helically coiled tubes. However, more data are needed to verify this
phenomenon. No obvious difference exists among the six tested
uids, especially in the laminar ow. During the turbulent ow, the
friction factor seems to increase with the nanoparticle concentra-
tion. Fig. 6a and b presents comparisons of the experimental fric-
tion factors with the predictive friction factors by the Ito equation
[26] and the Seban and McLaughlin equation [27] for laminar
ow and turbulent ow, respectively. Both equations can predict
the data points within a 30% error band. The Seban and
McLaughlin equation [27] tends to under-estimate the turbulent
friction factor, and this underestimated deviation increases with
Reynolds number and weight concentration of the nanoparticles.
3.2. Heat transfer in laminar ow
Fig. 7 demonstrates the relationship between Nu
b
(Pr
b
)
0.4
and
the inner tube Dean number De
b
(Re
b
(d
i
/D
c
)
0.5
) for laminar ow.
The subscript b indicates the average bulk temperature. All
Fig. 2. Rheological behavior of alumina nanouids at 20
C based on existing equations
in Table 2: (a) relative viscosity vs. volume concentration; (b) relative thermal con-
ductivity vs. volume concentration.
Fig. 3. Water experimental data before and after nanouid tests.
Z. Wu et al. / Applied Thermal Engineering 60 (2013) 266e274 270
properties used in the dimensionless numbers were calculated at
the average bulk temperature. The average bulk temperature was
estimated from the inner tube inlet and outlet temperatures. For
nanouids, the nanouid properties were used instead of those of
the base uid. Temperature effects were accounted for in the
Prandtl number Pr
b
. As shown in Fig. 7, the Nusselt number in-
creases with the Dean number. Tap water and the ve nanouids
present similar heat transfer characteristics. This indicates that the
net effect of nanoparticles on the heat transfer performance in
helically coiled tubes is probably insignicant. The thermal con-
ductivity increase by nanoparticles is benecial for heat transfer,
whereas the secondary ow intensity induced by centrifugal forces
may be reduced by nanoparticles due to the larger viscosity and
density of the nanouid compared to that of the base uid.
A new correlation for water and nanouids owing inside he-
lically coiled tubes for laminar ow was developed:
Nu
b
0:089*De
0:775
b
Pr
0:4
b
(9)
The coefcient of determination, R
2
, is equal to 0.995, indicating
that Eq. (9) ts the data very well.
Comparison of the newly proposed correlation Eq. (9) for
laminar ow in helically coiled tubes versus the other two existing
correlations: the Dravid et al. correlation [28] and the Kalb and
Seader correlation [29], is given in Fig. 8. It is found that Eq. (9) gives
the best prediction on the ground that all data points are located
within a 5% error band except those with Nu
b
< 13. The formulas
of the other two correlations and their applicable ranges are
presented in Table 3. The mean absolute error e
A
(%) and standard
deviation s
N
(%) of the three correlations are provided in Table 4.
3.3. Heat transfer in turbulent ow
Similar to the laminar ow, Fig. 9 presents the relationship be-
tween Nu
b
(Pr
b
)
0.4
and the inner tube Reynolds number Re
b
for the
turbulent ow. It is clear that Nu
b
(Pr
b
)
0.4
increases with the Re
b
and all the six uids show a similar trend. The Seban and
Fig. 5. The f
app
eRe relationship for the six tested uids.
Fig. 6. Darcy friction factor evaluations for (a) laminar ow, and (b) turbulent ow.
Fig. 7. Nu
b
(Pr
b
)
0.4
vs. De
b
for laminar ow.
Fig. 4. The f
app
eRe relationship for water and the laminar-to-turbulent transition.
Z. Wu et al. / Applied Thermal Engineering 60 (2013) 266e274 271
McLaughlin [27] correlation can predict the thermal behavior of
water and nanouids very accurately, with a mean absolute error
and a standard deviation of 2.60% and 3.11%, respectively. The
existing correlation can accurately reproduce the turbulent
convective heat transfer behavior of nanouids in helically coiled
tubes by adopting the properties of the nanouids in the analysis.
The nanouid density, specic heat, dynamic viscosity and thermal
conductivity were calculated by Eqs. (2) and (3), the Einstein
equation [19], and the Maxwell equation [20], respectively.
