International Communications in Heat and Mass Transfer 92 (2018) 51–55

Contents lists available at ScienceDirect

International Communications in Heat and Mass Transfer

journal homepage: www.elsevier.com/locate/ichmt

Dispersion regimes in alumina/water-based nanofluids: Simultaneous T

measurements of thermal conductivity and dynamic viscosity
Nizar Bouguerra, Sébastien Poncet*, Said Elkoun
Mechanical Engineering Department, Université de Sherbrooke, 2500 Bd de l’Université, Sherbrooke, QC J1K 2R1, Canada

A R T I C L E I N F O A B S T R A C T

Keywords: The present work proposes a detailed experimental analysis of the dispersion regimes within nanofluids. For this
Alumina/water-based nanofluids purpose, simultaneous measurements of thermal conductivity and dynamic viscosity are performed for alumina/
Dispersion regimes water-based nanofluids at volumetric concentrations φ ranging from 0.2 to 2%. Depending on the pH values, five
Thermal conductivity dispersion regimes have been identified for intermediate values of φ. The well-dispersed regime characterized by
Dynamic viscosity
a local maximum of the thermal conductivity and an absolute minimum of the dynamic viscosity is not recovered
Chain-like aggregation
for φ = 2%, while the chain-like agglomeration regime is not observed for φ = 0.2%. The results are corrobo-
rated by the distribution of the Mouromtseff number, which appears as a reliable parameter to identify the most
efficient nanofluids. For an optimized nanofluid, comparable efficiency may be achieved even at low con-
centration φ = 0.2%.

1. Introduction Studies are constantly carried out to better understand the under-
lying mechanisms responsible for the improved thermal conductivity of
Nanofluids (Nfs) are innovative heat transfer fluids with improved nanofluids [11, 12, 13, 14, 15]. The lack of adequate experimental
thermophysical properties obtained by suspending nanoparticles (Nps) methods allowing to correlate the macroscopic effects to the nanoscale
into conventional fluids. A complete, unique and well-grounded defi- behavior of the particles makes this task hard. In fact, most of the works
nition regrettably does not exist to date. This may be due to the fact that are based on theoretical models and when experiments exist, they focus
the field is relatively new and some concepts, especially heat transfer much more on the stability indicators (zeta potential, absorbency,
enhancement mechanisms, are not yet fully understood. Nevertheless, particle size …) (see in [16, 17, 18]). Unfortunately, these techniques
numerous definitions can be found, most of them derive from the first based mainly on the light scattering method are only suitable for very
definition proposed by Choi and Eastman [1] and enriched in a re- low particle concentrations far from the concentrations used in real
visited version [2] in which the authors emerge dispersion as key ele- applications. It remains a challenge and a subject of intense debate to
ment. Due to their small size (< 100 nm), nanoparticles benefit from a identify clearly the heat transfer mechanisms in nanofluids [19]: higher
large surface/volume ratio (S/V) allowing them to provide much higher specific surface area of nanoparticles, dispersion and electric charge,
thermal conductivity (k) than that predicted by the classical theories [1, Brownian motion, interfacial nanolayer, collision between particles and
3, 4, 5, 6, 7]. However, due to the challenges associated with the pre- linear nanoclusters like chains that can produce thermal bridges within
paration of similar nanofluids, the literature results are often contra- the nanofluid. On the contrary, the scientific community agrees that
dictory. As an example, Buongiorno et al. [8] proposed an experimental nanoparticle aggregation plays a significant role in thermal trans-
benchmark on the thermal conductivity of various nanofluids. The port [18]. Clustering is a complex process that depends on the prop-
authors did not observe noticeable enhancement due to the inaccuracy erties of the host fluid and the surface energy of the particles. The
of some measurement techniques [9, 10]. Actually, some methods, electric charges in the solution govern the level of aggregation and
originally designed for solid thermal conductivity measurements, are consequently the configuration of the particles within the base fluid
indeed not appropriate for liquids, being very affected by natural con- which can lead to different dispersion regimes, namely: well dispersed
vection. The KD2Pro Thermal Property Analyzer is the most significant (WD), weakly attracted (WA), chain-like agglomeration (CL Agg.),
example. This apparatus is erroneously frequently presented as a partially agglomerated (P Agg.) and fully agglomerated (F Agg.) [19].
measuring system based on the transient hot-wire (THW) method while Each of the possible mechanisms behind thermal conductivity en-
it is not as reliable. hancement may be either accentuated or attenuated by these different

*
Corresponding author.
E-mail address: [email protected] (S. Poncet).

https://doi.org/10.1016/j.icheatmasstransfer.2018.02.015

0735-1933/ © 2018 Elsevier Ltd. All rights reserved.

