Entropy generation and heatline visualization of magnetized thermal convection in an irregular enclosure filled with nanofluid and subject to non-uniform heating: FEM-machine learning hybrid approach

This investigation's primary objective is to analyze the fluid dynamics and heat transfer enhancement in an irregular enclosure. Several physical phenomena such as Alumina-water nanofluids, magnetic field, non-uniform heating and amplitude of non-uniform heating are taken into account. A thorough investigation using machine learning algorithms (MLAs) and computational fluid dynamics (CFD) is carried out. Initially, for CFD phase the coupled non-linear dimensionless governing equations of the considered problem are solved numerically by adapting a Finite Element method (FEM) using appropriate boundary conditions. The analysis considers a range of physical parameters, such as the Hartmann number (0–20), nanoparticle volume friction (0–0.04), Ryleigh number (103-105), offset temperature (0.1–1), frequency (1-10) and non-uniform heating amplitude (0.1–1). The numerical results are explicated in the form of streamlines, isotherms, heatlines structures. The impact of these factors on entropy generation is also explored. In addition, three popular MLAs, Random Forest (RF), Artificial Neural Network (ANN) and Support Vector Machine (SVM) were used to save the computational costs. The finding explore Hartmann number and nanoparticle volume friction has decreasing trend on total entropy production and heat transfer rate. The entropy production decreases up to 42.02 % for Hartman number and for nanoparticles it decreases 12.61 %. However, the non-uniform heating amplitude has opposite effect on irreversibility factor. Also, compared to other MLAs, the ANN model demonstrates more accurate predictions. The objective is to propose novel concept for enhancing the performance of cooling systems and optimizing energy utilization across diverse technical domain.