Design improvement of latent heat thermal energy storage in wavy channel enclosures using neural networks

Abstract

This study conducts an in-depth analysis of latent heat thermal energy storage (LHTES) in a uniquely designed wavy enclosure filled with anisotropic copper metal foam saturated by paraffin wax. It is subjected to a high-temperature fluid stream on its top and bottom wavy walls, with its side walls well-insulated. The enclosure serves as a test bed to scrutinize the melting dynamics of paraffin wax as a Phase Change Material (PCM). The primary relations for conservation of mass, momentum, and energy were written as partial differential equations and solved using the finite element method. Among the novel aspects of this work is the exploration of wavy wall topographies defined by wave amplitude and wave number. The investigation reveals that an increase in wave amplitude generally boosts key parameters like melt volume fraction, stored thermal energy, and average temperature in the LHTES system, albeit at the slight expense of reduced heat flux at the top wavy wall. Furthermore, the study investigates the impact of anisotropic angles in the copper foam and uncovers that a 90° anisotropic angle remarkably elevates all performance indicators without requiring additional material or weight. Increasing the anisotropic angle from 0° to 90° reduces the melting duration from 3750 s to 2500 s, representing a 33 % decrease in melting time. Leveraging high-accuracy neural network (NN) models, the research also offers insights into the system's response to variations in control variables like wave amplitude, wave number, and porosity. This approach significantly reduces computational time, allowing for a comprehensive analysis of many design configurations that would otherwise be computationally prohibitive.