Thermal Performance Optimization of a Phase Change Material (PCM)-Based Thermal Energy Storage System

Abstract

The mismatch between the availability of thermal energy and demand for it is a principal obstacle to the wider adoption of solar-thermal and waste-heat-recovery systems. Latent heat thermal energy storage (LHTES) using phase change materials (PCMs) offers compact, near-isothermal storage, yet the organic PCMs that store energy well conduct it poorly (0.15–0.25 W/m·K), which throttles charging and discharging power. This study addresses that limitation through a combined numerical, experimental and data-driven optimization of a vertical shell-and-tube LHTES module charged by hot water. A commercial paraffin wax (measured melting range 40.2–44.1 °C, latent heat 178.4 kJ/kg) was characterised by differential scanning calorimetry. Two enhancement strategies—longitudinal copper fins and dispersion of aluminium-oxide nanoparticles—were investigated jointly rather than in isolation. A transient conjugate model using the enthalpy-porosity formulation with buoyancy-driven convection was developed in ANSYS Fluent, made grid- and time-step-independent, and validated against fifteen calibrated thermocouples to within 6.8%. A Box–Behnken design was expanded to 180 cases to train a 4-12-8-2 artificial neural network that predicts complete melting time and stored energy with a testing coefficient of determination of 0.9921. Coupling the surrogate to a non-dominated sorting genetic algorithm (NSGA-II) generated the melting-time–stored-energy Pareto front, from which a preferred design was selected by TOPSIS. The optimum—eight fins of 22.5 mm height with 2.0% nanoparticle loading at 70 °C inlet—reduced complete melting time from 218.0 to 74.6 min (a 65.8% reduction) for only a 4.1% loss of stored energy, raised average charging power from 62 to 173 W, and improved exergy efficiency from 41.2% to 52.7%. Fin geometry and nanoparticle loading interact sub-additively, and the nanoparticle benefit saturates near 2% by mass.