A state-of-the-art-review on phase change materials integrated cooling systems for deterministic parametrical analysis, stochastic uncertainty-based design, single and multi-objective optimisations with machine learning applications

Renewable energy utilisation, latent energy storage, optimal system design, and robust system operation are critical elements for carbon-free buildings and communities. Machine learning methods are effective to assist the energy-efficient renewable systems during multi-criteria design and multi-level uncertainty-based operation periods. However, the current literature provides little knowledge on this topic. In this study, a state-of-the-art-review on phase change materials for cooling applications is presented, in terms of smart ventilations, intelligent PCMs charging/discharging, deterministic parametrical analysis, stochastic uncertainty-based performance prediction and optimisation. Furthermore, technical effectiveness of machine learning methods in single and multi-objective optimisations has been presented, through hybrid PCMs integrated renewable systems. Multivariables involved in the review include thermo-physical, geometrical and operating parameters of PCMs. Multi-criteria employed in the review include heat transfer rate, cooling energy storage density, heat storage and release efficiency, and indoor thermal comfort. The literature review presents technical challenges, such as tradeoff solutions between computational accuracy and efficiency, generic methods for effective selection amongst multi-diversified optimal solutions along the Pareto front, the general methodology for multi-level uncertainty quantification, smart controllers with accurate predictions under high-level parameters’ uncertainty and stochastic occupants’ behaviors. The future outlook and recommendations of machine learning methods in PCMs integrated cooling systems have also been presented as avenues for upcoming research.