Numerical study of tri-hybrid nanofluids in a rectangular cavity with an enclosed circle via COMSOL and Levenberg-Marquardt method

Abstract

This work focuses on a numerical simulation of magneto mixed convection transport in electrically conducting tri hybrid nanofluids that is enclosed in a two dimensional rectangular lid driven cavity with a cold circular obstacle. Dissipative processes due to viscous dissipation and Joule heating are taken into account and the non-dimensional governing equations are resolved by Galerkin finite-element method in COMSOL Multiphysics. The effects of the major controlling parameters, i.e. the Hartmann number $\left(0.1\le M\le 20\right)$, Reynolds number $\left(100\le Re\le 500\right)$, the Richardson number $\left(0.1\le Ri\le 10\right)$, and the nanoparticle volume-fraction coefficients ($0\le \varnothing \le 0.06$), on the flow structure and heat-transfer characteristics are systematically evaluated. These findings indicate that Ha increase inhibits fluid motion by the force of Lorentz forces, thus minimising convective exchange of heat at the moving heated wall. On the other hand, increased values of Re significantly increase fluid flow and thermal mixing resulting in increased local and mean Nusselt numbers. Tri-hybrid nanoparticles $\left(Au,Ag,Ti{O}_2\right)$ enhance the thermal capability of the base fluid by increasing the effective thermal conductivity, thereby, enhancing the overall heat-transfer rate in the base fluid. A high Richardson number works the flow field in the direction of buoyancy-dominated convection, dampens the contribution of forced-convection, and reduces the transfer of heat to that moving away of the upper moving wall. It uses an artificial neural network that has been trained using the Levenberg-Marquardt algorithm to forecasts and optimise the average Nusselt number, with excellent correspondence with computed data; the regression coefficient approaches one, and the mean squared error is small. High Reynolds number, low Hartmann number, low Richardson number, and moderate volume fractions of nanoparticle yield the best results with regard to heat-transfer performance, and thus, the study irrevocably supports the capability of integrating tri-hybrid nanofluids with data-driven optimization in the context of advanced thermal-management operations.