This research presents a numerical and machine learning steady three-dimensional magnetohydrodynamic ( MHD ) with heat generation to investigate the mixed convection in a cone-cubic enclosure. The enclosure is subjected to a uniform magnetic field, while a cone-shaped obstacle is positioned inside the cavity. The significance of the current study lies in understanding the synergistic interactions among internal heat generation, magnetic damping, and buoyancy in a complex geometrical enclosure and in deriving an efficient data-driven predictor for thermal transfer. The finite element method implemented in COMSOL Multiphysics is employed to perform numerical simulations, and the resulting dataset is used to train an Extreme Gradient Boosting (XGBoost) model for prediction of the Nusselt number. The results show that thermal transfer is positively correlated to Richardson number but shows an adverse relation to Hartman number and heat source with a smaller cone back radius always having superior thermal performance as compared to a larger cone. The XGBoost model mimics the numerical results well, which is a demonstration of the good predictive power. These insights provide useful guidelines for the design and optimization of MHD- based advanced thermal systems, e.g., cooling systems, heat exchangers and energy storage devices.
Ashique et al. (Mon,) studied this question.