N. Bouguerra et al. International Communications in Heat and Mass Transfer 92 (2018) 51–55

dispersion regimes. Controlling the agglomeration becomes then crucial present gap (5917.1 μm) by carrying out preliminary experiments on
for understanding the physics of nanofluids. Especially, always more distilled water. The results compare very well with the literature values
studies report that an optimum level of agglomeration may achieve a (uncertainty < 3%, repeatability < 0.5%). During the experiments on
maximum k enhancement by nanoclustering effect and that excessive nanofluids, ones evaluate μ of all samples as a function of γ̇. Whatever
particle clustering is unfavorable [18]. Many observations from scan- the operating conditions, the present nanofluids exhibit a Newtonian
ning electron microscopy (SEM) show the presence of chain-like ag- behavior over the range [0.01–100] s−1.
gregation [20, 21, 22]. Nevertheless, this technique cannot evaluate As explained above, one presents an experimental approach in-
clustering in real suspensions since it requires drying the base fluid. cluding a careful preparation step and the most appropriate measure-
Adding nanoparticles affect the other properties of the base fluid, ment techniques for alumina/water based nanofluids. Particular at-
particularly the dynamic viscosity (μ) [23]. Most of experimental stu- tention is paid to the synchronization of the thermal and rheological
dies deal with the thermal conductivity of colloids and few works ex- tests. The first step in the experimental procedure is the weighing of the
amine their viscosity, and when they exist, these data are rarely cou- nanoparticles. Colloids with an ultra-precise concentration are prepared
pled. The nanofluid viscosity is also very sensitive to agglomeration. by homogenizing particles in water using a magnetic stirrer. The ef-
Clustering has indeed an effect on the organization of the nanoparticles, fective dispersion is achieved out by applying ultrasonic vibrations.
by changing their average starting sizes, the particle size distribution, Using this method, an initial solution of 800 ml of nanofluid is sonicated
their shape and aspect ratio. As for k, each of these parameters has a for 12 h. Mahbubul et al. [27] showed that the sonification time has no
direct influence on μ [24]. As an example, Bhanushali et al. [25] in- longer influence on the viscosity of alumina/water-based nanofluids for
vestigated the effect of particle shape on the properties of nanofluids temperatures between 10 °C and 50 °C after 2 h. The required samples
using a range of distinct filler particle shapes. They concluded that volumes for THW and rheological measurements are 45 ml and 25 ml,
higher aspect ratio favors the thermal conductivity and is detrimental respectively. Therefore, the initial solution is divided into samples of
for the viscosity. Simultaneous measurements of k and μ could then 80 ml, the pH of each sample is adjusted to the desired value and a
inform us about the state of dispersion and bring response elements on sonication of 15 mn is carried out before each test. The homogenization
the heat transport mechanisms in nanofluids, which is the main ob- of the samples is performed with the Q700-Sonicator (Qsonica), while
jective of the present paper. Such experimental data, made simulta- the initial solution is sonicated using the UIP1000hdT (1000 W, Hiel-
neously on the same sample with accurate equipments, can also be scher Ultrasonics) which is more suited for large volumes. The next step
exploited to establish trustworthy global energy performance criteria is the calibration of the measuring instruments and the setting of the
allowing to select the proper nanofluids for use in energy systems. test temperature (25 °C). Once the test cell of each device is filled, si-
multaneous measurements of k and μ from the same sample can begin.
2. Experimental methods The above procedure is repeated for different pH values and volume
concentrations of nanoparticles (vol%). Temperature is fixed to 25 °C
The two-step method has demonstrated to be suitable to prepare for all measurements and controlled by incorporating a jacketed glass
oxide Nfs [2]. Alumina/water-based Nfs have been selected here for beaker connected to a thermostat bath circulator at all stages of pre-
