Dynamic Assessment and Optimization of Thermal Energy Storage Integration with Nuclear Power Plants Using Machine Learning and Computational Fluid Dynamics

This study integrates thermal energy storage (TES) systems with nuclear power plants (NPPs) utilizing phase change materials (PCMs). It employs computational fluid dynamics (CFD) simulations and machine learning techniques to improve the overall efficiency and profitability of NPPs. The novelty of this research extends beyond merely analyzing the influence of PCM thermophysical properties, design parameters, and input conditions on system performance. The objective is to develop a TES system that can be effectively integrated with NPPs by addressing critical challenges, including the dynamic assessment of input parameters and the utilization of available excess energy in response to real-time demand fluctuations. For the analysis, 2,500 CFD simulations were performed to assess the phase change behavior within a vertical annular channel. Key design factors, such as heat transfer fluid injection conditions and various PCM properties, were systematically analyzed. The extensive dataset from numerical simulations was utilized to employ an artificial neural network (ANN) and a multi-objective genetic algorithm (MOGA) for optimizing the TES system, with an emphasis on minimizing injection velocity, reducing charging duration, and maximizing stored energy. The numerical and machine learning models were also explicitly evaluated against experimental data under different conditions. The findings indicate that design parameters have a substantial effect on the performance of the TES system, with the ANN model effectively predicting PCM melting time and energy storage capacity. This study presents a complete framework for an optimized thermal energy storage integration strategy specifically designed for NPPS, offering a scalable and efficient method for improving energy storage and utilization of excess energy during off-peak hours.