Thermodynamic analysis of nanofluid with aggregated and non-aggregated nanoparticles

Regenerative cooling using nanofluids has been shown to improve the thermal performance of rocket engine thrust chambers and nozzle walls substantially. This study examines the use of a kerosene-alumina (Al₂O₃) nanofluid as an innovative coolant for semi- cryogenic rocket engines, emphasizing its superior thermophysical properties relative to conventional coolants, despite constraints in high-temperature environments. The focus is to analyze the thermodynamic behavior of nanofluid incorporating both aggregated and non-aggregated Al₂O₃ nanoparticles. To capture nanoparticles aggregation, modified Krieger-Dougherty and Maxwell-Bruggeman models are integrated into the formulation. The complex influences such as Hall current and flow through porous media are also taken into consideration. The governing equations with thermal and velocity slip boundary conditions are simplified using the lubrication approximation to enable efficient analysis and numerically simulated to collect reference datasets across eight nanofluid configurations. A machine learning employing Bayesian Regularization Back Propagation Scheme (BRBPS) is developed using partitioned simulation data (70 % training, 15 % testing, 15 % validation). Aggregated and non-aggregated nanofluids reveal similar patterns in entropy generation and velocity profiles. However, non-aggregated nanoparticles result in more thermodynamic irreversibility than aggregated nanoparticles when the conditions are the same. Bejan number decreases with higher permeability. Both aggregated and non-aggregated nanoparticles compress trapped bolus size with increasing nanoparticles volume fraction.