Heat Transfer Enhancement Using Al₂O₃, CuO, and Hybrid Al₂O₃–CuO Nanofluids in Water and Ethylene Glycol–Water Mixtures in an Automotive Radiator

The thermophysical properties of nanofluids were estimated using well-established theoretical models that have been validated and widely applied in previous research [23, 24]. The investigated nanofluid systems include:

• Nanoparticles: Al₂O₃, CuO, and a hybrid mixture (50% Al₂O₃ + 50% CuO);
• Base fluids: pure water and ethylene glycol/water mixtures with volume ratios of 20:80 and 40:60.

The thermophysical properties of the base fluids were obtained from established data sources [25, 26]. It is assumed that the nanoparticles are uniformly and stably dispersed within the base fluid, enabling the use of effective medium theory to estimate the bulk properties of the nanofluid. The following models were employed for the calculation:

Density:

$\rho_{n f}=\varphi \rho_p+(1-\varphi) \rho_{b f}$               (1)

Specific heat capacity:

$\left(\rho C_p\right)_{n f}=\varphi\left(\rho C_p\right)_p+(1-\varphi)\left(\rho C_p\right)_{b f}$                  (2)

Thermal conductivity (Maxwell model for sperical paricles):

$k_{n f}=\frac{k_p+(\Phi-1) k_{b f}-\varphi(\Phi-1)\left(k_{b f}-k_p\right)}{k_p+(\Phi-1) k_{b f}+\varphi\left(k_{b f}-k_p\right)} k_{b f}$             (3)

where, the shape factor is assumed to be Φ=3 for spherical particles [27, 28].

Viscosity (for pure water as the base fluid):

$\mu_{n f}=\mu_{b f}\left(123 \varphi^2+7.3 \varphi+1\right)$             (4)

where, the indices “nf”, “bf”, and “p” denote nanofluid, base fluid, and particles, respectively.

In the case of ethylene glycol/water mixtures as the base fluid, the same equations remain valid for nanoparticle volume fractions up to 3%, as supported by Sharma et al. [29]. The Maxwell model was selected for its proven accuracy with spherical particles at low concentrations (up to 7%), as validated in prior studies [23, 24]. Factors such as Brownian motion and particle agglomeration were not incorporated, as the single-phase homogeneous approximation assumes stable dispersion and negligible dynamic effects under laminar flow conditions, consistent with the low nanoparticle fractions and validated models used here [30]. Future work could extend to multiphase models for higher concentrations or unstable dispersions.

In addition to single-component nanofluids, this study also investigates hybrid nanofluids (Al₂O₃–CuO at a 50:50 ratio), which have demonstrated enhanced thermal performance. According to Sudarmadji et al. [31], such hybrid systems can achieve up to 83% improvement in heat transfer due to synergistic effects, including particle resonance and intensified Brownian motion.

The effective properties of hybrid nanofluids were computed using a weighted average approach:

Density of hybrid nanofluid:

$\rho_{n f}=\emptyset_1 \rho_{p 1}+\emptyset_2 \rho_{p 2}+\left(1-\emptyset_1-\emptyset_2\right) \rho_{b f}$               (5)

with $\emptyset_1=\emptyset_2=\emptyset / 2$.

Specific heat capacity of hybrid nanofluid:

$\begin{gathered}\left(\rho C_p\right)_{n f}=\emptyset_1\left(\rho C_p\right)_{p 1}+\emptyset_2\left(\rho C_p\right)_{p 2}+\left(1-\emptyset_1\right.  \left.-\emptyset_2\right)\left(\rho C_p\right)_{b f}\end{gathered}$            （6）

where, p1, p2 refer to Al₂O₃ and CuO.

These models have been reported to agree well with experimental data in numerous studies [25, 32] and were adopted as input parameters for the CFD simulations conducted in ANSYS Fluent. The full list of thermophysical properties used in the simulations is summarized in Table 1.

Table 1. Thermal properties of base fluids, nanoparticles and nanofluids

4.1 Calculation of average heat transfer coefficient and average Nusselt number

The average convective heat transfer coefficient havg and average Nusselt number Nuavg were determined using Newton's law of cooling and thermo-fluid parameters obtained from CFD simulations. The total heat transfer from the coolant to the inner tube wall is expressed as:

$Q=h_{a v g} A_s\left(T_b-T_s\right)$                (10)

where, A_s is the inner surface area of the tube, T_s denotes the average tube wall temperature, and T_b is the average coolant temperature.

