Thermal Performance of Selected Nanofluids with Surfactants

Nanofluids have been proposed for use as working fluids in order to increase energy efficiency. While a large number of studies have been performed, there is comparatively high variation in reported physical property and heat transfer enhancement data. In addition, comparatively few thermal enhancement studies consider the effects of surfactants that may be required to ensure sufficient stability of the nanofluid over time. In this study, nanofluids were prepared by combining different nanoparticles, base fluids and surfactants and were subsequently evaluated for stability using the sedimentation method. Based on the sedimentation trials and viscosity measurements, three nanofluids (aluminium oxide/water; activated carbon/CTAB/water; copper oxide/ARB/water) were selected for thermal performance enhancement experiments. Thermal enhancement performance was tested in a closed loop with two double pipe heat exchangers and all physical properties required in the calculations were measured directly as part of this study. The carbon/CTAB/water nanofluid had the highest heat transfer enhancement index (a comparison of increased heat transfer rate against increased pressure drop) of the three nanofluids considered. Sedimentation of nanoparticle agglomerates was most noticeable with the aluminium oxide/water nanofluid that did not contain a surfactant.


Introduction
In 1993 Choi proposed the use of a suspension of nano-meter sized particles in heat transfer fluids and is often credited with introducing the term 'nanofluid' [1].Choi and Eastman [2] reported that the thermal conductivity of copper nanoparticles suspended in water was enhanced by a factor of 1.5 and 3.5 compared to water for volume fractions of 5% and 20% respectively; substantially greater than those measured for milli-or micro-metre sized particles.Further experiments with copper nanoparticles in acidified ethylene glycol also showed significant increases in thermal conductivity [3].However, they also observed that the thermal conductivity of nanofluids decreased with time, which was most likely due to agglomeration and/or sedimentation of the nanoparticles.
Similarly encouraging results were observed by other researchers and subsequently the use of nanofluids for heat transfer enhancement has become a very active area of research [4][5][6].However, despite many potential applications being identified for nanofluids, to date it does not appear that nanofluids have received widespread usage outside the research environment, and there appears to be little in the way of standardisation of nanofluid technology.One obvious barrier is the relatively high cost of nanoparticles, being typically in the region of hundreds to thousands of US dollars per kg [6].However, it is also probable that uncertainty surrounding the true enhancement value of the nanoparticles may be another significant barrier, since reported thermal properties and heat transfer enhancement values vary significantly for the same nanoparticle [5].To illustrate this point, Fig. 1 shows reported Nusselt number (Nu) values for alumina/water nanofluids along with the Dittus-Boelter equation [7] plotted for water over a range of Reynolds numbers (Re).While all the studies reported significantly higher Nu than for water, there is high variation in Nu for similar volume fraction concentrations of the alumina/water nanofluid.This is also true of thermal property data; the reported thermal conductivities of nanofluids in particular vary widely.It is difficult to determine whether this is due to real thermal conductivity differences that may be caused by particle size and morphology and interaction with base fluid [8] or simply due to measurement errors/ uncertainties.
Another challenge that needs to be overcome before there can be widespread uptake of nanofluids is how to manage or minimise the agglomeration of nanoparticles in suspension that leads to sedimentation over time [9][10][11][12][13][14].Simple methods have been proposed such as adding stabilising agents (surfactants) to the base fluid before the suspension of nanoparticles to lower the interfacial forces between the fluid molecules and the nanoparticles thereby improving their stability [15,16].This has been shown to be effective in a number of practical applications, e.g., for enhancing heat transfer rates in heat pipes [17,18] and in boiling and condensation processes [19].However, even with the addition of surfactants there is no guarantee of permanent stability [20].
A feature of many of the thermal enhancement studies [21,22] is that they make use of literature values for thermal properties without necessarily making it clear how the particular values were chosen from amongst the wide ranges reported in the literature.In addition, few take into consideration the problem of sedimentation and most make no mention of stabilising agents such as surfactants.It appears that, in general, characterisation and physical properties measurement of nanoparticles and nanofluids are performed by materials scientists and chemists, whereas thermal and mechanical engineering researchers carry out the experiments on the application of nanofluids, and there is not always close collaboration or communication between the two groups [5].Therefore, the aim of this study is to assess the heat transfer enhancement performance of selected nanofluids with particular emphasis on the effect of surfactants, while using only directly measured physical property data, in order that uncertainty surrounding physical property data from the literature may be avoided.

