Nanofluids, consisting of suspended nanoparticles, offer significant potential for enhancing heat transfer in thermal systems. However, predicting heat transfer rates remains challenging, particularly under varying temperatures, due to the common assumption of constant thermophysical properties. This study addresses this limitation by incorporating temperature-dependent properties into a numerical analysis of natural convection in water-based nanofluids inside a rectangular enclosure. A single-phase model evaluates the effects of nanoparticle material, volume fraction, and enclosure aspect ratio on heat transfer. Three nanoparticle materials (TiO2, CDiamond, Ag) are considered over Rayleigh numbers \(\left(\text{Ra}\right)\) from \({10}^{3}\) to \({10}^{6}\) . The findings reveal that incorporating variable properties improves Nusselt number predictions, particularly at lower \(\text{Ra}\) and nanoparticle concentrations. The variable properties model corrects Nu underprediction by up to \(39\%\) at low \(\text{Ra}\) and reduces overpredictions at high \(\text{Ra}\) by up to \(7\%\) compared to constant property models. The effect of nanoparticle concentration varies with \(\text{Ra}\) , influencing heat transfer trends. Silver nanoparticles achieve the highest heat transfer performance, with enhancements up to \(43\%\) , while TiO2 shows the lowest. Aspect ratio \((\text{AR}\) ) also significantly impacts heat transfer, with \(\text{AR}=1\) yielding the greatest enhancement with TiO2 compared to \(\text{AR}=1.5\) or \(4\) . These results highlight the limitations of assuming constant properties and demonstrate the interplay of nanoparticle concentration, \(\text{Ra}\) , material, and \(\text{AR}\) in nanofluid convection. Thus, incorporating temperature-dependent models is essential for accurately predicting nanofluid heat transfer rates, particularly in applications such as electronics cooling and advanced heat exchangers.