their experimental flexibility unlike metallic Nfs which must absolutely paration.
be prepared with the one-step method. The Al2O3 nanoparticles are
purchased from US Research Nanomaterials and exhibit the following
properties: purity of 99.9%, averaged diameter of 50 nm. 3. Results and discussion
Water-based nanofluids allow a direct control on the degree of in-
teraction between nanoparticles by regulating the pH. The preparation Figs. 1–4 show the effective thermal conductivity (knf/kbf) and dy-
step requires indeed to use the appropriate surfactant, to control the pH namic viscosity (μnf/μbf) of the nanofluids as a function of pH for dif-
of the solution and to disperse the Nps by ultrasound techniques. In a ferent volume percent concentration (from 0.2% to 2%). Here, knf and
former study [26], a careful experimental investigation on the stability μnf are the thermal conductivity and dynamic viscosity, respectively, of
of Al2O3-water based nanofluids has identified the pH as a key para- the nanofluid whereas kbf and μbf are that of the base fluid. Examination
meter for dispersing and stabilizing suspensions. The pH is controlled of the results represented by all of these measuring points makes it
here using hydrochloric acid (HCl) and sodium hydroxide (NaOH) in possible to correlate the consequences of the agglomeration on k and μ.
analytical grade. Being temperature dependent, it is measured with Indeed, changes in the thermal and rheological behaviors occur si-
LabQuest 2 from Vernier (Beaverton, USA) coupled to a temperature multaneously in the same pH ranges. This can be explained by the
probe. One will mainly focus, in the following section, on the optimal structural modifications and the variation of the nanoparticle's ar-
pH range [4.5–6.5] identified by Bouguerra et al. [26]. The pH of the rangement in the base fluid. The pH of the solutions affects the surface
mother solution is set to 5.5 and pH is later adjusted to the desired charge of the particles, which cause their repulsion or attraction, re-
value during the primary magnetic stirring. sulting in many states of colloidal dispersion. A high surface charge
The thermal conductivity of suspensions is measured with the THW- results in well-dispersed particles, while a fully developed agglomera-
L1 Liquid Thermal Conductivity System from Thermtest Instruments. tion occurs when the surface charges vanish (point of zero charge).
This device offers an accurate and fast measure avoiding the error due Between these two limit states, different levels of charge can lead to
to natural convection. The measuring cell is integrated into a system other dispersions regimes. The WD regime is characterized by the
controlling the temperature with an accuracy of 0.1 °C (heat ex- smallest particle size, which results in a decrease of μ and is accom-
changer + thermostat bath circulator). The principle, description and panied by an increase in k. An inverse variation of these two quantities
advantage of THW-method are detailed in [26]. Before each series of is observed during the F.Agg. regime, which leads to a larger particle
measurements, calibration and validation steps have been performed size. Consequently, the combined analysis of k and μ curves can provide
using distilled water and glycerin test fluids. The repeatability and the reliable information on the dispersion state.
uncertainty of the measurements remain better than 0.5%. Given the opposite variations of k and μ, it is not clear whether a
The rheological properties of nanofluids are performed on a stress nanofluid is efficient or not. It is then necessary to find a global para-
controlled rheometer (TA HR-2) using a DIN coaxial cylinder geometry. meter able to quantify its global quality for heat exchanges. The
A Peltier cylinder regulates temperature with an accuracy of 0.1 °C. For Mouromtseff number (Mo) [28] is a factor of merit based on four
low viscosity fluids, the rheometer may have a maximum beyond which properties of the fluid: density (ρ), dynamic viscosity (μ), thermal
it is no longer able to measure the angular velocity accurately. Then, conductivity (k) and heat capacity (Cp). Higher values of Mo indicate
one first delimits the validity range of the shear rate (γ̇) relative to the higher heat transfer capabilities.