The average heat transfer coefficient was calculated directly from the heat flux density using:

$h_{a v g}=\frac{q^{\prime \prime}}{T_b-T_s}$                      (11)

Here, $q^{\prime \prime}$ represents the mean heat flux density across the tube wall (W/m²), obtained from the Total Surface Heat Flux field using the Area-Weighted Average function (Surface Integrals tool) in Ansys Fluent. The tube wall temperature Ts was similarly determined by area-weighted averaging of the Wall Temperature field data. The mean fluid temperature Tb was computed as a mass-weighted average across the entire fluid domain using the Volume Integrals > Mass-Weighted Average function.

The average Nusselt number was subsequently calculated as:

$N u_{\text {avg }}=\frac{h_{\text {avg }} D_h}{k}$            (12)

where, D_h is the hydraulic diameter and k is the thermal conductivity of the fluid.

4.2 Simulation model verification

To validate the accuracy of the CFD model, the present study replicated simulations using pure water as the working fluid, adopting the geometric configuration and boundary conditions reported by Hussein et al. [34]. The thermophysical properties of pure water were also taken from the same source. Unlike the original study, which simplified the model by neglecting the tube wall thickness and excluding the fins, the current model incorporates these features to more accurately represent the actual structure of an automotive radiator. Nusselt number data from the reference study were extracted using the WebPlotDigitizer software, as the original publication did not provide tabulated data. The simulation results were then compared with the reference data, as shown in Figure 3. The results demonstrate that the present model successfully reproduces the increasing trend of the Nusselt number with Reynolds number. The average deviation between the two datasets of Nusselt number is approximately 10.3%, which is considered acceptable in light of the differences in physical modeling and the potential uncertainties introduced during data extraction from graphs.

4.3 Effect of ethylene glycol on the convective heat transfer coefficient

Figure 4 represents the variation of the convective heat transfer coefficient (h) with inlet velocity for three working fluids: pure water, a 20:80 ethylene glycol/water (EG/W) mixture, and a 40:60 EG/W mixture. The results indicate that h decreases as the ethylene glycol content in the mixture increases. This observation agrees with the experimental work of Chen et al. [35], which demonstrated that ethylene glycol reduces heat transfer efficiency due to its lower thermal conductivity and higher viscosity. Pure water consistently exhibits the highest h value among the three fluids at any given inlet velocity. The 20:80 EG/W mixture shows intermediate performance, whereas the 40:60 EG/W mixture exhibits the lowest heat transfer coefficient across the investigated range of velocities. This behavior can be attributed to the combined effects of increasing viscosity and decreasing thermal conductivity as the ethylene glycol concentration rises, which hinder effective heat transfer between the liquid flow and the pipe wall. Moreover, h increases with inlet velocity in all cases, following the theory of forced convection heat transfer. Higher velocities enhance flow turbulence, thereby improving the convective heat transfer coefficient. However, the rate of increase in h is noticeably lower for mixtures with higher EG content, reflecting the detrimental effect of ethylene glycol on overall heat transfer performance.

333.png

Figure 3. Nusselt number versus Reynolds number for the current CFD model compared with Hussein et al. [34]

444.png

Figure 4. Effect of ethylene glycol concentration on the convective heat transfer coefficient

4.4 Effect of Al₂O₃ and CuO nanoparticle concentrations in pure water on the convective heat transfer coefficient

Figures 5 and 6 depict the variation of the convective heat transfer coefficient (h) with Reynolds number for different concentrations of Al₂O₃ and CuO nanoparticles dispersed in pure water. The investigated nanoparticle volume concentrations include φ = 0 (pure water), 0.01 (1%), 0.03 (3%), 0.05 (5%), and 0.07 (7%). The results indicate that the heat transfer coefficient increases markedly with rising nanoparticle concentration across the entire range of Reynolds numbers considered. At a given Reynolds number, nanofluids containing Al₂O₃ or CuO exhibit significantly higher h values compared to pure water. The enhancement effect becomes more pronounced at higher concentrations, particularly at φ=0.07, where the heat transfer coefficient reaches its maximum value within the investigated range. The maximum enhancements are approximately 32.18% for Al₂O₃ and 28.51% for CuO relative to pure water (see Tables 2 and 3), demonstrating the significant impact of nanoparticle addition on the thermal performance of the base fluid.

