Journal of Thermal Analysis and Calorimetry

https://doi.org/10.1007/s10973-019-08838-w (0123456789().,-volV)(0123456789().,-volV)

Thermal conductivity prediction of nanofluids containing CuO

nanoparticles by using correlation and artificial neural network
Ali Komeilibirjandi1 • Amir Hossein Raffiee2 • Akbar Maleki3 • Mohammad Alhuyi Nazari4 •

Mostafa Safdari Shadloo5

Received: 18 July 2019 / Accepted: 13 September 2019

 Akadémiai Kiadó, Budapest, Hungary 2019

Abstract
Nanofluids are employed in different thermal devices due to their enhanced thermophysical features which lead to
noticeable heat transfer augmentation. One of the major reasons of the heat transfer improvement by using the nanofluids is
their increased thermal conductivity. Several methods have been applied to estimate this property of nanofluids such as
correlations and artificial neural networks (ANNs). In the present paper, group method of data handling (GMDH) and a
mathematical correlation are proposed for forecasting the thermal conductivity of nanofluids containing CuO nanoparticles.
The inputs of the both models are the base fluids’ thermal conductivities, concentration, temperature and nanoparticle
dimension. Comparison of the forecasted data by these two approaches revealed more favorable performance of GMDH.
The values of R-squared in the cases where polynomial and ANN were utilized were 0.9862 and 0.9996, respectively.
Moreover, the average absolute relative deviation values were 5.25% and 0.881% for the indicated methods, respectively.
According to these statistical values, it is concluded that employing the ANN-based regression leads to more confident
model for forecasting the TC of the nanofluids containing CuO nanoparticles.

Keywords Nanofluid  GMDH  Thermal conductivity  Artificial neural network

List of symbols yexperimental

i
Measured data in experiment
ai ith coefficient ypredicted Predicted value by the model
i
x Input vector
x1 Temperature of nanofluid (C)
Abbreviations
x2 Volume fraction of solid phase (%)
AARD Average absolute relative deviation
x3 Size of particles (nm)
ANN Artificial neural network
x4 Thermal conductivity of the base fluid
GMDH Group method of data handling
(W m-1 K-1)
MSE Mean square error
y Output vector
RD Relative deviation

& Akbar Maleki Introduction

akbar.maleki20@yahoo.com; a_maleki@shahroodut.ac.ir;
a_maleki@ut.ac.ir
Nanofluids have attracted the researchers’ attention in the
1
Department of Civil, Geo and Environmental Engineering, field of thermal engineering due to their ability in
Technical University of Munich, Munich, Germany improvement of heat transfer [1–4]. Their ability in aug-
2
School of Mechanical Engineering, Purdue University, mentation of heat transfer is attributed to some reasons
West Lafayette, IN, USA such as increment in thermal conductivity (TC) in single
3
Faculty of Mechanical Engineering, Shahrood University of phase and nucleation site increase in boiling heat transfer
Technology, Shahrood, Iran [5–7]. For instance, Ramezanizadeh et al. [8] added nickel
4
Department of Renewable Energies, Faculty of New Science nanoparticles into the working fluid of a thermosyphon and
and Technologies, University of Tehran, Tehran, Iran observed reduction in its thermal resistance. Improvement
5
CORIA-UMR 6614—Normandie University, CNRS in the performance of the thermosyphon was mainly
University and INSA of Rouen, 76000 Rouen, France

123
 A. Komeilibirjandi et al.

attributed to the increment in both TC and nucleation sites. to acceptable forecasting in some cases, considering the
Similar conclusion was drawn by Nazari et al. [9] in the size of particles leads to the increase in the accuracy of the
case of using graphene oxide nanofluid in a pulsating heat proposed model.
pipe. In addition to two-phase thermal mediums, nanofluids Different mathematical approaches are used in order to
have shown the ability of heat transfer augmentation in forecast and model the properties of nanofluids [32, 33].
single-phase devices [10–12]. For instance, Shirzad et al. Artificial intelligence is a powerful tool for various goals
[13] numerically investigated the performance of a pillow such as optimization and modeling the complex systems
plate heat exchanger filled with various nanofluids and [34–37]. Intelligence models are applicable for predicting
concluded that applying nanofluids can increase the heat the dependent values by using appropriate inputs with
transfer coefficient. In addition to the cooling appliances, perfect precision [38–41]. These models are useful for wide
nanofluids are used in different energy systems as shown in variety of applications such as predicting the thermal per-
Fig. 1. Employing nanofluids in renewable energy tech- formance of heat exchangers and energy systems
nologies leads to higher energy extraction, size reduction [10, 42, 43], energy consumption prediction [44], carbon
and reliability improvement [14–16]. dioxide emission [45], weather forecasting [44], etc. In
TC, specific heat and dynamic viscosity are among the addition, these approaches are able to forecast nanofluids’
factors with noticeable impact on the heat transfer ability TC and dynamic viscosity with acceptable precision
of the fluids [23, 24]. Due to the importance of TC in the [33, 46, 47]. Ahmadi et al. [48] employed LSSVM-GA
enhancement of heat transfer, it is crucial to distinguish the method to estimate the TC of Al2 O3 /EG nanofluid. The
factors influencing this property. According to the literature forecasted values by the proposed model were very close to
review, size and concentration of solid particles, tempera- the measured TCs, and the R-squared was 0.9902. In
ture and the type of base fluid are among the most addition to the SVM-based methods, artificial neural net-
important factors [25–27]. In addition to the mentioned works (ANNs) are appropriate tools for prediction of
ones, method of preparation, shape of the solid structures nanofluids’ TC. In a study [29], TC ratio of CuO/EG was
and the pH of the nanofluids affect their thermophysical predicted by utilizing GMDH and LSSVM-GA. The R-
and rheological features [25]. Generally, temperature and squared values of the mentioned approaches were 0.994
concentration increment results in improved TC. More- and 0.991, respectively. In addition to the high accuracy of
over, nanofluids with the more conductive base fluids have the GMDH approach in forecasting the behavior of com-
higher TC. In the majority of previous researches con- plex system, it has some features such as no requirement
ducted in the TC modeling of nanofluids, concentration and for predefinition of numbers of layers, neurons in hidden
temperature have been used as inputs [28]. In some recent layer and active neurons [49] which make it favorable for
studies, size of particles is added as another variable to predictive models. In addition, GMDH networks do not
achieve more comprehensive regression [29, 30]. Accord- suffer from the training data overfitting [50].
ing to the study performed by Ahmadi et al. [31], although The majority of the models represented for estimating
using the temperature and concentration as the inputs leads the TC are applicable for the nanofluids with a single type
of base fluid [51]. Proposing a model with the ability of
forecasting the TC for various base fluids would be very
useful for the researchers. In the current article, GMDH
Solar ANN, as an efficient algorithm, and a mathematical cor-
collectors relation, as an easy-to-use method, are employed to model
the TC of nanofluids with dispersed CuO particles to pro-
vide applicable models. In addition, the confidence of the
Solar water Fuel cell proposed models is compared on the basis of different
heater Applicaons cooling statistical criteria to produce detailed insight into the
of nanofluids employed forecasting models. The base fluids of the
in energy models nanofluids are water, ethylene glycol (EG) and
systems engine oil. The inputs of the models are concentration and
size of solid phase, TC of the base fluid and the tempera-
ture. All the data utilized for modeling are extracted from
PV cooling
Geothermal experimental studies.
facilies