52
N. Bouguerra et al. International Communications in Heat and Mass Transfer 92 (2018) 51–55

Fig. 1. (a) Ratios of the effective thermal conductivity and dynamic viscosity as a Fig. 2. Same legend as Fig. 1 for φ = 0.5%.
function of pH for φ = 1%; (b) corresponding efficiency ratio.

liquid inside the interfacial nanolayer approaches that of the solids,

ρ0.8k 0.67Cp0.33 nanolayer can lead to an increase in k. This contribution is increasingly
Mo =
μ0.47 (1) pronounced when the particle size is decreasing [11]. Even if the role of
the Brownian motion in the increase of the thermal conductivity is
Figs. 1–4 display also the efficiency ratio of suspensions (Monf/Mobf)
defined as the Mouromtseff number of the nanofluid divided by that of
the base fluid (water). To calculate Monf, the experimental data of knf
and μnf are used, while ρnf and Cp,nf are evaluated as a function of the
volumetric fraction (φ) by applying Eqs. (2) and (3) [29]:
ρnf = φρnp + (1 − φ) ρbf (2)

ρnf Cp, nf = φρnp Cp, np + (1 − φ) ρbf Cp, bf (3)

where the subscripts nf, np and bf refer to nanofluid, nanoparticle and
base fluid, respectively. Note that using the improved model of Shar-
ifpur et al. [30] instead of Eq. (2) would lead to similar results.
The results can be classified into three categories according to the
dispersion regimes reflected by the number of k-peaks. Figs. 1 and 2
illustrate the characteristics of the nanofluids for φ = 1% and 0.5%,
respectively. These intermediate volume fractions exhibit all the dis-
persion regimes encountered here. They are characterized by the pre-
sence of two peaks of k accompanied by a peak and a valley of μ. This
constitutes the 2-peak configuration. For small (φ = 0.2%) or large
(2%) concentrations, one of the two peaks of k disappears revealing two
other different configurations with an isolated peak of k (Figs. 3 and 4).
Fig. 1 shows the existence of a first peak of k for a pH value between
5 and 5.5 while μ exhibits its lowest level. This region should corre-
spond logically to the WD regime. These monodispersed nanoparticles
promote the improvement of k since they accentuate some mechanisms
proposed to explain the atypical enhancement. Indeed, this particle's
arrangement benefits from a high specific surface. Surface/volume ratio
can increase by several orders of magnitude, which emphasizes the
importance of the nanoscale [2]. In this configuration, the liquid-solid
interfaces are more important, which leads to an increase in the number
of interfacial nanolayers. In fact, liquid in contact with a solid forms an
Fig. 3. Same legend as Fig. 1 for φ = 0.2%.
ordered nanolayer of solid-like structure. As the organization of the

53
N. Bouguerra et al. International Communications in Heat and Mass Transfer 92 (2018) 51–55

different. Indeed, the k-peak relative to WD state is more pronounced

for φ = 0.5%, while that of CL Agg. is more marked for φ = 1%. The
amplitude of these two peaks is balanced at a volumetric fraction
φ = 0.7%, which is not present here for sake of clarity.
In Fig. 3, which groups the data relating to φ = 0.2%, chain-like
agglomeration is not observed. The WD state is the only particle's ar-
rangement able to provide an increase in k at this concentration level.
The disappearance of the CL Agg. regime may be due to the lack of
particles. Indeed, the chains can be formed locally in reduced number
(slices of bridge) but not enough to create a complete path network able
to ensure the transmission of the heat flux efficiently from end to end.
In Fig. 4, which groups the data relating to φ = 2%, the WD state is
not observed. An increase of μ accompanies the enhancement of k and
reveals the configuration of the isolated k-peak resulting from the CL
Agg. regime. This high level of φ can increase the probability of inter-
particles collisions, thus complicating the appearance of the mono-
dispersed state.
The distributions of the Mouromtseff number Mo of all studied
concentrations show that, when the WD regime exists, it must be
prioritized. Compared to the CL agg. regime, this dispersion state pre-
sents better overall energy performance. The increase of φ does not lead
to any significant improvement in Mo. When the dispersion is opti-
mized, comparable efficiency can be obtained even at very low φ levels
(0.2%). However, in applications where thermal requirements prevail,
and where pumping costs are less important, nanofluids having a high
concentration may be beneficial since they provide an enhancement of
k by more than 23% for only φ = 2%. The enhancement of k is reached
thanks to the CL agg. structure which yields to a non-penalizing μ in-
crease since the nanofluid remains more efficient than the base fluid.
Fig. 4. Same legend as Fig. 1 for φ = 2%.

