D.Y. Chen, Y.Y. Qiao
Shanghai Jiao Tong University,
China
Keywords: Phase change material, Thermal Energy Storage, Bayesian network, Lattice Boltzmann method, Predictive modeling
Summary:
This study presents a novel approach integrating the Lattice Boltzmann Method (LBM) with Bayesian Network (BN) to accelerate computational fluid dynamics (CFD) simulations for phase-change material (PCM) heat transfer in thermal energy storage (TES) systems. PCM-based TES units are crucial in enhancing energy efficiency and addressing renewable energy intermittency. However, the computational expense of high-fidelity CFD simulations often limits their integration into real-time predictive control frameworks for TES systems. In the proposed methodology, LBM serves as the primary numerical tool to simulate the complex melting and solidification processes of PCM, providing detailed insights into flow dynamics and heat transfer mechanisms. BN is incorporated to reduce computational demands by acting as a surrogate model that captures key dependencies and uncertainties within the system. By learning from LBM-generated data, BN predicts critical parameters such as phase transition dynamics and temperature distributions with high accuracy, enabling significant acceleration of simulations without compromising fidelity. This hybrid framework facilitates the seamless integration of CFD models into predictive control algorithms, offering more accurate and adaptive control strategies for TES systems.