Fig. 1 Main applications of nanofluids in energy systems [14, 17–22]

123
Thermal conductivity prediction of nanofluids containing CuO nanoparticles by using correlation…

2 3
Methods 1 x1p x1p x21p x21p x1p x1p
6 x22p x22q x2q 7
A ¼ 41 x2p x2q x2p 5 ð4Þ
Mathematical correlations are employed to model the 1 xnp xnq x2np x2nq xnp xnq
systems and features of the materials. These types of
models have some advantages including simple structure One method to solve this matrix equation (Aa ¼ y) is using
and ease of utilization; however, due to their weakness in singular value decomposition (SVD) method. By using
considering the interaction of variables affecting the out- SVD method, the unknown coefficient a is calculated by
puts, they are not appropriate for complicated systems. Eq. 5 [52].
ANNs are presented as intelligence approaches for mod-  1
eling the systems with complex relationship between the a ¼ AT A AT y ð5Þ
inputs. In general, Voltera–Kolmogorov–Gabor (VKG) In Eq. 5, AT is the transpose of the matrix A. With this
polynomials (Eq. 1) can be used in order to model complex method, unknown coefficient, a; can be calculated in most
systems containing set of data with multiple inputs and one of the cases. If the matrix ðAT AÞ is non-invertible,
output [52]. Tikhonov method is applied to solve the equation.
Xn n X
X n n X
X n X n
In the design of GMDH neural network, the aim is to
y ¼ a0 þ ai x i þ aij xi xj þ aijk xi xj xj
prevent the divergence of the network and relate shape and
i¼1 i¼1 j¼1 i¼1 j¼1 k¼1
þ  structure of the network to one or some numerical
parameters in a way that by changing these parameters, the
ð1Þ network structure changes as well. Evolution methods such
In Eq. 1, x ¼ ðx1 ; x2 ; . . .; xn Þ are input vectors, y is model as genetic algorithm have a vast application in different
output, and ai are coefficients of polynomial. VKG poly- steps of designing the neural networks due to their unique
nomials are approximated by second-order polynomials. capabilities in finding the optimum values and investigat-
These second-order polynomials are constructed based on ing the possibility in unpredictable spaces. In this paper,
the dual combinations of network inputs. GMDH algorithm the genetic algorithm is used to design the shape of the
is proposed based on this idea as the learning method for neural network and determine its coefficients [53]. In order
modeling complex systems [52, 53]. to make the GMDH neural networks widely recognized,
GMDH neural network has a structure of a multilayered the constraint of using adjacent layer in constructing the
and feed-forward network containing set of neurons created next level must be excluded. This type of neural networks
from linkage of different input pairs via second-order is called GS in which to build new levels all the previous
polynomial. Each layer in this network is established from layers (as well as the input layer) are used [52, 53].
one or more processing units that each of them has two The criteria used in the current study for the evaluation
inputs and one output. These units actually play the role of of the regressions are R-squared, relative deviation (RD),
components constructing the model and are assumed as a mean square error (MSE) and average absolute relative
second-order polynomial as given in Eq. 2 [52]. deviation (AARD) which are defined as [55, 56]:
Pi¼n  experimental 2
y^n ¼ a0 þ a1 x1 þ a2 x2 þ a3 x1 x2 þ a4 x21 þ a5 x22 ð2Þ i¼1 y  y predicted
i i
R2 ¼ 1  P  2
Unknown parameters of GMDH algorithm are coefficients i¼n experimental
i¼1 yi  yexperimental
of the polynomial in Eq. 2. In order to calculate the output
y^i for each input vector x ¼ ðx1 ; x2 ; . . .; xn Þ based on Eq. 2, yexperimental
i  ypredicted
i
RDi ¼  100
the average of squared error (Eq. 3) must be minimized yexperimental
i
[54]. i¼n  2
1X
X
n MSE ¼ yexperimental  y predicted

e¼ ðy^i  yi Þ2 ð3Þ n i¼1 i i

 
i¼n  experimental 
i¼1
1X yi  ypredicted
i 
AARD ¼   100
To find the minimal of the error, partial derivative of Eq. 3 n i¼1  experimental
yi 
is used. By employing Eq. 2 in this partial derivative, a
matrix equation (Aa ¼ y) is achieved. In this equation, where yexperimental
i and ypredicted
i refer to the values of the data
a ¼ fa0 ; a1 ; a2 ; a3 ; a4 ; a5 g, y ¼ fy1 ; . . .; ym gT , and matrix A obtained in experimental studies and determined by the
is given in Eq. (4) [54]. model, respectively.

123
 A. Komeilibirjandi et al.

Results and discussion applicable for different nanofluids, the TC of the base fluids
must be considered as another input. Generally, for the
Since the performance of a model in forecasting the output same concentration, temperature and size of particles, the
significantly depends on the input data, the variables used nanofluids would have higher TC in the cases where the
as inputs must be carefully defined. The majority of the particles are suspended in more conductive fluids (as
researches carried out to model the TC of nanofluids have shown in Fig. 3).
focused on a single type of nanofluid [57]. In these types of In the present research, the data are extracted from
studies, the temperature and volume fraction of solid phase several experimental studies. The ranges of the inputs and
have been used as the inputs due to the noticeable depen- the base fluids are given in Table 1. The base fluids of the
dency of TC on them [58, 59]. As shown in Fig. 2, incre- investigated nanofluids are ethylene glycol (EG), water and
ment in both of these factors leads to TC enhancement. engine oil. In order to consider the impact of the base
Another factor considered in the recent studies, in order to fluids’ TC on the outputs of the model, their values for each
reach more favorable precision and comprehensiveness, is fluids at 20 C are used as one of the inputs. Moreover,
the dimension of particles suspended in the base fluid. In temperature, volume fraction and the size of solid phase are
addition, since the present models are designed to be the other inputs. In the first step, a model is proposed by