4. Conclusions
controversial, one can note that the WD regime accentuates this me-
chanism. Indeed, the diffusion coefficient for isolated Brownian particle The dispersion regimes in alumina/water-based nanofluids have
is inversely proportional to its diameter [13, 16, 18]. Moreover, the WD been studied experimentally by simultaneous measurements of thermal
regime allows better distribution of the particles in the host fluid, which conductivity and dynamic viscosity using advanced and appropriate
reduces the interparticle distances and limits the zones of low k in the techniques for volume fractions of nanoparticles between φ = 0.2% and
nanofluid [2]. Even if, most theoretical models dedicated to the visc- 2% at 25 °C. By varying the pH of the solutions between 4 and 8, five
osity of nanofluids connect it only to the volume concentration of na- dispersion patterns have been identified except for the two extremum
noparticles [24], the experimental studies which have treated the effect concentrations. At φ = 0.2%, the chain-like agglomeration is not re-
of the stability parameters on the rheological properties confirm that a covered, whereas at φ = 2%, it was not possible to highlight the well-
well dispersed suspension results in a decrease of μ. Conversely, ex- dispersed regime. The Mouromtseff number was then used to identify
cessive agglomeration leads to an increase in μ [31]. Large aggregates the most efficient nanofluid for practical applications [35]. If the na-
have a size that makes them losing the advantageous characteristics of noparticles are well dispersed, comparable efficiency may be achieved
nanoparticles and leads to a deterioration of k. So, it is possible to even at low φ values (φ = 0.2%).
identify the pH range between 6.5 and 8 in Fig. 1 as a F.Agg. region. A A considerable number of research papers focus on the thermal
second k-peak is recorded in the pH range between 5.5 and 6. Unlike conductivity k of nanofluids. However, the dynamic viscosity μ is an-
the first one, a peak of μ accompanies it. The values of μ in this range other important property that requires the same attention due to its
rise sharply exceeding even those of the F.Agg. zone and announcing critical effect on the overall performance of thermal energy plants.
the existence of a particular aggregation. This is in perfect agreement Consequently, the dynamic viscosity of Nfs should be systematically
with the optimized agglomeration theory suggested by Prasher et al. examined before use in heat transfer applications. Indeed, viscosity is a
[32, 33]. According to this model, the optimized aggregation size, principal flow property of fluids. Pumping power or pressure drop in
which is not the smallest, can lead to the unexpected enhancements of k convective heat transfer are directly linked to the viscosity of fluids. In
thanks to special aggregates. They are found in the literature under the other words, the dual examination of the thermal conductivity and the
name of chain-like aggregation, fractal-like aggregates and linear ag- viscosity is fundamental to determine the thermo-fluidic behavior of
gregation [20, 21, 22, 32, 33, 34]. This aggregation structure in the nanofluids.
form of chains can create highly conducting path network (bridges) that
can conduct heat more efficiently [13, 16, 18, 19]. Unlike F.Agg., where
all the advantages of the WD state are lost, this chain structuring con- Acknowledgments
tinues to benefit from a relatively high surface/volume ratio, which
ensures a large surface area for exchange with the liquid and allows the This project is part of the research program of the NSERC Chair on
nanolayer to occupy a larger portion of the volume. On the other side, Industrial Energy Efficiency, established at Université de Sherbrooke in
this organization can nullify the effect of the Brownian motion since the 2014, with the support of Hydro-Québec, Natural Resources Canada
size of structures moves away from that of the molecules of the base and Rio Tinto Alcan. The Sigma Energy Storage company and the
fluid. Canadian Foundation for Innovation (John R. Evans Leaders Fund
Fig. 2 shows the same trends as Fig. 1. The amplitudes are however number 34582) are also gratefully acknowledged.