555.png

Figure 5. Effect of Al₂O₃ nanoparticle concentration on h in pure water

4.5 Comparison of convective heat transfer coefficients of nanofluids containing Al₂O₃ and CuO nanoparticles in pure water

Figure 7 illustrates the variation of the convective heat transfer coefficient with Reynolds number for nanofluid solutions comprising Al₂O₃ and CuO nanoparticles at different volume concentrations (1%, 3%, 5%, and 7%) in pure water. Overall, both types of nanofluids exhibited a marked enhancement in heat transfer performance compared to pure water. This enhancement increased progressively with Reynolds number and nanoparticle concentration. A comparison between the two nanoparticle types revealed that the Al₂O₃-based nanofluid achieved higher heat transfer coefficients than the CuO-based nanofluid under identical operating conditions. However, within the simulation parameters of this study, the maximum observed difference between the two nanofluids was approximately 3% (see Table 3), suggesting that the disparity in heat transfer performance remained relatively small within the investigated range.

666.png

Figure 6. Effect of CuO nanoparticle concentration on h in pure water

777.png

Figure 7. Comparison of convective heat transfer coefficients of nanofluids containing Al₂O₃ and CuO nanoparticles in pure water

Table 2. Convective heat transfer coefficient (h) of pure water at various Reynolds numbers

Table 3. Heat transfer coefficients of pure water-based nanofluids at different concentrations     

Table 4. Heat transfer coefficients of 40:60 (EG/W)-based nanofluids at different concentrations

Table 5. Effect of Al₂O₃ nanoparticles on heat transfer in water and EG/W solution at various Reynolds numbers

4.6 Comparison of convective heat transfer coefficients of nanofluids containing Al₂O₃ and CuO nanoparticles in a 40%EG/60%Water base fluid

Figure 8 illustrates the variation of the convective heat transfer coefficient (h) with Reynolds number for nanofluids containing Al₂O₃ and CuO nanoparticles at volume concentrations of 1% and 3%, using a base fluid of 40% ethylene glycol and 60% water (40%EG/60%Water). Similar to the results obtained with pure water, both nanofluids exhibited significant enhancements in the convective heat transfer coefficient compared to the base fluid without nanoparticles. This enhancement increased with Reynolds number and nanoparticle concentration.

888.png

Figure 8. Comparison of convective heat transfer coefficients of nanofluids containing Al₂O₃ and CuO nanoparticles in a 40%EG/60%Water base fluid

However, consistent with previous observations, the CuO nanofluid demonstrated a lower improvement in heat transfer performance compared to the Al₂O₃ nanofluid under identical conditions. Notably, when comparing the heat transfer enhancement efficiency of a nanofluid containing 3% Al₂O₃ in two different base fluids at Re = 1750, it was found that the enhancement in the 40%EG/60%Water mixture was 12.9%, which was higher than that in pure water (11.65%) (see Tables 4 and 5). This finding indicates that selecting a base fluid with an appropriate ethylene glycol ratio, combined with an optimal nanoparticle concentration, can yield superior heat transfer efficiency compared to using pure water as the dispersion medium.

Table 5 presents the convective heat transfer performance of nanofluids containing Al₂O₃ nanoparticles dispersed in two different base fluids: pure water and a 40:60 ethylene glycol/water (EG/W) mixture. The results were evaluated over a Reynolds number range of 250 to 1750, with nanoparticle volume concentrations of 1% and 3%. The addition of Al₂O₃ nanoparticles significantly increased the convective heat transfer coefficient in both base fluids. At a concentration of 1%, the heat transfer efficiency improved from approximately 3.08% to 3.40% in pure water, and from 3.48% to 3.71% in the EG/W mixture. When the nanoparticle concentration increased to 3%, the efficiency gain became more pronounced, reaching 11.65% for pure water and 12.90% for EG/W at Re = 1750. These results highlight the positive and consistent effect of Al₂O₃ nanoparticles in enhancing heat transfer performance across a range of flow rates.

Although pure water exhibited a higher absolute heat transfer coefficient due to its superior thermal conductivity, the relative improvement rate (%) resulting from the addition of nanoparticles was consistently higher in the EG/W mixture. This can be attributed to the inherently lower heat transfer capacity of EG/W, where the incorporation of nanoparticles produced a relatively more pronounced effect. For instance, at Re = 1500 with a nanoparticle concentration of 3%, the enhancement reached 12.81% in EG/W compared to 11.56% in pure water.

From an application standpoint, these findings suggest that Al₂O₃ nanofluids in an EG/W medium can provide relatively greater benefits in terms of heat transfer efficiency, making them particularly suitable for systems that utilize base fluids with low inherent thermal conductivity. However, it is important to note that increasing the nanoparticle concentration may lead to higher viscosity and pressure losses, which could negatively impact the overall thermo-hydraulic performance of the system. Therefore, future investigations should consider additional factors such as pressure drop and pumping power to obtain a more comprehensive assessment of operational efficiency.