(a) (b)

0.31
Thermal conductivity/W m–1 K–1
0.85
Thermal conductivity/W m–1 K–1

0.3
0.8
0.29

0.75 0.28

0.27
0.7
0.26
0.65
0.25

0.24
2 2

Vo 1.5 60 Vo 1.5 60
lu 50 lum 50
me 1 40 ec 1
co on 40
nc 30
/°C ce 30
en 0.5
ture ntr 0.5 tur e/°C
tra pera pera
20 ati 20
tio 0 on
n/% 10 Tem /%
0
10 Tem

(c)
0.18
Thermal conductivity/W m–1 K–1

0.17

0.16

0.15

0.14

0.13
2

Vo 1.5 60
lum 1 50
ec 40
on 30
cen 0.5 e/°C
r
tra
tion 0
20
pe ratu
/% 10
Tem

Fig. 2 Effects of concentration and temperature on the TC of a CuO/water, b CuO/EG, and c CuO/engine oil nanofluids [60]

123
Thermal conductivity prediction of nanofluids containing CuO nanoparticles by using correlation…

(a) 1 model, is equal to 0.31859. The obtained coefficients of the

0.9 CuO/Engine Oil CuO/Water CuO/EG abovementioned correlations are given in Table 2.
Thermal conducvity/W m–1 K–1

One of the conventional criteria used for the assessment

0.8
of regression is R-squared value. The closeness of this
0.7 value to 1 means high accuracy of the model. In Fig. 4, the
0.6 R-squared of the polynomial regression is represented by
0.5
comparing the outputs and the measured TC in the exper-
imental studies. As shown in Fig. 4, the value of the R-
0.4
squared in the case where polynomial is utilized is equal to
0.3 0.9862. The majority of the predicted TC values are located
0.2 in the vicinity of the actual data, which means accept-
0.1
able accuracy of the regression.
In order to gain deeper insight, the relative deviation
0
0 0.5 1 1.5 2 (RD) of the regression for each data is shown in Fig. 5. As
Concentraon/Vol% it is can be observed, the maximum absolute RD (ARD) is
(b) 1
approximately 22.5%; however, most of the data are well
predicted and their ARD is lower than 5%. The high value
Thermal conducvity/W m–1 K–1

0.9 CuO/Engine Oil CuO/Water CuO/EG

of the maximum ARD can be due to the weakness of the
0.8
proposed polynomial regression in considering the com-
0.7
plex interactions of the inputs. This problem can be solved
0.6
by employing ANNs which have more advanced structure
0.5
and are able to more accurate forecasting.
0.4
As indicated in the previous sections, the mathematical
0.3 regression is not as accurate as the ANNs in modeling the
0.2 systems. In this regard, GMDH shell software is applied for
0.1 modeling the TC of the nanofluids with various base fluids
0
0 0.5 1 1.5 2
based on GMDH ANN approach. The obtained relation-
Concentraon/Vol% ships between the inputs and the TC of the considered
nanofluids are given in ‘‘Appendix’’. In Fig. 6, the data are
Fig. 3 TC of the nanofluids with various base fluids. a 10 C, b 50 C compared to determine the R-squared. In this case, the R-
squared is equal to 0.9996. Comparing the R-squared
Table 1 The variation ranges of the inputs [60–67] determined in the cases where polynomial regression and
Inputs Range GMDH are employed reveals the higher accuracy of the
regression by using the ANN.
Base fluids Engine oil, water and ethylene glycol Similar to the regression obtained by the polynomial, the
Temperature/C 10–70 values of RD are determined for each data index in the case
Volumetric concentration/% 0–14.725 of employing GMDH ANN for the regression, as shown in
Size/nm 18.6–55 Fig. 7. The outputs are closer to the actual data when the
GMDH ANN is used in comparison with the polynomial.
More favorable precision of the model by using the ANN
can be due to its more complex structure, which leads to
applying a polynomial. The structure of polynomial is
better consideration of inputs interaction. In this case, the
shown in Eq. 6.
maximum ARD is approximately 6.3% which is much
TC ¼ A  x1 þ B  x2 þ C  x3 þ D  x4 þ E  x21 lower than the similar value when the polynomial regres-
þ F  x22 þ G  x23  H  x24 þ I  x1  x2 þ J  x1 x3 sion is used. Moreover, the RD for the majority of the data
is in the bound of ± 2%. These values of RD reveal the
þ K  x1  x4 þ L  x2  x3 þ M  x2  x4
reliability and appropriate precision of the GMDH in
þ N  x3  x4 þ O forecasting the TC values of the nanofluids with CuO
ð6Þ nanoparticles.
In order to assess the overall performance of the pro-
where x1 , x2 , x3 and x4 are temperature, volume fraction,
posed regressions in modeling the TC of the nanofluids,
size and the TC of the base fluids, respectively. In this
average absolute relative deviation (AARD) values of the
equation, the value of O, as the constant of the proposed
models are compared. As shown in Fig. 8, AARD values

123
 A. Komeilibirjandi et al.

Table 2 Coefficients of the correlation

A = 0.012691 B = -0.01054 C = 0.008593 D = 0.728172 E = 1.75 9 10-6 F = -0.00035 G = 4.7 9 10-5
H = 0.13926 I = 0.00045 J = -0.00025 K = 0.004144 L = 0.000441 M = -0.00029 N = 0.002965

1 6
Outputs of the model/W m–1 K–1

R² = 0.9862
4
0.8

Relave deviaon/%
2
0.6
0
0 50 100 150 200 250
0.4 –2

–4
0.2
–6
0
0 0.2 0.4 0.6 0.8 1 –8
Experimental TC/W m–1 K–1 Data index

Fig. 4 Comparison of the TC values by using polynomial regression Fig. 7 Relative deviation of the regression by using the GMDH ANN

25 6

20
5
15
Relave deviaon/%

10 4
AARD/%

5
0 3
0 50 100 150 200 250
–5
2
– 10
– 15
1
– 20
– 25 0
Data index Polynomial regression GMDH ANN

Fig. 5 Relative deviation of the regression by using the polynomial Fig. 8 Comparison of AARD values for the correlation and ANN
regressions
1.2
Outputs of the model/W m–1 K–1

