A sustainable solution to the selection of thermal energy storage materials for concentrated solar power applications and predicted suitable machine learning algorithm

This study investigates the development of a carbonate-based eutectic phase change material (PCM) and evaluates the hot corrosion behavior of containment alloys for high-temperature concentrated solar power (CSP) applications. A eutectic mixture of sodium carbonate (55 wt%) and diatomite (45 wt%) was thermally characterized using DSC, TGA, and FTIR through which confirming its stability up to 800 °C with an exothermic peak at 849.1 °C and an enthalpy change of 48.1 J/g. Hot corrosion tests were conducted at 800 °C for exposure durations up to 500 h on Inconel 686, Hastelloy C2000, Inconel 59, and SS316 in the eutectic PCM environment. Amongst, Inconel 59 exhibited superior corrosion resistance, with the corrosion rate decreasing significantly from 627.15 μm/year at 100 h to 25.09 μm/year at 500 h, indicating the formation of a stable and protective oxide scale. In contrast, SS316 showed severe degradation, with corrosion rates increasing from 1756.01 μm/year to 3706.44 μm/year due to the formation of non-protective iron oxides and chromium depletion. XRD, SEM, and EDS analyses revealed the dominant formation of NiO and Cr 2 O 3 protective layers in Ni-based alloys, while SS316 exhibited porous Fe 2 O 3 and FeCr 2 O 4 phases. Based on the experimental corrosion data, machine learning models were applied to predict corrosion behavior, with the stochastic gradient descent (SGD) algorithm demonstrating reliable performance (R 2 = 0.923). The results revealed that Inconel 59 is a promising containment material for carbonate-based thermal energy storage systems and demonstrate the potential of machine learning approaches for corrosion prediction in CSP environments.