No abnormal heat transfer enhancement exists in our case
because of small nanoparticle/uid slip ow. The liquid around the
nanoparticles can be regarded as a continuumbecause the Knudsen
number, Kn, which is dened as the ratio of the water molecule
mean free path to the nanoparticle diameter (l/d
p
0.3/40), is
relatively small. Nanoparticle rotation can be ignored due to the
very lowrotational Peclet number which was developed by Ahuja
[30] to evaluate particle rotation under the effect of shear stress.
Buongiorno [3] stated that Brownian diffusion and thermophoresis
may become important as slip mechanisms. Thus, these two slip
mechanisms were checked. For a 40-nm alumina nanoparticle and
a temperature of 300 K, the magnitudes of the time a nanoparticle
needs to diffuse a length equal to its diameter for Brownian diffu-
sion and thermophoresis can be estimated to 10
4
and 10
1
s,
respectively, which are much longer than that for turbulent
transport, 10
7
s [3]. A temperature gradient of 10
4
K/m was esti-
mated in our case to calculate the thermophoretic velocity. Due to
the prepared homogeneous nanouid and negligible Brownian
diffusion and thermophoresis during experiments, diffusiophoresis
can also be ignored. Therefore, turbulent transport occurs without
any slip effects and nanoparticles move homogeneously with the
base uid. It is concluded that the tested nanouids can be treated
as homogeneous uids. Additional effects of nanoparticles, e.g.,
Brownian motion, thermophoresis and diffusiophoresis, on the
convective heat transfer characteristics of the nanouids are
negligible compared to the dominant thermophysical properties of
the nanouids.
3.4. Figure of merit
A gure of merit r h
n
/h
f
for the heat transfer coefcient ratio of
the nanouid over the base uid is adopted to compare the heat
transfer performance of the nanouid to that of the base uid, as
given in Yu et al. [31]. If r >1, the nanouid is benecial for the heat
transfer coefcient. Because water and the ve tested nanouids
can be accurately reproduced by Eq. (9) for the laminar owand the
Seban and McLaughlin correlation [27] for the turbulent ow, the
gure of merit can be obtained based on Eq. (9) and the Seban and
McLaughlin correlation [27]. It should be noted that the appropriate
property equations mentioned above in Section 2.2 should be used.
Different comparison bases can be used to obtain the gure of
merit, such as constant Reynolds number basis and constant ow
velocity basis [31]. However, the constant Reynolds number basis
can be misleading because the net result for the constant Reynolds
number comparison is a combination of the nanouid property
effect and the owvelocity effect. Due to the higher viscosity of the
nanouid, the ow velocity in the nanouid is generally higher
than that of the base uid at the same Reynolds number, which
provides an advantage for the nanouid over the base uid. If the
base uid is to be pumped at the same ow velocity as the nano-
uids, it may approach or exceed the thermal performance of the
nanouid. The result based on constant Reynolds number will be
more misleading at higher estimated or measured relative viscos-
ity. This misleading effect can be seen clearly in Fig. 10. Over 40%
Fig. 8. Comparison of Eq. (9) with two existing correlations.
Table 3
Description of three existing heat-transfer correlations.
Authors Correlations
Laminar ow
Dravid et al. [28] Applicable range: 50 < De
b
< 2000, 5 < Pr
b
< 175.
Nu
b
0:65
De
b
_
0:76$Pr
0:175
b
Kalb and Seader [29] Applicable range: 80 < De
b
< 1200, 0.7 < Pr
b
< 5.
Nu
b
0:913$De
0:476
b
$Pr
0:200
b
Turbulent ow
Seban and
McLaughlin [27]
Applicable range: 6000 < Re
b
< 65,600, 2.9 < Pr
b
<
5.7, D
c
/d 17, 104
Nu
b
0:023$Re
0:85
b
Pr
0:4
b
d=Dc
0:1
Table 4
Evaluation of Eq. (9) and two existing correlations for laminar ow.
Fluid type Dravid et al.
[28]
Kalb and
Seader [29]
Eq. (9)
e
A
a
s
N
b
e
A
s
N
e
A
s
N
Water 15.2 12.8 13.2 15.4 2.27 2.87
Nanouid 0.78% wt. 16.7 13.4 14.1 16.2 2.24 3.29
Nanouid 2.18% wt. 18.4 13.9 13.5 15.7 2.18 3.06
Nanouid 3.89% wt. 18.2 13.5 13.1 16.1 1.95 2.68
Nanouid 5.68% wt. 16.2 12.2 12.5 14.6 2.58 4.10
Nanouid 7.04% wt. 18.0 13.8 13.4 16.6 2.75 4.06
a
e
i
_
NuexpNucal
Nuexp
_
$100; e
A
1
NP
$
NP
i 1
NuexpNucal
Nuexp
$100.
b
s
N
N
P
i1
ei eR
2
Np1
_
; e
R
1
NP
$
NP
i 1
_
NuexpNucal
Nuexp
_
$100.