R² = 0.9996 studies focused on the other nanofluids. In addition to

1
AARD, MSE values for both regressions are calculated to
0.8 evaluate the models based on this criterion. The calculated
MSE values in the cases where the mathematical correla-
0.6
tion and GMDH ANN are employed are approximately
0.4 9:28  104 and 2:36  105 , respectively. According to
all of the considered statistical criteria, it is demonstrated
0.2
that applying the GMDH method results in more accurate
0 regression compared with the correlation. This means that
0 0.2 0.4 0.6 0.8 1 1.2
the predicted data by the ANN are closer to the values
Experimental TC/W m–1 K–1
determined by the correlation as shown in Fig. 9. The
Fig. 6 Comparison of the TC values by GMDH ANN accuracy of the ANN-based models is more obvious for the
nanofluids with higher TC values.
for the correlation and the ANN are approximately 5.25%
and 0.881%, respectively. These values reveal much more
perfection of the GMDH ANN in forecasting the TC of the
nanofluids, which is similar to the results of the previous

123
Thermal conductivity prediction of nanofluids containing CuO nanoparticles by using correlation…

Output of the models/W m–1 K–1 1 N323 ¼ 0:00163454  N350  5:08638

GMDH ANN
0.9  N350  N353  32:6077 þ ðN350Þ2  32:5894
0.8 Correlaon
þ N353  6:08958
0.7 p ffiffiffiffiffi pffiffiffiffiffi 2
N350 ¼ 0:293485  3
x4  1:64301 þ ð 3 x4 Þ ; 2:37782
0.6
þ N377  0:182398
0.5
N60 ¼ 0:0214562  N197  N77  4:09698
0.4
þ ðN197Þ2 2:28313 þ N77  0:854643
0.3
0.2
þ ðN77Þ2 1:96742
0.1
N77 ¼ 0:0235764  N332  0:243521
0.1 0.3 0.5 0.7 0.9  N332  N185  54:0984
Experimental data/W m–1 K–1
þ ðN332Þ2  26:1746 þ N185  1:40135
Fig. 9 Comparison the data obtained by the models and actual values þ ðN185Þ2 27:7079
N185 ¼ 0:000925156 þ N266  0:717883
Conclusions þ N266  N284  9:61885
 ðN266Þ2 5:14743 þ N284  0:272103
Due to the crucial role of nanofluids’ thermal conductivity  ðN284Þ2 4:43229
pffiffiffiffiffi
(TC) in their thermal behavior, it is important to propose N284 ¼ 0:0131173 þ 3 x2  0:0105856
p ffiffiffiffi
ffi
models with reliable performance in predicting this ther- þ 3 x2  N320  0:0680691
mophysical feature. In this article, the accuracies of group þ N320  0:782995 þ ðN320Þ2 0:140752
pffiffiffiffiffi
method of data handling (GMDH) artificial neural network N320 ¼ 1:18285 þ 3 x3  0:809801
pffiffiffiffiffi
(ANN) and a polynomial correlation in modeling the TC of þ 3 x3  N334  0:122827
nanofluids, containing CuO nanoparticles with three types pffiffiffiffiffi 2
 ð 3 x3 Þ 0:13057 þ N334  0:458137
of base fluids, were evaluated and compared. The þ ðN334Þ2 0:0931282
employed inputs for the both models were temperature, TC pffiffiffiffiffi
N266 ¼ 0:740175  3 x4  4:94909
of the base fluids, size and volume fraction of the nano- p ffiffiffiffi
ffi pffiffiffiffiffi 2
 3 x4  N335  14:4033 þ ð 3 x4 Þ 7:61967
sized solid phase. According to the values of statistical
þ N335  6:04715 þ ðN335Þ2 6:09104
criteria, including AARD and R-squared, employing
N335 ¼ 0:878954 þ N356  1:16337
GMDH ANN resulted in much more accuracy. The deter-
 N356  N380  0:405506
mined values of R-squared for the ANN and polynomial
þ ðN356Þ2 0:0176024 þ N380  3:83443
were 0.9996 and 0.9862, respectively. The values of
AARD for the mentioned methods were approximately  ðN380Þ2 4:01012
0.881% and 5.25%, respectively. N197 ¼ 0:947113 þ N380  4:34746
þ N380  N302  0:391864
 ðN380Þ2 4:90199 þ N302  0:896456
Appendix  ðN302Þ2 0:0559059
N302 ¼ 0:00439602  N339  1:49386
 N339  N357  67:8386
N156 ¼ 0:0638187 þ N250  2:30293
þ ðN339Þ2 37:3589 þ N357  2:47456
þ N250  N264  23:5634  ðN250Þ2 13:7886
þ ðN357Þ2 30:4492
 N264  1:68318  ðN264Þ2 9:33539
N339 ¼ 0:00924584 þ N345  0:716211
N264 ¼ 0:0432462 þ N377  0:0323191
þ N345  N375  0:256732
þ N377  N315  0:403573 þ N315  0:482274
þ ðN345Þ2 0:179385 þ N375  0:0486508
þ ðN315Þ2 0:363301
N129 ¼ 0:0617473  N344  N217  27:9366
N315 ¼ 0:806049 þ N345  0:801919
þ ðN344Þ2 13:1372 þ N217  1:40366
þ N345  N380  0:592981  ðN345Þ2 0:0493165
þ ðN217Þ2 14:3235
þ N380  3:74782  ðN380Þ2 4:28298
N250 ¼ 0:00466728  N335  0:446895
 N335  N357  27:6001 þ ðN335Þ2 15:9285
þ N357  1:43754 þ ðN357Þ2  11:6728

