A machine learning approach to predicting the heat convection and thermodynamics of an external flow of hybrid nanofluid
Abstract This study numerically investigates heat convection and entropy generation in a hybrid nanofluid (Al2O3-Cu-water) flowing around a cylinder embedded in porous media. An artificial-neural-network is used for predictive analysis, in which numerical data are generated to train an intelligence algorithm and to optimize the prediction errors. Results show that the heat transfer of the system increases when the Reynolds number, permeability parameter, or volume fraction of nanoparticles increases. However, the functional forms of these dependencies are complex. In particular, increasing the nanoparticle concentration is found to have a non-monotonic effect on entropy generation. The simulated and predicted data are subjected to particle swarm optimization to produce correlations for the shear stress and Nusselt number. This work demonstrates the capability of artificial intelligence algorithms in predicting the thermohydraulics and thermodynamics of thermal and solutal systems.