54
N. Bouguerra et al. International Communications in Heat and Mass Transfer 92 (2018) 51–55

Conflicts of interest for heat transfer enhancement using nanofluids, Appl. Therm. Eng. 108 (2016)
720–739.
[19] E. Michaelides, Transport properties of nanofluids. A critical review, J. Non-Equilib.
None. Thermodyn. 38 (1) (2013) 1–79.
[20] H. Patel, S. Das, T. Sundararajan, A. Sreekumaran Nair, B. George, T. Pradeep,
References Thermal conductivities of naked and monolayer protected metal nanoparticle based
nanofluids: manifestation of anomalous enhancement and chemical effects, Appl.
Phys. Lett. 83 (14) (2003) 2931–2933.
[1] S. Choi, J. Eastman, Enhancing thermal conductivity of fluids with nanoparticles, [21] H. Zhu, C. Zhang, S. Liu, Y. Tang, Y. Yin, Effects of nanoparticle clustering and
Proc. of the ASME Int. Mech. Eng. Congress & Exposition, San Francisco, California, alignment on thermal conductivities of Fe3O4 aqueous nanofluids, Appl. Phys. Lett.
USA, 1995, pp. 99–105. 89 (2) (2006) 023123.
[2] S. Das, S. Choi, W. Yu, T. Pradeep, Nanofluids: Science and Technology, John Wiley [22] G. Paul, P. Das, I. Manna, Maneuvering the chain agglomerates of colloidal su-
and sons Inc., Hoboken, USA, 2008. perparamagnetic nanoparticles by tunable magnetic fields, Appl. Phys. Lett. 105
[3] J. Eastman, S. Choi, L. Thompson, S. Lee, Enhanced thermal conductivity through (18) (2014) 183108.
the development of nanofluids, Mater. Res. Soc. Symp. Proc., Boston, USA, 1996. [23] K. Bashirnezhad, S. Bazri, M. Safaei, M. Goodarzi, M. Dahari, O. Mahian, A.
[4] M. Liu, M. Lin, I. Huang, C. Wang, Enhancement of thermal conductivity with Dalkilica, S. Wongwises, Viscosity of nanofluids: a review of recent experimental
CuO for nanofluids, Chem. Eng. Technol. 29 (1) (2006) 72–77. studies, Int. Commun. Heat Mass Transf. 73 (2016) 114–123.
[5] Y. Hwang, H. Park, J. Lee, W. Jung, Thermal conductivity and lubrication [24] J. Meyer, S. Adio, M. Sharifpur, P. Nwosu, The viscosity of nanofluids: a review of
characteristics of nanofluids, Curr. Appl. Phys. 6 (2006) 67–71. the theoretical, empirical, and numerical models, Heat Transf. Eng. 37 (5) (2016)
[6] W. Yu, H. Xie, L. Chen, Y. Li, Investigation of thermal conductivity and viscosity 387–421.
of ethylene glycol based ZnO nanofluid, Thermochim. Acta 491 (1) (2009) 92–96. [25] S. Bhanushali, N. Jason, P. Ghosh, A. Ganesh, G. Simon, W. Cheng, Enhanced
[7] H. Mintsa, G. Roy, C. Nguyen, D. Doucet, New temperature dependent thermal thermal conductivity of copper nanofluids: the effect of filler geometry, ACS Appl.
conductivity data for water-based nanofluids, Int. J. Therm. Sci. 48 (2) (2009) Mater. Interfaces 9 (22) (2017) 18925–18935.
363–371. [26] N. Bouguerra, A. Khabou, S. Poncet, S. Elkoun, Thermal conductivity of Al2O3/
[8] J. Buongiorno, D. Venerus, T. Mckrell, A benchmark study on the thermal con- water-based nanofluids: revisiting the influences of pH and surfactant, Int. J. Mech.,
ductivity of nanofluids, J. Appl. Phys. 106 (9) (2009) 94312-1–94312-14. Aerosp., Ind., Mechatronic Manufact. Eng. 10 (12) (2016) 1849–1858.
