This study analyzed the heat transfer potential of nanofluids (suspension of Al2O3 nanoparticles in 60:40 mixtures of ethylene glycol and water by volume) compared to the base fluid by modeling the flow in both laminar and turbulent flow regimes in a flat tube of a radiator using both single-phase and Eulerian–Eulerian multiphase approaches of computational fluid dynamics (CFD) simulation. Nusselt number is calculated by varying the Reynolds number for both the models and compared with the theoretical correlations and the experimental data available in the literature. The effect of change in volumetric concentrations, inlet velocities, inlet temperatures, and particle sizes on the heat transfer performance parameters of nanofluids was evaluated. The multiphase approach showed a 5–45% greater increase in Nusselt number than the single-phase approach for the same volume fraction up to 1% and thereafter multiphase approach showed an even higher Nusselt number. The multiphase model revealed that raising the volume fraction up to 2% increases the Nusselt number of nanofluid by approximately thrice that of the base fluid but after that further increase in volume fraction does not show any significant increment. With the increase in Reynolds number, the Nusselt number, as well as surface heat transfer coefficient, increased for both laminar and turbulent flow regimes. Skin friction coefficient, pressure drop across inlet and outlet, and pumping power required increased slightly with the increase in volume fractions. Also, on varying the nanoparticle size, superior performance was observed at the particle size of 25 nm.