Nanoparticle/Surfactant Pairing and Physical Properties Measurement
Nanofluids considered for this study were combinations of the following base fluids, nanoparticles, and surfactants: • Base fluids-water, ethylene glycol, n-hexane • Nanoparticles-alumina (Al 2 O 3 ), copper oxide (CuO), activated carbon (C) • Surfactants-sodium lauryl sulphate (SDS), cetyltrimethyl ammonium bromide (CTAB), sodium dodecyl benzene sulfonate (SDBS), Arabinogalactan (ARB) The nanoparticles listed were chosen because they were available commercially and each has been reported to enhance thermal conductivity beyond what is predicted by classic (Maxwell) theory [5].The surfactants selected are in common use and were readily available to the authors.Hexane was chosen as a nonpolar base fluid in case difficulties were encountered with achieving suspension of nanoparticles in polar base fluids; however, since no difficulties were encountered with any of the nanoparticles, hexane was not persisted with beyond preliminary experimentation.
Prior to the heat transfer enhancement trials, a study of the effectiveness of the different surfactants for achieving nanofluid stability was performed using the sedimentation method [23][24][25].Stability was assessed for the three nanoparticles with each of the four surfactants in both water and ethylene glycol, as has been reported previously [26].In each case masses of surfactants equal to the mass of nanoparticles were used for nanoparticle concentrations in the range of 0.75 to 1.5 volume %.On average the samples containing ARB remained in suspension the longest for both water and ethylene glycol, with carbon/ARB remaining in suspension in water for 30 days (compared to less than 1 day, for some nanofluids without a surfactant).
After the sedimentation trials were completed the viscosities of the nanofluids were measured, as has been reported previously [26].It was observed that although it provided the greatest stability ARB caused significant increases in viscosity (e.g., the carbon/ARB/water nanofluid had viscosity 3 times greater than that of carbon/ water), which increases the pumping power requirement.Taking into consideration the increase in stability against the increase in viscosity, three of the nanofluids were selected for consideration in the heat transfer enhancement trials: carbon/CTAB/ water, CuO/ARB/water and Al 2 O 3 /water.
The true densities of the activated carbon, alumina and copper oxide nanoparticles were measured using the Archimedes principle and the heat capacities of the nanoparticles were measured using Differential Scanning Calorimetry (DSC), and the results are summarised in Table 1.As the surfactants dissolved in water, it was not possible to measure the true density via Archimedes principle, so instead the mixture densities of water + CTAB and water + ARB for a range of concentrations of the surfactant was measured.The densities of the nanofluids were calculated from the density of the water/surfactant mixture and the density of the nanoparticle via Eq. 1 [27]: where ρ is density and v is volume fraction.The specific heat capacities of the nanofluids were calculated from the specific heat capacities of the components using Eq. 2 [28]: where c p is specific heat capacity and x is mass fraction.Mass fractions for the ith component of the nanofluid (x i ) are related to volume fractions (v i ) by the following relationship [29]: The thermal conductivities of the nanofluids were measured with a Hukseflux™ TP08 thermal conductivity probe; however, it was difficult to achieve repeatable results, despite the fact that repeatable results for water were consistently obtained that were within the experimental uncertainty of the literature value [30].A second thermal conductivity measurement device, a Thermtest TC-30™, was also used; however, similar problems were encountered, and as a result no thermal conductivity data are reported in this paper.The lack of reliable measured thermal conductivity . data did not prevent the assessment of thermal performance of the nanofluids; however, it meant that results are presented in terms of heat transfer coefficients, rather than the more common practice of presenting Nusselt numbers.