Fig. 9. Nu
b
(Pr
b
)
0.4
vs. Re
b
for turbulent ow.
Z. Wu et al. / Applied Thermal Engineering 60 (2013) 266e274 272
heat transfer enhancement can be obtained for the 7.04% wt.
nanouid compared to water for the constant Reynolds number
basis, as shown in Fig. 10a. The viscosity in the Reynolds number of
Fig. 10a was calculated by the Williams et al. equation [12] instead
of the Einstein equation [19] used in the above analysis. The Wil-
liams et al. equation [12] gives a larger viscosity increase for
nanouids than the Einstein equation [19], which induces a more
obviously misleading result of the heat transfer enhancement.
However, from Fig. 10b at the same ow velocity, the heat transfer
enhancement of nanouids over the base uid is much less than
that of the constant Reynolds number. Thus, the anomalous heat
transfer enhancement shown in Fig. 10a is just an analysis artifact.
The method based on the constant Reynolds number comparison
should not be used.
Based on the constant ow velocity basis, the following equa-
tions for evaluation of heat transfer enhancement can be generated
from Eq. (9) and the Seban and McLaughlin [27] correlation:
h
n
h
f
_
r
n
r
f
_
0:775
_
k
n
k
f
_
0:6_
m
f
m
n
_
0:375
_
c
pn
c
pf
_
0:4
(10)
for laminar ow, and
h
n
h
f
_
r
n
r
f
_
0:85
_
k
n
k
f
_
0:6_
m
f
m
n
_
0:45
_
c
pn
c
pf
_
0:4
(11)
for turbulent ow. The gure of merit h
n
/h
f
is a function of
the properties of the nanouid and water. Thus, heat transfer
enhancement of the nanouid relative to the base uid depends on
temperature, base uid type, nanoparticle properties, nanoparticle
concentrations and other factors that affect the nanouid proper-
ties, such as surfactants, nanoparticle agglomeration, etc. According
to Eq. (10), the enhancement of the inner tube heat transfer coef-
cient will be 0.37%, 1.03%, 1.93%, 2.67% and 3.38% for the 0.78% wt.,
2.18% wt., 3.89% wt., 5.68% wt. and 7.04% wt. alumina/water
nanouid at an average bulk temperature of 20
C for laminar ow,
respectively. For turbulent ow at the average bulk temperature of
20
C, according to Eq. (11), the heat transfer enhancement will be
0.37%, 1.05%, 1.92%, 2.72% and 3.43% for the 0.78% wt., 2.18% wt.,
3.89% wt., 5.68% wt. and 7.04% wt. alumina/water nanouid,
respectively, when compared with water at the same ow velocity.
These low values of enhancement are not very clear in Fig. 10b, due
to measurement uncertainties and the enhancement decrease
induced by the reduction of secondary ow intensity by nano-
particles. Secondary ow intensity mitigation may neutralize the
benet from the thermal conductivity increase. Overall, the
enhancement is not very promising for heat transfer purposes
when the pressure drop penalty and the cost of nanoparticles need
to be considered. To achieve high heat transfer enhancement,
development of newnanoparticle materials which can enhance the
thermal conductivity greatly and increase the viscosity slowly or
even reduce viscosity simultaneously will be very promising.
The results of this study differ from those of previous studies by
Akhavan-Behabadi et al. [14] and Mukesh Kumar et al. [15], which
may be attributed to the following reasons. The main reason is that
the heat transfer enhancement comparison was based on constant
Reynolds number or constant Dean number in Akhavan-Behabadi
et al. [14] and Mukesh Kumar et al. [15], which, as stated above,
may give misleading results. The other reason might be that
different base uids, nanoparticles and nanouid preparation
methods used in the above investigations presented different ef-
fects on physical properties and secondary ow eld, therefore
giving rise to different results.