Moreover, in contrast to the findings reported by Sudarmadji et al. [31], where CuO nanofluids demonstrated superior heat transfer performance over Al₂O₃, the present study reveals that Al₂O₃ outperforms CuO under steady flow conditions and at higher concentrations (3%). This disparity underscores that the heat transfer performance of different nanoparticle types is strongly influenced by specific operating conditions, channel geometry, and base fluid properties.

101.png

102.png

103.png

104.png

105.png

106.png

Figure 9. Temperature distribution along the cooling tube for two types of nanofluids with different base fluids at Re = 1750

Figure 9 illustrates the temperature distribution obtained from simulations of fluid flow in the cooling tube at a Reynolds number of Re = 1750. The analyzed cases involve nanofluids containing CuO and Al₂O₃ nanoparticles at volume concentrations of 3% and 7%, dispersed in two base fluids: pure water and a 40:60 ethylene glycol/water mixture (EG/W).

The temperature contours depict the distribution of temperature along the tube length as the coolant circulates through the system. The results reveal a gradual decrease in temperature along the nanofluid flow direction, indicating heat transfer from the fluid to the tube wall and subsequently to the surrounding environment. These findings are consistent with the work of Tijani and bin Sudirman [32], where heat is transferred from the high-temperature coolant through the flat tube and fins via a combination of convective heat transfer within the fluid and conduction through the solid material. The continuous and uniform temperature profiles, without the presence of localized hot spots, suggest that the flow remains stable and that the heat transfer problem has been effectively resolved within the computational domain. Furthermore, the contour plots visually highlight the extent and intensity of heat exchange on the tube surface, thereby identifying the locations of effective cooling zones along the system’s length.

4.7 Comparison of heat transfer efficiency of three types of nanoparticles: CuO, Al₂O₃, and Al₂O₃ + CuO in two base fluids

Figures 10 and 11 illustrate the variation of the convective heat transfer coefficient for three types of nanoparticles, CuO, Al₂O₃, and a hybrid mixture of Al₂O₃ + CuO, dispersed in two base fluids: pure water and a 40:60 ethylene glycol/water (EG/W) solution. In all cases, the nanoparticle concentration was maintained at 3% by volume, while the Reynolds number ranged from 250 to 1750 to simulate different flow regimes.

1010.png

Figure 10. Comparison of heat transfer efficiency of three types of nanoparticles in pure water

1111.png

Figure 11. Comparison of heat transfer efficiency of three types of nanoparticles in the base fluid (40:60 EG/W)

The simulation results reveal a consistent trend across both base fluids: The Al₂O₃ nanofluid exhibits the highest heat transfer coefficient, followed by the hybrid nanofluid (Al₂O₃ + CuO), with the CuO nanofluid showing the lowest performance. This trend remains largely unaffected by changes in Reynolds number, suggesting that the type of nanoparticle has a more pronounced influence on heat transfer enhancement than the flow rate.

Although hybrid nanofluids are often expected to combine the advantages of their individual components, the present findings indicate that the heat transfer performance of the Al₂O₃ + CuO hybrid is not superior to that of the pure Al₂O₃ nanofluid. The lower performance of the Al₂O₃–CuO hybrid nanofluid compared to pure Al₂O₃ may be attributed to the fixed 50:50 mixing ratio used in this study. This mixing ratio may not be optimal for maximizing synergistic thermal effects. Additionally, particle–particle interactions and possible reduction in effective thermal conductivity due to CuO's lower thermal conductivity than Al₂O₃ could also contribute. Prior studies [18, 21] indicate that hybrid nanofluids only outperform single-component nanofluids when composition, dispersion stability, and flow conditions are carefully optimized. This suggests that the enhancement mechanism in nanofluids is not simply additive; rather, it depends on several factors, including the mixing ratio of the constituent nanoparticles, their dispersion stability in the base fluid, particle–particle interactions, and specific flow conditions. Some previous studies have also reported that hybrid nanofluids demonstrate significant performance improvements only when configuration parameters and operating conditions are carefully optimized. Therefore, the design and application of hybrid nanofluids should be approached comprehensively and selectively, rather than assuming inherently superior performance.

The authors gratefully acknowledge the support of Thai Nguyen University of Technology through Project T2024-TS22.