Machine learning-assisted study on modulating aluminosilicate framework of high-silica coal ash via calcium extracted from waste oyster shells

This study addresses excessively high ash fusion temperatures (AFTs) in high‑silicon‑aluminum coal from Guizhou, limiting its application in entrained-flow gasification. Using Jinsha (JS) and Nayong (NY) coals, we investigate waste oyster shell (OS) as a calcium-based flux and its regulation mechanism. Analyses including ash fusion tests, SEM, XRF, XRD, and FactSage simulations were performed to examine melting behavior, mineral evolution, and microstructural changes upon oyster shell ash (OSA) addition. An exploratory machine-learning analysis was performed, providing qualitative support for factors controlling flow temperature (FT). Results show that OSA addition effectively reduces AFTs of both JS ash (JSA) and NY ash (NYA). At 17.5 wt% OSA, FTs of JSA and NYA decrease significantly to 1370 °C from >1500 °C and 1468 °C, respectively. Mechanistically, the fluxing effect on JSA is attributed to Ca 2+ -induced disruption of the aluminosilicate network. For NYA, effective FT reduction occurs only above a 10.5 wt% OSA threshold, due to multicomponent eutectic formation. Machine learning identifies the Ca/Si ratio as the most dominant factor negatively correlated with FT, offering data-driven support for experimental observations. This study provides an economically viable fluxing option for clean utilization of high‑silicon‑aluminum coal and a theoretical basis for waste OS resource recovery.