Application of an artificial intelligence model for natural convection of nano-encapsulated phase change materials (NEPCMs) confined in a porous square enclosure with an inclined elliptical heated block

This work examines the thermal behavior and performance of a porous enclosure filled with Nano-Encapsulated Phase Change Materials (NEPCMs) using an artificial intelligence model and numerical simulation. The advantages of Phase Change Materials (PCMs) and nanofluids are combined in this novel material. The enclosure features an inclined elliptical obstacle at the center, heating the cavity, while the right vertical wall serves as a cooling source. The governing equations are transformed into a dimensionless form and solved using a Finite Element Method (FEM) code. Various influential parameters affecting flow and thermal characteristics are explored, including fusion temperature, Stephan number, nanoparticles' volume fraction, obstacle inclination angle, and Rayleigh and Darcy numbers. The findings show that when the fusion temperature is set at 0.5 and the Stephan number is low, the best heat transfer performance occurs. Additionally, introducing only 1% of NEPCMs into the base fluid improves heat transfer by up to 20%. A Levenberg-Marquardt backpropagation model is employed for Artificial Neural Network (ANN) training based on numerical data to further enhance the analysis. The model features two hidden layers with an optimal number of neurons, achieving a high R2 of 0.991.