Phase change material (PCM)-based latent heat thermal energy storage architecture enabled by multi-stage annular fins: Shapley additive explanations-based sensitivity analysis and intelligent methods

Phase change material (PCM)-based latent heat thermal energy storage (LHTES) systems have attracted considerable attention because of the large amount of energy they can store and their ability to bridge the gap between thermal energy supply and demand. Nonetheless, their practical application remains limited because PCMs conduct heat poorly, resulting in slow charging and incomplete utilization of available latent heat capacity. In this study, a novel LHTES architecture incorporating a multi-stage annular fin configuration was proposed to enhance heat-transfer performance while maintaining effective PCM utilization. The proposed system consists of a separable HTF plate and PCM storage enclosures, where the staged annular fins provide multiple conductive pathways and improve thermal distribution within the PCM domain. A comprehensive numerical and artificial intelligence-assisted optimization framework was developed to determine the optimal fin geometry. An artificial neural network (ANN) surrogate model was developed from CFD-generated samples to predict stored thermal energy under short-term and long-term charging conditions. The ANN model was then used with NSGA-II to identify favorable solutions within the design domain, while the TOPSIS method was used to select the final optimal configuration. Furthermore, SHAP analysis was used to interpret the contribution of the geometric parameters to thermal storage performance. The results showed that the developed ANN model accurately predicted thermal storage capacity, achieving correlation coefficients of 0.994 and 0.996 for short-term and long-term charging conditions, respectively. SHAP analysis showed that the first-stage annular fin radius dominated short-term energy storage performance, whereas the second-stage annular fin radius contributed most under long-term charging conditions. The optimized annular-fin configuration achieved a stored thermal energy of approximately 29,174 kJ after 360 min, compared with 13,075 kJ for the fin-free system, an enhancement of approximately 123.1%. Moreover, the optimized design increased the final liquid fraction from 0.307 to 0.869, an improvement of approximately 183% in PCM melting utilization.