This paper aims to evaluate how artificial neural network (ANN)-based multilayer perception (MLP) predicts the behavior of the system under different conditions, optimizes thermal systems for desired performance and analyzes and designs buoyancy-driven convection systems in a computationally efficient way. In view of such a significant background, the objective of the current study deals with an ANN-based MLP on buoyancy-driven convection within a nano-encapsulated phase change material (NEPCM)-loaded inclined enclosure comprising a non-Newtonian fluid subject to an inclined magnetic field and an Arrhenius exothermic reaction.
Governing equations would be solved using finite element method. The ANN-based MLP technique is implemented to predict the thermal behavior of the system based on data generated from numerical simulations. The studied parameters and their ranges adopted are power-law index (n), Frank-Kamenetskii number (FK), thermal conductivity ratio parameter (k*), Rayleigh number (Ra), Hartmann number (Ha), Stefan number (Ste) and solid volume fraction.
The major findings from the present computational fluid dynamics and ANN analysis are that Nusselt numbers (Nulocal, Nuave) augment by 1016.02% due to an increase in Ra from 104 to 106. When the conducting blocks are shifted from case A to case B, Nulocal and Nuave fall and they ameliorate with shifting of the blocks from case B to case C. Ecological coefficient of performance upsurges due to a rise in power-law index, while it diminishes due to an increase in FK and Ha. The significance of the work includes its diverse usages in various fields such as solar energy systems, electronic cooling, thermal energy storage in buildings and heating, ventilation and air conditioning systems.
To the best of our knowledge, no such analysis has been conducted to date.