Thermal Performance Assessment Apparatus and Procedure
After identifying Al 2 O 3 /water, CuO/ARB/water, and C/CTAB/water as promising nanofluids based on their stability and comparatively low viscosity, a test apparatus was constructed to perform the heat transfer performance assessment.
A number of different apparatus have been used to measure heat transfer rates in nanofluids [31][32][33]; however, the most common seems to be the double-pipe heat exchanger [21,22,[34][35][36].The attraction of this device is that it is relatively simple and cost-effective to fabricate, and so a double-pipe heat exchanger test rig was built for this study.
The experimental test rig (Fig. 2) contained a closed test loop consisting of two identical double-pipe heat exchangers (DPHEX) that contained the working fluid (tube-side) and were connected to a centrifugal pump and an electromagnetic flow meter.Prior to running experimental trials, the system was charged with water and monitored for 3 days to make sure that there was no leakage.The entire test loop was insulated using rubber foam (R value 0.7 m 2 •K•W −1 of 20 mm thickness) to minimise heat loss.A heat exchanger with a large surface area would be preferable for experimental purposes to maximise heat transfer rates.However, the larger the surface the larger the volume of nanofluids required to achieve a given concentration, which in turn means higher cost.It was decided that the size of the heat exchangers should be similar to those used in previous studies (e.g., Han et al. [21] and Sonawane et al. [22]), with a 1 m length and approximately 1-in.diameter.The dimensions of the heat exchangers that were fabricated from stainlesssteel grade 316 for this work are shown in Table 2.The heat exchangers were both installed in the counter-flow configuration.
Cold water was pumped through the shell-side of DPHEX 1 (Fig. 2) by a pump and a magnetic flow meter was used to measure the flow rates in the shell-side.The hot water supply came from a tank that was heated by a steam coil.T-type thermocouples were used to record temperatures at locations indicated in Fig. 2.An icepoint test was used to provide absolute calibration, while a water bath test was used for relative calibration between thermocouples over a range of temperatures [37].Flowrates were measured using Endress Hauser Promag 50P electromagnetic flowmeters and the pressure drop across the test loop was measured using a differential pressure sensor (Yokogawa, model EJA110A) and all measured data was processed using a DAQ 34970A data acquisition system (Agilent technologies Inc.).
An experimental run was started by charging the test loop of the rig with the working fluid being tested.The bubble vent (Fig. 2) was used to determine when the test loop was bubble free.Once charged, the pump in the test loop was turned on to circulate the working fluid.The shell-side flow rates were fixed at 55.0 L/min because it was the maximum steady flow rate that could be achieved whereas flow rates on the tube-side were varied between 1.9 L/min and 71.0 L/min.After a tubeside flowrate was selected, the system was allowed to reach steady state after which time the flowrates, temperatures and pressure drop were recorded for 30 min before changing the tube-side flow rate.A typical run of the experiment with measurements at 6 to 7 different tube-side flowrates took 6 h to 7 h.
At the end of each run the test loop pump was flushed with water to wash all nanoparticles out until clear wash-water was observed.Three to four experimental runs for each fluid were carried out to provide replicate datapoints for a given flowrate.The average heat transfer rate Qavg was obtained by calculating the enthalpy changes of the cooling water and nanofluid through the heat exchanger [38]: where ṁ is the mass flow rate, T is temperature with the subscript number referring to the temperature measurement location in Fig. 2. The average overall heat transfer coefficient U was evaluated as [38]: where A is the heat transfer surface area and ΔT LM is the log mean temperature difference [38]: where ΔT A is the difference between fluid (tube-side) inlet and cold tap water (annulus) outlet temperatures and ΔT B is the difference between fluid (tube-side) outlet and cold tap water (annulus) inlet temperatures.Temperature-dependent thermal properties of water used in the calculation of heat transfer rates were calculated from the International Association for the Properties of Water and Steam (IAPWS) functions [39,40].
Friction factor 'f' for the test closed loop was obtained from the Fanning friction factor [41] equation as: where ΔP is the measured pressure drop and l is the length of the heat exchanger tube, and u is the average fluid velocity.
In this study a root-sum-square approach proposed by Kline and McClintock [42] was used to estimate measurement uncertainty.If R is any defined function of independent variables (a, b, c…), the uncertainty δR of the function was obtained from: The uncertainty estimates for each measured variable are shown in Table 3.