4. Conclusions
Heat transfer characteristics of tap water and ve alumina
nanouids with 0.78% wt., 2.18% wt., 3.89% wt., 5.68% wt. and 7.04%
wt. nanoparticle concentrations were experimentally investigated
for both laminar ow and turbulent ow inside a double-pipe he-
lically coiled heat exchanger. Effect of nanoparticles on the critical
Reynolds number is negligible. By using Eqs. (2) and (3), the Ein-
stein equation [19] and the Maxwell equation [20] to calculate
nanouid properties at the average bulk temperature, tap water
and the ve nanouids showed similar pressure drop and heat
transfer performances, which indicates that the net effect of
nanoparticles on the heat transfer performance in helically coiled
tubes is probably insignicant. Secondary ow intensity mitigation
due to the larger viscosity and density of the nanouid may
neutralize the benet from the thermal conductivity increase. A
new correlation was developed for laminar ow for water and
nanouid owing inside helically coiled tubes, which can predict
the experimental data very accurately. The applicable range of Eq.
(9) is: 100 < De
b
< 1300, 4.0 < Pr
b
< 7.0, F < 2.0%. For turbulent
ow, the Seban and McLaughlin correlation [27] can predict the
thermal behavior of water and the ve nanouids very accurately
using the properties of nanouids. For both laminar ow and tur-
bulent ow, no anomalous heat transfer enhancement was found.
The heat transfer enhancement of the ve nanouids over tap
water ranges from 0.37% to 3.43% for the constant ow velocity
basis for both laminar and turbulent ows. Figure of merit based on
the constant Reynolds number can be misleading and should not be
used to evaluate heat transfer enhancement because the net result
Fig. 10. Heat transfer coefcient comparisons: (a) h vs. Re
b
for the constant Reynolds
number basis, with dynamic viscosity calculated by the Williams et al. equation [12],
and (b) h vs. u for the constant ow velocity basis.
Z. Wu et al. / Applied Thermal Engineering 60 (2013) 266e274 273
for the constant Reynolds number comparison is a combination of
the nanouid property effect and the ow velocity effect. Possible
additional effects of nanoparticles, e.g., Brownian motion, ther-
mophoresis and diffusiophoresis, on the convective heat transfer
characteristics of the nanouids were negligible compared to the
dominant thermophysical properties of the nanouids. No multi-
phase phenomenon was found and the tested alumina nanouids
can be treated as homogeneous uids.
Acknowledgements
Financial support from the Swedish Energy Agency is gratefully
acknowledged. Special thanks to Mr. Ingjald Andreasson for his
work on the experimental rig.
References
[1] T.J. Rennie, G.S. Raghavan, Numerical studies of a double-pipe helical heat
exchanger, Applied Thermal Engineering 26 (2005) 1266e1273.
[2] Y. Mori, W. Nakayama, Study on forced convective heat transfer in curved
pipes, International Journal of Heat and Mass Transfer 8 (1965) 67e82.
[3] J. Buongiorno, Convective transport in nanouids, ASME Journal of Heat
Transfer 128 (2006) 240e250.
[4] R. Sarkar, K.Y. Leong, H.A. Mohammad, A review on applications and chal-
lenges of nanouids, Renewable and Sustainable Energy Reviews 15 (2011)
1646e1668.
[5] S.M.S. Murshed, K.C. Leong, C. Yang, A combined model for the effective
thermal conductivity of nanouids, Applied Thermal Engineering 29 (2009)
2477e2483.
[6] A.S. Dalkilic, N. Kayaci, A. Celen, M. Tabatabaei, O. Yildiz, W. Daungthongsuk,
S. Wongwises, Forced convective heat transfer of nanouids e a review of the
recent literature, Current Nanoscience 8 (2012) 949e969.
[7] G. Huminic, A. Huminic, Application of nanouids in heat exchangers: a
review, Renewable and Sustainable Energy Reviews 16 (2012) 5625e5638.
[8] R. Taylor, S. Coulombe, T. Otanicar, P. Phelan, A. Gunawan, W. Lv, G. Rosengarten,
R. Prasher, H. Tyagi, Small particles, big impacts: a review of the diverse appli-
cations of nanouids, Journal of Applied Physics 113 (2013) 011301.
[9] A. Sergis, Y. Hardalupas, Anomalous heat transfer modes of nanouids: a
review based on statistical analysis, Nanoscale Research Letters 6 (2011) 391.
[10] Y. Xuan, Q. Li, Investigation on convective heat transfer and ow feature of
nanouids, ASME Journal of Heat Transfer 125 (2003) 151e155.
[11] E.V. Timofeeva, W. Yu, D.M. France, D. Singh, J.L. Routbort, Nanouids for heat
transfer: an engineering approach, Nanoscale Research Letters 6 (2011) 182.