123
 A. Komeilibirjandi et al.

N217 ¼ 0:0372683  N298  2:21564 p ffiffiffiffiffi pffiffiffiffiffi 2

N331 ¼ 0:676945 þ 3
x4  2:13489  ð 3 x4 Þ 0:948832
 N298  N299  3:82076  N334  0:303469 þ ðN334Þ2 0:92214
þ ðN298Þ2 3:83466 þ N299  2:97188 N286 ¼ 0:0171763 þ N332  0:675503 þ N332  N376  0:481898
þ ðN299Þ2 0:242383 þ ðN332Þ2 0:116863 þ N376  0:0098905
N298 ¼ 0:00461328  N338  0:605155 N376 ¼ 3:96757  1014 þ N377  1
 N338  N357  37:0172 N332 ¼ 0:254285 þ N334  1:1415  N334  N380  0:346046
þ ðN338Þ2 20:7828 þ N357  1:65003 þ ðN357Þ2 16:1369 þ N380  0:997576  ðN380Þ2 0:884078
pffiffiffiffiffi pffiffiffiffiffi pffiffiffiffiffi pffiffiffiffiffi pffiffiffiffiffi
N357 ¼ 0:224904  3 x4  1:55475  3 x4  N375  0:51028 N334 ¼ 0:665183  3 x1  0:151091 þ 3 x1  3 x4  0:212101
pffiffiffiffiffi 2 p ffiffiffiffi
ffi p ffiffiffiffi
ffi pffiffiffiffiffi 2
þ ð 3 x4 Þ 2:47322 þ N375  0:443408 2
þ ð 3 x1 Þ 0:00618198  3 x4  1:99001 þ ð 3 x4 Þ 2:16703
pffiffiffiffiffi pffiffiffiffiffi
N375 ¼ 1:3152 þ 3 x2  3 x3  0:0459091 N136 ¼ 0:0109802  N247  0:896005  N247  N346  19:3321
p ffiffiffiffi
ffi pffiffiffiffiffi 2
þ 3 x3  1:11126  ð 3 x3 Þ 0:186381 þ ðN247Þ2 11:8366 þ N346  1:81279 þ ðN346Þ2 7:56095
2
N338 ¼ 0:0129124 þ N345  N356  43:7902  ðN345Þ 21:8442 N346 ¼ 0:483901 þ N354  1:11989  N354  N379  0:267761
þ N356  0:911248  ðN356Þ2 21:7936 þ N379  2:11768  ðN379Þ2 2:224
pffiffiffiffiffi pffiffiffiffiffi
N344 ¼ 0:477016 þ N353  1:12265  N353  N379  0:265719 N379 ¼ 8:47397 þ 3 x3  3:10228  3 x3  N380  2:08197
pffiffiffiffiffi 2
þ N379  2:10893  ðN379Þ2 2:25038  ð 3 x3 Þ 0:336827 þ N380  16:7518  ðN380Þ2 9:85072
N353 ¼ 0:0220018 þ N356  0:994495 þ N377  0:0584774 N354 ¼ 1:37254  1013 þ N356  1
N48 ¼ 0:000454786  N69  N71  0:870543 pffiffiffiffiffi
N356 ¼ 0:325471 þ 3 x2  0:0275276
þ ðN69Þ2 0:868246 þ N71  1:00227 pffiffiffiffiffi pffiffiffiffiffi 2
 3 x4  1:60857 þ ð 3 x4 Þ 2:35855
N71 ¼ 0:0109597  N139  N214  50:1732 þ ðN139Þ2 25:6696 N247 ¼ 0:818308 þ N380  3:6748  ðN380Þ2 4:05446
2
þ N214  0:921561 þ ðN214Þ 24:5636 þ N329  1:06662  ðN329Þ2  0:060733
N214 ¼ 0:0144738 þ N331  1:08299  N331  N269  19:656 N329 ¼ 0:0216857 þ N345  0:60242 þ N345  N377  0:531008
þ ðN331Þ2 7:87523 þ ðN269Þ2 11:6828 þ ðN345Þ2 0:173633 þ N377  0:0189504
N269 ¼ 0:285722 þ N380  1:126  N380  N303  0:586986 pffiffiffiffiffi
N377 ¼ 0:0244881 þ 3 x1  0:0503244
p ffiffiffiffi
ffi p ffiffiffiffi
ffi pffiffiffiffiffi
 ðN380Þ2 1:0272 þ N303  1:25418  3 x1  3 x2  0:0191276 þ 3 x2  0:274573
pffiffiffiffiffi pffiffiffiffiffi pffiffiffiffiffi pffiffiffiffiffi pffiffiffiffiffi 2
N303 ¼ 0:00379619 þ 3 x2  0:0252342 þ 3 x2  N334  0:0472854 N345 ¼ 1:09654 þ 3 x3  3 x4  0:427947  ð 3 x3 Þ 0:0443035
þ N334  0:771192 þ ðN334Þ2 0:17568 p ffiffiffiffi
ffi p ffiffiffiffi
ffi 2
 3 x4  3:60702 þ ð 3 x4 Þ 2:68179
N69 ¼ 0:0119576 þ N273  1:07403  N273  N139  29:6595 p ffiffiffiffiffi pffiffiffiffiffi pffiffiffiffiffi
N380 ¼ 0:967148  x1  1:74116 þ 3 x1  3 x3  0:435543þ
3

þ ðN273Þ2 13:7071 þ ðN139Þ2 15:8416 pffiffiffiffiffi 2 pffiffiffiffiffi pffiffiffiffiffi 2

pffiffiffiffiffi pffiffiffiffiffi ð 3 x1 Þ 0:0204996 þ 3 x3  1:57949  ð 3 x3 Þ 0:452209:
N139 ¼ 1:31772  3 x4  7:24065  3 x4  N268  15:7399
pffiffiffiffiffi 2
þ ð 3 x4 Þ 9:82624 þ N268  6:84983 þ ðN268Þ2 6:10165
N268 ¼ 0:9162 þ N380  4:16699 þ N380  N336  0:265379
 ðN380Þ2 4:69022 þ N336  0:98972  ðN336Þ2 0:0976737
pffiffiffiffiffi pffiffiffiffiffi References
N336 ¼ 0:0171419 þ 3 x2  0:0106666 þ 3 x2  N345  0:0751379
2
þ N345  0:748366 þ ðN345Þ 0:170354
1. Mahian O, Kianifar A, Heris SZ, Wen D, Sahin AZ, Wongwises
N273 ¼ 0:0680299 þ N380  0:191049  N380  N304  0:643235
S. Nanofluids effects on the evaporation rate in a solar still
þ N304  1:20139 þ ðN304Þ2 0:0835203 equipped with a heat exchanger. Nano Energy. 2017;36:134–55.
N304 ¼ 0:0193152 þ N334  0:777182 þ N334  N377  0:252188 https://doi.org/10.1016/J.NANOEN.2017.04.025.
2. Ramezanizadeh M, Alhuyi Nazari M, Ahmadi MH, Açıkkalp E.
þ ðN334Þ2 0:109905 þ N377  0:0983817 Application of nanofluids in thermosyphons: a review. J Mol Liq.
N118 ¼ 0:0155467  N136  0:337627  N136  N223  22:2581 2018;272:395–402. https://doi.org/10.1016/J.MOLLIQ.2018.09.101.
þ ðN136Þ2 12:3103 þ N223  1:23063 þ ðN223Þ2 10:046 3. Mahian O, Kianifar A, Kleinstreuer C, Al-Nimr MA, Pop I, Sahin
AZ, et al. A review of entropy generation in nanofluid flow. Int J
N223 ¼ 0:0208803  N286  2:33259 þ N286  N299  41:8394
Heat Mass Transf. 2013;65:514–32. https://doi.org/10.1016/J.
 ðN286Þ2 15:7271 þ N299  3:48829  ðN299Þ2 26:2913 IJHEATMASSTRANSFER.2013.06.010.
N299 ¼ 0:0212474 þ N331  0:802107 þ N331  N376  0:275152 4. Karimipour A, D’Orazio A, Shadloo MS. The effects of different
nano particles of Al2O3 and Ag on the MHD nano fluid flow and
þ ðN331Þ2 0:0724277 þ N376  0:0897678
heat transfer in a microchannel including slip velocity and