Thermal Performance Results and Practical Aspects Effects of Nanofluids
Prior to the experiments with nanofluids, a trial with deionised water was performed to provide a basis for comparison against the nanofluids.Subsequently trials for each of the selected nanofluids were performed as described in the previous section and the results are shown in Fig. 3. ( All three nanofluids produced higher overall heat transfer coefficients (U) than water at any Re, with increases in the range of 10% to 50%, depending on Re.At higher Re (e.g., above 50 000) the C/CTAB/water nanofluid and the alumina/water nanofluid both performed significantly better than the CuO/ARB/water nanofluid.
Figure 4 shows the pressure drops for the three selected nanofluids and water over a range of Re.The nanofluids experienced significantly higher pressure drops than water, as expected due to the increase in viscosity caused by the presence of the nanofluid and surfactant.The CuO/ARB/water nanofluid produced the highest pressure drop, perhaps caused more by the ARB surfactant than the CuO nanoparticles themselves, since ARB was shown to cause significant increases in viscosity [26].It is noteworthy that while the pressure drop results for water followed the expected dependence on Re (ΔP ∝ Re 2 ), this was not observed for any of the nanofluids, for  which the dependence of pressure drop on Re might be fitted with a power law relationship with a power less than unity.The reason for this was not discovered as part of this study but would make for an interesting investigation.The results displayed in Figs. 3 and 4 clearly show that the nanofluids produce significant heat transfer enhancement at the expense of significantly increased pressure drop.A variety of different methods have been proposed for evaluating heat transfer enhancement when incorporating the effect of increased pressure drop.In the case of single-phase flow with fixed heat transfer area and constant cross-sectional area of flow, Webb and Kim [43] used the Chilton-Colburn factor (j) to represent the heat transfer rate [38] where h is the convective heat transfer coefficient, ρ, c p , μ, and v are the density, specific heat capacity, viscosity and the velocity of the fluid respectively, St is the Stanton Number and Pr is the Prandtl Number.Because the tube inside diameter in this study was constant, the heat transfer coefficient could be written as [38]: where G is the tube-side mass flow velocity (i.e., the product of density and velocity, ρv).Comparing the product of the heat transfer coefficient and surface area (hA) of the enhanced case, relative to that of reference case (subscript ref): The pumping power Π may be calculated as [43]: ( Fig. 4 Measured pressure drops for the selected nanofluids and water over a range of Re Writing Eq. 12 as a ratio relative to the smooth surface, yields: Elimination of mass velocities (G/G ref ) from Eqs. 11 and 13, gives: As the heat transfer area was held constant in this study the ratio A/A s was 1.0 and hence all area terms could be removed from Eq. 14.
Webb and Kim's heat transfer enhancement index η was developed from Eq. 14 and is defined as: A number of other researchers [24,[44][45][46][47][48][49] used a modified form of Webb and Kim's enhancement index based on Nusselt numbers (Nu) rather than the Chilton-Colburn j-factor or the heat transfer factor.However, as the Nu were not determined in this study due to uncertainty surrounding the measurement of the thermal conductivity of the nanofluids, the definition of the heat transfer enhancement index in this study was based on overall heat transfer coefficients with water as the reference case, i.e.: Figure 5 shows the enhancement index defined by Eq. 16 for each of the selected nanofluids over a range of Re.The C/CTAB/water nanofluid had the highest heat transfer enhancement index of the three nanofluids considered for any Re value.Both C/CTAB/water and Al 2 O 3 /water had heat transfer enhancement indices greater than 100% at higher Re, whereas the CuO/ARB/water nanofluid only had a heat transfer enhancement index of 100% at the highest Re considered.For a given mass fraction of nanoparticles, the volume fraction of carbon is higher than of Al 2 O 3 and CuO due to its significantly lower density (Table 1), and hence the thermal conductivity enhancement effect will be greater.In addition, CTAB had a significantly smaller impact on viscosity than ARB, and these two factors are likely to have been major reasons why C/CTAB/water performed the best.
In absolute terms the heat transfer enhancement indices are not that promising since the nanofluids are only more effective than water for comparatively high Re values (above 70 000 to 80 000 for CuO and Al 2 O 3 respectively), higher than the majority of practical applications.However, while Eq. ( 16) is derived from first principles it does not necessarily reflect the true cost-benefit trade-off of the increased heat transfer achievable.As it is defined, the pumping cost is effectively discounted 160 Page 12 of 15 by a power of 1/3 relative to the heat transfer enhancement.If both the ratio of overall heat transfer coefficients and the ratio of friction factors were weighted according to true costs, the performance indices of the nanofluids might be more favourable (or possibly less favourable).However, it is unlikely that the order of the three nanofluids in terms of effectiveness (C/CTAB > Al 2 O 3 > CuO/ARB) would change.
It is also worth mentioning that if the concentrations of the nanoparticles were increased, the performance enhancement indices defined by Eq. ( 16) would increase, but at the expense of greater cost of the nanoparticles.

