Effect of particle size on second law of thermodynamics analysis of Al2O3 nanofluid: Application of XGBoost and gradient boosting regression for prognostic analysis

Abstract

In this study, the current research delves into the influence of nanoparticle size on turbulent forced convective heat transfer, entropy generation, and friction factor. The investigation focused on three different sizes (30 nm, 50 nm, and 80 nm) of Al2O3 nanoparticles (NPs) suspended in a water-based nanofluid (NF) with a 1 vol% concentration flow in a circular tube. The nanoparticles (NPs) were characterized using various characterization techniques. The stability and pH of the NF were determined, and its viscosity (VST) and thermal conductivity (TC) were measured at a temperature of 60 °C. Heat transfer experiments were conducted with varying particle sizes and Reynolds number (Re), maintaining a fluid inlet temperature of 60 °C. The results indicated that the NF containing 30 nm particles exhibited higher VST and TC compared to the other samples and the base fluid. The maximum enhancement in Nu for Al2O3 (30 nm) and Al2O3 (80 nm) NFs is 60.7 and 18.5 % greater than that of base fluid, respectively. The maximum and minimum total entropy generation (Sgen,T) value of 0.499 and 0.286 observed for base fluid and Al2O3 NF (30 nm), respectively at low Re. The highest friction factor enhancement for Al2O3 NF (30 nm) exceeded by 9.4 % compared to the base fluid, and the maximum thermal performance factor observed for Al2O3 NF (30 nm) was 1.57. Finally, regression analysis was employed to establish correlations for estimating Nu and friction factor values. Prognostic models were developed using two sophisticated machine learning algorithms, XGBoost and Gradient Boosting Regression (GBR). Both models demonstrated exceptional prediction abilities, achieving over 99 % accuracy rates based on the experimental data.