123
Thermal conductivity prediction of nanofluids containing CuO nanoparticles by using correlation…

temperature jump. Phys E Low-Dimens Syst Nanostruct. 19. Natarajan E, Sathish R. Role of nanofluids in solar water heater.
2017;86:146–53. https://doi.org/10.1016/J.PHYSE.2016.10.015. Int J Adv Manuf Technol. 2009. https://doi.org/10.1007/s00170-
5. Gandomkar A, Saidi MH, Shafii MB, Vandadi M, Kalan K. 008-1876-8.
Visualization and comparative investigations of pulsating ferro- 20. Ebaid MSY, Ghrair AM, Al-Busoul M. Experimental investiga-
fluid heat. Appl Therm Eng. 2017. https://doi.org/10.1016/j.appl tion of cooling photovoltaic (PV) panels using (TiO2) nanofluid
thermaleng.2017.01.068. in water–polyethylene glycol mixture and (Al2O3) nanofluid in
6. Alhuyi Nazari M, Ahmadi MH, Ghasempour R, Shafii MB. How water–cetyltrimethylammonium bromide mixture. Energy Con-
to improve the thermal performance of pulsating heat pipes: a vers Manag. 2018;155:324–43. https://doi.org/10.1016/J.
review on working fluid. Renew Sustain Energy Rev. ENCONMAN.2017.10.074.
2018;91:630–8. https://doi.org/10.1016/j.rser.2018.04.042. 21. Meibodi SS, Kianifar A, Mahian O, Wongwises S. Second law
7. Safaei MR, Safdari Shadloo M, Goodarzi MS, Hadjadj A, analysis of a nanofluid-based solar collector using experimental
Goshayeshi HR, Afrand M, et al. A survey on experimental and data. J Therm Anal Calorim. 2016;126:617–25. https://doi.org/10.
numerical studies of convection heat transfer of nanofluids inside 1007/s10973-016-5522-7.
closed conduits. Adv Mech Eng. 2016;8:168781401667356. 22. Ehyaei MA, Ahmadi A, Assad MEH, Hachicha AA, Said Z.
https://doi.org/10.1177/1687814016673569. Energy, exergy and economic analyses for the selection of
8. Ramezanizadeh M, Alhuyi Nazari M, Ahmadi MH, Chau K. working fluid and metal oxide nanofluids in a parabolic trough
Experimental and numerical analysis of a nanofluidic ther- collector. Sol Energy. 2019;187:175–84. https://doi.org/10.1016/
mosyphon heat exchanger. Eng Appl Comput Fluid Mech. J.SOLENER.2019.05.046.
2019;13:40–7. https://doi.org/10.1080/19942060.2018.1518272. 23. Said Z, Rahman SMA, El Haj Assad M, Alami AH. Heat transfer
9. Nazari MA, Ghasempour R, Ahmadi MH, Heydarian G, Shafii enhancement and life cycle analysis of a shell-and-tube heat
MB. Experimental investigation of graphene oxide nanofluid on exchanger using stable CuO/water nanofluid. Sustain Energy
heat transfer enhancement of pulsating heat pipe. Int Commun Technol Assess. 2019;31:306–17. https://doi.org/10.1016/J.
Heat Mass Transf. 2018;91:90–4. https://doi.org/10.1016/j. SETA.2018.12.020.
icheatmasstransfer.2017.12.006. 24. Said Z, El Haj Assad M, Hachicha AA, Bellos E, Abdelkareem
10. Baghban A, Kahani M, Nazari MA, Ahmadi MH, Yan W-M. MA, Alazaizeh DZ, et al. Enhancing the performance of auto-
Sensitivity analysis and application of machine learning methods motive radiators using nanofluids. Renew Sustain Energy Rev.
to predict the heat transfer performance of CNT/water nanofluid 2019;112:183–94. https://doi.org/10.1016/J.RSER.2019.05.052.
flows through coils. Int J Heat Mass Transf. 2019;128:825–35. 25. Ahmadi MH, Mirlohi A, Alhuyi Nazari M, Ghasempour R. A
https://doi.org/10.1016/J.IJHEATMASSTRANSFER.2018.09. review of thermal conductivity of various nanofluids. J Mol Liq.
041. 2018. https://doi.org/10.1016/j.molliq.2018.05.124.
11. Hemmat Esfe M, Saedodin S, Mahian O, Wongwises S. Heat 26. Hemmat Esfe M, Saedodin S, Bahiraei M, Toghraie D, Mahian
transfer characteristics and pressure drop of COOH-functional- O, Wongwises S. Thermal conductivity modeling of MgO/EG
ized DWCNTs/water nanofluid in turbulent flow at low concen- nanofluids using experimental data and artificial neural network.
trations. Int J Heat Mass Transf. 2014;73:186–94. https://doi.org/ J Therm Anal Calorim. 2014;118:287–94. https://doi.org/10.
10.1016/J.IJHEATMASSTRANSFER.2014.01.069. 1007/s10973-014-4002-1.
12. Safaei M, Ahmadi G, Goodarzi M, Safdari Shadloo M, Goshayeshi H, 27. Hemmat Esfe M, Saedodin S, Mahian O, Wongwises S. Thermal
Dahari M, et al. Heat transfer and pressure drop in fully developed conductivity of Al2O3/water nanofluids. J Therm Anal Calorim.
turbulent flows of graphene nanoplatelets–silver/water nanofluids. 2014;117:675–81. https://doi.org/10.1007/s10973-014-3771-x.
Fluids. 2016;1:20. https://doi.org/10.3390/fluids1030020. 28. Ramezanizadeh M, Alhuyi Nazari M, Ahmadi MH, Lorenzini G,
13. Shirzad M, Ajarostaghi SSM, Delavar MA, Sedighi K. Improve Pop I. A review on the applications of intelligence methods in
the thermal performance of the pillow plate heat exchanger by predicting thermal conductivity of nanofluids. J Therm Anal
using nanofluid: numerical simulation. Adv Powder Technol. Calorim. 2019. https://doi.org/10.1007/s10973-019-08154-3.
2019;30:1356–65. https://doi.org/10.1016/J.APT.2019.04.011. 29. Ahmadi M-A, Ahmadi MH, Fahim Alavi M, Nazemzadegan MR,
14. Ramezanizadeh M, Alhuyi Nazari M, Hossein Ahmadi M, Chen Ghasempour R, Shamshirband S. Determination of thermal con-
L. A review on the approaches applied for cooling fuel cells. Int J ductivity ratio of CuO/ethylene glycol nanofluid by connectionist
Heat Mass Transf. 2019;139:517–25. https://doi.org/10.1016/J. approach. J Taiwan Inst Chem Eng. 2018. https://doi.org/10.
IJHEATMASSTRANSFER.2019.05.032. 1016/J.JTICE.2018.06.003.
15. Salavati Meibodi S, Kianifar A, Niazmand H, Mahian O, 30. Ahmadi MH, Alhuyi Nazari M, Ghasempour R, Madah H, Shafii
Wongwises S. Experimental investigation on the thermal effi- MB, Ahmadi MA. Thermal conductivity ratio prediction of
ciency and performance characteristics of a flat plate solar col- Al2O3/water nanofluid by applying connectionist methods. Col-
lector using SiO2/EG–water nanofluids. Int Commun Heat Mass loids Surf A Physicochem Eng Asp. 2018. https://doi.org/10.
Transf. 2015;65:71–5. https://doi.org/10.1016/J.ICHEATMAS 1016/j.colsurfa.2018.01.030.
STRANSFER.2015.02.011. 31. Ahmadi MH, Hajizadeh F, Rahimzadeh M, Shafii MB, Chamkha
16. Toghyani S, Afshari E, Baniasadi E, Shadloo MS. Energy and AJ. Application GMDH artificial neural network for modeling of
exergy analyses of a nanofluid based solar cooling and hydrogen Al2O3/water and Al2O3/ethylene glycol thermal conductivity. Int
production combined system. Renew Energy. 2019;141:1013–25. J Heat Technol. 2018;36:773–82.
https://doi.org/10.1016/J.RENENE.2019.04.073. 32. Sepyani K, Afrand M, Hemmat Esfe M. An experimental eval-
17. Ahmadi MH, Ramezanizadeh M, Nazari MA, Lorenzini G, uation of the effect of ZnO nanoparticles on the rheological
Kumar R, Jilte R. Applications of nanofluids in geothermal: a behavior of engine oil. J Mol Liq. 2017;236:198–204. https://doi.
review. Math Model Eng Probl. 2018;5:281–5. https://doi.org/10. org/10.1016/J.MOLLIQ.2017.04.016.
18280/mmep.050402. 33. Ahmadi Nadooshan A, Hemmat Esfe M, Afrand M. Evaluation of
18. Kasaeian A, Eshghi AT, Sameti M. A review on the applications of rheological behavior of 10W40 lubricant containing hybrid nano-
nanofluids in solar energy systems. Renew Sustain Energy Rev. material by measuring dynamic viscosity. Phys E Low-Dimens
2015;43:584–98. https://doi.org/10.1016/J.RSER.2014.11.020. Syst Nanostruct. 2017;92:47–54. https://doi.org/10.1016/J.
PHYSE.2017.05.011.