[9] H. Xie, J. Wang, T. Xi, Y. Liu, F. Ai, Q. Wu, Thermal conductivity enhancement of [27] I. Mahbubul, R. Saidur, M. Amalina, M. Niza, Influence of ultrasonication
suspensions containing nanosized alumina particles, J. Appl. Phys. 91 (7) (2002) duration on rheological properties of nanofluid: an experimental study with alu-
4568–4572. mina-water nanofluid, Int. Commun. Heat Mass Transf. 76 (2016) 33–40.
[10] J. Huang, X. Wang, Q. Long, X. Wen, Y. Zhou, L. Li, Influence of pH on the stability [28] I. Mouromtseff, Water and forced-air cooling of vacuum tubes nonelectronic pro-
characteristics of nanofluids, Proc. Symposium on Photonics and Optoelectronics, blems in electronic tubes, Proc. Inst. Radio Eng. 30 (4) (1942) 190–205.
Wuhan, China, 2009, pp. 1–4. [29] Y. Xuan, W. Roetzel, Conceptions for heat transfer correlation of nanofluids, Int. J.
[11] P. Keblinski, S. Phillpot, S. Choi, J. Eastman, Mechanisms of heat flow in sus- Heat Mass Transf. 43 (19) (2000) 3701–3707.
pensions of nano-sized particles (nanofluids), Int. J. Heat Mass Transf. 45 (4) (2002) [30] M. Sharifpur, S. Yousefi, J. Meyer, A new model for density of nanofluids in-
855–863. cluding nanolayer, Int. Commun. Heat Mass Transf. 78 (2016) 168–174.
[12] P. Keblinski, J. Eastman, D. Cahill, Nanofluids for thermal transport, Mater. Today [31] P. Keblinski, R. Prasher, J. Eapen, Thermal conductance of nanofluids: is the
8 (6) (2005) 36–44. controversy over? J. Nano Res. 10 (7) (2008) 1089–1097.
[13] M. Chandrasekar, S. Suresh, A review on the mechanisms of heat transport in [32] R. Prasher, W. Evans, P. Meakin, J. Fish, P. Phelan, P. Keblinski, Effect of
nanofluids, Heat Transf. Eng. 30 (14) (2009) 1136–1150. aggregation on thermal conduction in colloidal nanofluids, Appl. Phys. Lett. 89 (14)
[14] V. Terekhov, S. Kalinina, V. Lemanov, The mechanism of heat transfer in nano- (2006) 88–90.
fluids: state of the art (review). Part 1. Synthesis and properties of nanofluids, [33] R. Prasher, P. Phelan, P. Bhattacharya, Effect of aggregation kinetics on the
Thermophys. Aeromech. 17 (1) (2010) 1–14. thermal conductivity of nanoscale colloidal solutions (nanofluid), Nano Lett. 6 (7)
[15] H. Machrafi, G. Lebon, The role of several heat transfer mechanisms on the en- (2006) 1529–1534.
hancement of thermal conductivity in nanofluids, Contin. Mech. Thermodyn. 28 (5) [34] J. Sui, P. Zhao, B. Bin-Mohsin, L. Zheng, X. Zhang, Z. Cheng, Y. Chen, G. Chen,
(2016) 1461–1475. Fractal aggregation kinetics contributions to thermal conductivity of nano-suspen-
[16] J. Wang, R. Zheng, J. Gao, G. Chen, Heat conduction mechanisms in nanofluids sions in unsteady thermal convection, Sci. Rep. 6 (1) (2016) 39446.
and suspensions, Nano Today 7 (2) (2012) 124–136. [35] N. Che Sidik, I. Adamu, M. Jamil, G. Kefayati, R. Mamat, G. Najafi, Recent
[17] M. Patil, J. Seo, S. Kang, M. Lee, Review on synthesis, thermo-physical property, progress on hybrid nanofluids in heat transfer applications: a comprehensive re-
and heat transfer mechanism of nanofluids, Energies 9 (10) (2016) 1–17. view, Int. Commun. Heat Mass Transf. 78 (2016) 68–79.
[18] R. Vidonscky Pinto, F. Augusto, S. Fiorelli, Review of the mechanisms responsible

55