Practicality of the Use of Nanofluids for Heat Transfer Purposes
During this study, several practical challenges were encountered while testing nanofluids in the closed flow loop.Each time a new nanofluid was to be tested, it required a lot of water to hose and wash out the previously tested fluid from the closed test loop, which also revealed that large, aggregated particles formed for all tested nanofluids, but most significantly for the Al 2 O 3 /H 2 O nanofluid which had no surfactant during testing.The cleaning process was time consuming too because it typically took a full day to wash nanofluid out of all parts in the closed loop.
In addition, the pump that was used to circulate nanofluids in the closed loop would not run smoothly at the beginning of each test run and had to be restarted several times, especially for the Al 2 O 3 /water nanofluid that had no surfactant.It is suspected that nanoparticles may have clogged the pump, as has been encountered by other researchers [35].It was also observed that nanoparticles tended to block pressure taps causing the differential pressure cell to display an error signal.
It was difficult to reuse a nanofluid because of the nanoparticle aggregates that formed and the large quantity of flush-water that was required to purge the system of a particular nanofluids.These practical problems were significantly greater for the So, while it has been shown again that nanoparticles can significantly increase rates of heat transfer, when taking the increased cost of pumping and the problems caused by agglomeration and sedimentation of the nanoparticles into consideration it is perhaps not surprising that nanofluids are yet to become widespread in industry.

Conclusions
Three nanofluids (aluminium oxide/water; activated carbon/CTAB/water; copper oxide/ARB/water) were selected for thermal performance enhancement experiments after a preliminary study on the effect of surfactants on stability and viscosity.
• All three nanofluids were able to achieve significantly higher overall heat transfer coefficients than water; however, all three also had significantly higher pressure drops with the highest increase being recorded for CuO/ARB/water.• The carbon/CTAB/water nanofluid had the highest heat transfer enhancement index.• Sedimentation of nanoparticle agglomerates was observed for all nanofluids but was most noticeable with the aluminium oxide/water nanofluid that did not contain a surfactant.

Fig. 1
Fig. 1 Nusselt numbers for alumina/water nanofluids over a range of Reynolds numbers, with the predictions of the Dittus-Boelter equation to represent the curves for water

Fig. 2
Fig. 2 Schematic diagram of the closed flow loop

Fig. 3
Fig.3Comparison of overall heat transfer coefficients obtained with water and selected nanofluids tubeside of the double pipe heat exchanger, with water held at constant velocity (Re = 9900) on the shell-side hA h ref A ref = jAG j ref A ref G ref .

Fig. 5
Fig. 5 Heat transfer enhancement index for selected nanofluids over a range of Re

Table 3
Summary of estimated uncertainties