123
 A. Komeilibirjandi et al.

34. Zhang W, Maleki A, Rosen MA. A heuristic-based approach for approach. J Therm Anal Calorim. 2019;135:271–81. https://doi.
optimizing a small independent solar and wind hybrid power org/10.1007/s10973-018-7035-z.
scheme incorporating load forecasting. J Clean Prod. 2019. 49. Zendehboudi A, Saidur R, Mahbubul IM, Hosseini SH. Data-
https://doi.org/10.1016/J.JCLEPRO.2019.117920. driven methods for estimating the effective thermal conductivity
35. Maleki A, Khajeh MG, Rosen MA. Two heuristic approaches for of nanofluids: a comprehensive review. Int J Heat Mass Transf.
the optimization of grid-connected hybrid solar–hydrogen sys- 2019;131:1211–31. https://doi.org/10.1016/J.IJHEATMAS
tems to supply residential thermal and electrical loads. Sustain STRANSFER.2018.11.053.
Cities Soc. 2017;34:278–92. https://doi.org/10.1016/J.SCS.2017. 50. Pham DT, Liu X. Modelling and prediction using GMDH net-
06.023. works of Adalines with nonlinear preprocessors. Int J Syst Sci.
36. Maleki A, Hajinezhad A, Rosen MA. Modeling and optimal design 1994;25:1743–59. https://doi.org/10.1080/00207729408949310.
of an off-grid hybrid system for electricity generation using various 51. Ramezanizadeh M, Alhuyi Nazari M. Modeling thermal con-
biodiesel fuels: a case study for Davarzan, Iran. Biofuels. ductivity of Ag/water nanofluid by applying a mathematical
2016;7:699–712. https://doi.org/10.1080/17597269.2016.1192443. correlation and artificial neural network. Int J Low-Carbon
37. Malekan M, Khosravi A, Goshayeshi HR, Assad MEH, Garcia Technol. 2019. https://doi.org/10.1093/ijlct/ctz030.
Pabon JJ. Thermal resistance modeling of oscillating heat pipes 52. Zoqi M, Ghamgosar M, Ganji M, Fallahi S. Application of
for nanofluids by artificial intelligence approach. J Heat Transf. GMDH and genetic algorithm in fraction in biogas from landfill
2019. https://doi.org/10.1115/1.4043569. modeling. J Environ Sci Technol. 2016;18:1–12.
38. Maleki A, Khajeh MG, Rosen MA. Weather forecasting for 53. Ahmadi MH, Ahmadi MA, Mehrpooya M, Rosen MA. Using
optimization of a hybrid solar-wind–powered reverse osmosis GMDH neural networks to model the power and torque of a
water desalination system using a novel optimizer approach. stirling engine. Sustainability. 2015;7:2243–55. https://doi.org/
Energy. 2016;114:1120–34. https://doi.org/10.1016/J.ENERGY. 10.3390/su7022243.
2016.06.134. 54. Pourkiaei SM, Ahmadi MH, Hasheminejad SM. Modeling and
39. Gholipour Khajeh M, Maleki A, Rosen MA, Ahmadi MH. Electricity experimental verification of a 25 W fabricated PEM fuel cell by
price forecasting using neural networks with an improved iterative parametric and GMDH-type neural network. Mech Ind.
training algorithm. Int J Ambient Energy. 2018;39:147–58. https://doi. 2016;17:105. https://doi.org/10.1051/meca/2015050.
org/10.1080/01430750.2016.1269674. 55. Ramezanizadeh M, Ahmadi MA, Ahmadi MH, Alhuyi Nazari M.
40. Ahmadi Nadooshan A, Hemmat Esfe M, Afrand M. Prediction of Rigorous smart model for predicting dynamic viscosity of Al2O3/
rheological behavior of SiO2-MWCNTs/10W40 hybrid nanolu- water nanofluid. J Therm Anal Calorim. 2018. https://doi.org/10.
bricant by designing neural network. J Therm Anal Calorim. 1007/s10973-018-7916-1.
2018;131:2741–8. https://doi.org/10.1007/s10973-017-6688-3. 56. Ramezanizadeh M, Ahmadi MH, Nazari MA, Sadeghzadeh M,
41. Pakatchian MR, Saeidi H, Ziamolki A. CFD-based blade shape Chen L. A review on the utilized machine learning approaches for
optimization of MGT-70(3)axial flow compressor. Int J Numer modeling the dynamic viscosity of nanofluids. Renew Sustain
Methods Heat Fluid Flow. 2019. https://doi.org/10.1108/HFF-10- Energy Rev. 2019;114:109345. https://doi.org/10.1016/J.RSER.
2018-0603. 2019.109345.
42. Ahmadi MH, Tatar A, Alhuyi Nazari M, Ghasempour R, 57. Hemmat Esfe M, Motahari K, Sanatizadeh E, Afrand M, Rosta-
Chamkha AJ, Yan W-M. Applicability of connectionist methods mian H, Reza Hassani Ahangar M. Estimation of thermal con-
to predict thermal resistance of pulsating heat pipes with ethanol ductivity of CNTs-water in low temperature by artificial neural