[12] W. Williams, J. Buongiorno, L.W. Hu, Experimental investigation of turbulent
convective heat transfer and pressure loss of alumina/water and zirconia/
water nanoparticle colloids (nanouids) in horizontal tubes, ASME Journal of
Heat Transfer 130 (2008) 042412.
[13] W. Yu, D.M. France, E.V. Timofeeva, D. Singh, J.L. Routbort, Convective heat
transfer of nanouids in turbulent ow, in: Carbon Nano Materials and
Applications Workshop, South Dakota, Oct. 30eNov. 1, 2011.
[14] M.A. Akhavan-Behabadi, M.F. Pakdaman, M. Ghazvini, Experimental investi-
gation on the convective heat transfer of nanouid ow inside vertical heli-
cally coiled tubes under uniform wall temperature condition, International
Communications in Heat and Mass Transfer 39 (2012) 556e564.
[15] P.C. Mukesh Kumar, J. Kumar, S. Suresh, K. Praveen Babu, Heat transfer
enhancement in a helically coiled tube with Al
2
O
3
/water nanouid under
laminar ow condition, International Journal of Nanoscience 11 (2012)
1250029.
[16] H.A. Mohammed, K. Narrein, Thermal and hydraulic characteristics of nano-
uid ow in a helically coiled tube heat exchanger, International Communi-
cations in Heat and Mass Transfer 39 (2012) 1375e1383.
[17] K. Narrein, H.A. Mohammed, Inuence of nanouids and rotation on heli-
cally coiled tube heat exchanger performance, Thermochimica Acta 564
(2013) 13e23.
[18] A.P. Sasmito, J.C. Kurnia, A.S. Mujumdar, Numerical evaluation of laminar heat
transfer enhancement in nanouid ow in coiled square tubes, Nanoscale
Research Letters 6 (2011) 376.
[19] A. Einstein, Eine neue bestimmung der molekul-dimension (A new determi-
nation of the molecular dimensions), Annalen der Physik 19 (1906) 289e306.
[20] J.C. Maxwell, A Treatise on Electricity and Magnetism, second ed., Clarendon
Press, Oxford, UK, 1881.
[21] S.E. Maiga, S.J. Palm, N.C. Tam, R. Gilles, G. Nicolas, Heat transfer enhance-
ments by using nanouids in forced convection ows, International Journal of
Heat and Fluid Flow 26 (2005) 530e546.
[22] D.A. Drew, S.L. Passman, Theory of Multicomponent Fluids, Springer, Berlin,
1999.
[23] D. Wen, Y. Ding, Experimental investigation into convective heat transfer of
nanouids at entrance region under laminar ow conditions, International
Journal of Heat and Mass Transfer 47 (2004) 5181e5188.
[24] Z. Zhang, H. Gu, M. Fujii, Effective thermal conductivity and thermal diffusivity
of nanouids containing spherical and cylindrical nanoparticles, Experimental
Thermal and Fluid Science 31 (2007) 5593e5599.
[25] B. Sundn, Introduction to Heat Transfer, WIT Press, Southampton, 2012.
[26] H. Ito, Laminar ow in curved pipes, Applied Mathematics and Mechanics 11
(1969) 653e663.
[27] R.A. Seban, E.F. McLaughlin, Heat transfer in tube coils with laminar
and turbulent ow, International Journal of Heat and Mass Transfer 6 (1963)
387e395.
[28] A.N. Dravid, K.A. Smith, E.W. Merrill, P.L.T. Brian, Effect of secondary uid
motion on laminar ow heat transfer in helically coiled tubes, Advances in
Chemical Engineering 17 (1971) 1114e1122.
[29] C.E. Kalb, J.D. Seader, Heat and mass transfer phenomena for viscous ow in
curved circular tubes, International Journal of Heat and Mass Transfer 15
(1972) 801e817.
[30] A. Ahuja, Augmentation of heat transport in laminar ow of polystyrene
suspensions, Journal of Applied Physics 46 (1975) 3408e3425.
[31] W. Yu, D.M. France, E.V. Timofeeva, D. Singh, J.L. Routbort, Thermophysical
property-related comparison criteria for nanouid heat transfer enhancement
in turbulent ow, Applied Physics Letters 96 (2010) 213109.
Z. Wu et al. / Applied Thermal Engineering 60 (2013) 266e274 274