Machine Learning-Optimized Phase Change Material Selection for Building Energy Storage in Hot-Arid Climates

摘要

Phase change materials (PCMs) offer passive thermal energy storage for reducing cooling loads in buildings, but optimal PCM selection for specific climates remains challenging due to the complex interplay of melting temperature, latent heat, thermal conductivity, and cost. We develop a Bayesian optimization framework coupled with a validated EnergyPlus building energy model to identify optimal PCM properties for representative hot-arid cities (Riyadh, Phoenix, Alice Springs). The framework evaluates 2,500+ PCM candidates from a curated database and identifies that PCMs with melting points of 26-28°C and latent heat >200 J/g achieve maximum annual cooling energy savings of 32-41% when integrated as 15mm wallboard layers. A cost-benefit analysis shows payback periods of 4.2-6.8 years at current PCM prices, decreasing to 2.1-3.5 years with projected 2030 manufacturing costs.