by using neural networks. Int J Heat Mass Transf. 2018. https:// network and correlation. Int Commun Heat Mass Transf.
doi.org/10.1016/j.ijheatmasstransfer.2018.06.085. 2016;76:376–81. https://doi.org/10.1016/J.ICHEATMAS
43. Ahmadi MH, Sadeghzadeh M, Raffiee AH, Chau K. Applying STRANSFER.2015.12.012.
GMDH neural network to estimate the thermal resistance and 58. Adhami Dehkordi R, Hemmat Esfe M, Afrand M. Effects of
thermal conductivity of pulsating heat pipes. Eng Appl Comput functionalized single walled carbon nanotubes on thermal per-
Fluid Mech. 2019;13:327–36. https://doi.org/10.1080/19942060. formance of antifreeze: an experimental study on thermal con-
2019.1582109. ductivity. Appl Therm Eng. 2017;120:358–66. https://doi.org/10.
44. Zhang W, Maleki A, Rosen MA, Liu J. Sizing a stand-alone 1016/J.APPLTHERMALENG.2017.04.009.
solar-wind-hydrogen energy system using weather forecasting 59. Asadi A, Asadi M, Rezaniakolaei A, Rosendahl LA, Afrand M,
and a hybrid search optimization algorithm. Energy Convers Wongwises S. Heat transfer efficiency of Al2O3-MWCNT/ther-
Manag. 2019;180:609–21. https://doi.org/10.1016/J.ENCON mal oil hybrid nanofluid as a cooling fluid in thermal and energy
MAN.2018.08.102. management applications: an experimental and theoretical
45. Rezaei MH, Sadeghzadeh M, Alhuyi Nazari M, Ahmadi MH, investigation. Int J Heat Mass Transf. 2018;117:474–86. https://
Astaraei FR. Applying GMDH artificial neural network in mod- doi.org/10.1016/J.IJHEATMASSTRANSFER.2017.10.036.
eling CO2 emissions in four nordic countries. Int J Low-Carbon 60. Agarwal R, Verma K, Agrawal NK, Duchaniya RK, Singh R.
Technol. 2018;13:266–71. https://doi.org/10.1093/ijlct/cty026. Synthesis, characterization, thermal conductivity and sensitivity
46. Mohamadian F, Eftekhar L, Haghighi Bardineh Y. Applying of CuO nanofluids. Appl Therm Eng. 2016;102:1024–36. https://
GMDH artificial neural network to predict dynamic viscosity of doi.org/10.1016/j.applthermaleng.2016.04.051.
an antimicrobial nanofluid. Nanomed J. 2018;5:217–21. https:// 61. Liu M-S, Lin MC-C, Huang I-T, Wang C-C. Enhancement of
doi.org/10.22038/NMJ.2018.05.00005. thermal conductivity with CuO for nanofluids. Chem Eng Tech-
47. Hemmat Esfe M, Bahiraei M, Hajmohammad MH, Afrand M. nol. 2006;29:72–7. https://doi.org/10.1002/ceat.200500184.
Rheological characteristics of MgO/oil nanolubricants: experi- 62. Lee S, Choi SU-S, Li S, Eastman JA. Measuring thermal con-
mental study and neural network modeling. Int Commun Heat ductivity of fluids containing oxide nanoparticles. J Heat Transf.
Mass Transf. 2017;86:245–52. https://doi.org/10.1016/J. 1999;121:280. https://doi.org/10.1115/1.2825978.
ICHEATMASSTRANSFER.2017.05.017. 63. Wang X, Xu X, Choi SU S. Thermal Conductivity of nanopar-
48. Ahmadi MH, Ahmadi MA, Nazari MA, Mahian O, Ghasempour ticle—fluid mixture. J Thermophys Heat Transf. 1999;13:474–80.
R. A proposed model to predict thermal conductivity ratio of https://doi.org/10.2514/2.6486.
Al2O3/EG nanofluid by applying least squares support vector 64. Liu M, Lin M, Wang C. Enhancements of thermal conductivities
machine (LSSVM) and genetic algorithm as a connectionist with Cu, CuO, and carbon nanotube nanofluids and application of

123
Thermal conductivity prediction of nanofluids containing CuO nanoparticles by using correlation…

MWNT/water nanofluid on a water chiller system. Nanoscale Res 67. Khedkar RS, Sonawane SS, Wasewar KL. Influence of CuO
Lett. 2011;6:297. https://doi.org/10.1186/1556-276X-6-297. nanoparticles in enhancing the thermal conductivity of water and
65. Das SK, Putra N, Thiesen P, Roetzel W. Temperature dependence monoethylene glycol based nanofluids. Int Commun Heat Mass
of thermal conductivity enhancement for nanofluids. J Heat Transf. 2012;39:665–9. https://doi.org/10.1016/j.icheatmas
Transfer. 2003;125:567. https://doi.org/10.1115/1.1571080. stransfer.2012.03.012.
66. Mintsa HA, Roy G, Nguyen CT, Doucet D. New temperature
dependent thermal conductivity data for water-based nanofluids. Publisher’s Note Springer Nature remains neutral with regard to
Int J Therm Sci. 2009;48:363–71. https://doi.org/10.1016/J. jurisdictional claims in published maps and institutional affiliations.
IJTHERMALSCI.2008.03.009.

123