Finite element and neural network modeling of thermal energy storage and entropy behavior in a wavy porous triangular enclosure with nano-encapsulated phase change materials
Yasir Ul Umair Bin Turabi, Shafee Ahmad, Siddeeq Ahmad
Abstract
Efficient thermal energy storage is crucial for sustainable technologies, including solar energy harvesting, electronic device cooling, and battery thermal management. This study investigates the thermal and entropy behavior within a magnetohydrodynamic natural convection environment filled with nano-encapsulated phase change materials (NEPCMs) in a wavy porous triangular enclosure containing a centrally embedded cold cylinder. The main objective is to optimize heat transfer performance and energy storage capabilities through geometric and thermophysical enhancements, while also minimizing irreversibility. The finite element method (FEM) is employed for numerical simulation, while an artificial neural network (ANN), trained using the Levenberg–Marquardt algorithm, provides high-accuracy predictive modeling. Results reveal that increasing Rayleigh number, wall undulations, and NEPCM volume fraction significantly enhance the Nusselt number, indicating improved convective heat transfer. Entropy generation analysis shows that optimal Stefan number and fusion temperature minimize irreversibility. The ANN model achieves near-perfect agreement with FEM data [regression (R) = 0.999 99; mean square error ≈ 0.0014], offering a reliable predictive framework. This integrated computational intelligent approach presents a novel pathway for designing high-efficiency latent heat thermal energy storage systems. The findings hold promise for advanced applications in smart renewable energy systems, electronic cooling devices, and battery management technologies.