Experimental studies on latent heat capacity of hybrid nano-enhanced phase change materials using artificial neural network for energy storage applications

Abstract

The present study investigates the enhancement of latent heat capacity and thermal stability in hybrid nano-enhanced solid–solid phase change materials (SS-PCMs) using Neopentyl Glycol (NPG) as the base material. The key contribution of this work lies in incorporating copper oxide (CuO) and titanium dioxide (TiO₂) nanoparticles to optimize thermal performance and ensure long-term stability. CuO (1 wt.%) and TiO₂ (0.1, 0.3, 0.5,0.7 wt%) were introduced into the matrix, and the thermal properties were systematically evaluated using Differential Scanning Calorimetry (DSC) and Thermogravimetric Analysis (TGA) before and after 500 thermal cycles. The optimal composition, consisting of 1 wt% CuO and 0.3 wt% TiO₂, demonstrated an initial latent heat capacity of 117 J/g, which increased to 123 J/g post-cycling, indicating exceptional thermal stability and phase retention. To further enhance predictive capabilities and reduce experimental costs, an artificial neural network (ANN) model was developed using the Keras API in Python to estimate thermal behaviour. The model achieved a high coefficient of determination (R2 = 0.9479) and a low root-mean-square error (RMSE = 2.0307), underscoring its accuracy and reliability. These findings establish the efficacy of hybrid nanoparticle incorporation in improving SS-PCMs’ thermal properties and emphasise the viability of machine learning as a robust predictive tool, mitigating the time and economic constraints associated with extensive experimental investigations.