Machine-Learning Enhanced Analysis of Mixed Biothermal Convection of Single Particle and Hybrid Nanofluids within a Complex Configuration

Transport phenomena in a hybrid or single-particle nanofluid over a conical body embedded inside a porous medium are investigated. The fluid contains homogeneously mixed nanoparticles and live cells that are able to migrate, collectively sculpturing a thermo-biosolutal system. Transport processes including mixed convection as well as species and cell transfer are simulated using a similarity technique. As the problem involves a large number of parameters with complicated interactions, machine learning is applied to predict a wide range of parametric variations. The simulation data are used to build an intelligent tool based on an artificial neural network to predict the behavior of the system. This also aids the development of precise correlations for nondimensional parameters dominating the transport phenomena. The results indicate that lower values of the motile Lewis number and a higher mixed convection parameter enhance the Nusselt number. However, it is contained respectively by the increment of the Peclet number and increases in the bio Rayleigh number. It is further shown that an increase in the Prandtl number enhances the Sherwood number and makes the motile microorganisms more uniform. The Peclet number directly influences the transport of heat, mass, and microorganisms. This study clearly demonstrates the abilities of combining numerical simulations with machine learning to significantly extend and enrich analysis of problems with large numbers of variables. The findings also pave the way for predicting behaviors of complex thermo-biosolutal systems without resorting to computationally demanding simulations.

Automated summary

This study investigates transport phenomena in a hybrid or single-particle nanofluid over a conical body embedded inside a porous medium. Using a similarity technique, the researchers simulated transport processes including mixed convection, species transfer, and cell transfer. Machine learning was applied to predict a wide range of parametric variations, and the simulation data were used to build an intelligent tool based on an artificial neural network. The findings demonstrate the abilities of combining numerical simulations with machine learning to significantly extend and enrich analysis of problems with large numbers of variables.