Numerical analysis of natural convection in a partially heated trapezoidal cavity filled with hybrid nanofluid: A comparative study using ANN and RSM approaches

This study presents a comprehensive investigation of natural convective heat transfer in a partially heated trapezoidal cavity filled with a C u − M g O − H 2 O hybrid nanofluid, employing both a numerical approach based on the finite element method (FEM) and an artificial neural network (ANN) technique. The novelty of this work lies in the combined use of hybrid nanofluids with advanced computational and data-driven methods to predict and optimize thermal performance in complex cavity geometries. Copper ( C u ) and magnesium dioxide ( M g O ) nanoparticles dispersed in water ( H 2 O ) constitute the hybrid nanofluid. The effects of key parameters-including the nanoparticle volume fraction ( 0 ≤ ϕ ≤ 0.06 ), Hartmann number ( 0 ≤ H a ≤ 100 ), and Rayleigh number ( 10 3 ≤ R a ≤ 10 6 )-are thoroughly analyzed. The results are presented using 2D and 3D contour plots, revealing that the use of C u − M g O − H 2 O hybrid nanofluid enhances the thermal transfer rate by up to 6 % compared to the base fluid. Additionally, the applied magnetic field reduces the average Nusselt number ( N u a v ), while an increase in R a elevates it by up to 5.28 %. Furthermore, the application of response surface methodology (RSM) provides a predictive relationship between N u a v and the governing parameters, illustrated through 2D and 3D response surfaces. The combined use of FEM and ANN offers a robust and efficient framework for analyzing thermal systems and fluid flow, emphasizing the novelty of integrating numerical and machine learning techniques with hybrid nanofluids for enhanced heat transfer performance.