Improving phase change heat transfer in an enclosure filled by uniform and heterogenous metal foam layers: A neural network design approach

Phase change materials (PCMs) inherently store and release large amounts of energy during phase transitions. In this research, the potential of two metal foam (MF) layers in enhancing the thermal energy storage unit's heat transfer was probed, with one layer having distinct attributes at an anisotropic angle, omega Utilizing the finite element method to understand the system dynamics, model accuracy was affirmed through rigorous checks. The impact of the heterogeneous parameter (0 < Kn < 0.3), heterogeneous angle (0 < omega < 90 degrees), and porosity 0.9 < epsilon < 0.975 was addressed on the melting process. To circumvent the high simulation costs, an artificial neural network (ANN) was trained on 7838 data points. Noteworthy findings indicate that a slight 7.5 % increase in porosity can reduce the melting time by 66 %. Moreover, the 0 degrees anisotropic angle emerged as the most efficient in heat transfer due to its superior thermal properties. The incorporation of ANN analytics was a pivotal shift, bypassing the traditionally high computational demands of phase change heat transfer studies. Once fully trained, the ANN adeptly demonstrated melting volume fraction (MVF) nuances under varied conditions. Further, optimal melting efficiencies were pinpointed at the omega = 0 degrees angle, with a specific porosity zone, epsilon similar to 0.925, showing minimal MVF and the benefits of a higher porosity (epsilon = 0.94) becoming evident at t = 3000 s. Ultimately, this investigation harmoniously integrates traditional analytical tools with ANN technology, offering profound insights into PCM heat transfer dynamics and laying the groundwork for future energy-efficient thermal storage solutions.