Application of Artificial Neural Networks and Factorial Design Analysis for Predicting the Interaction of Influencing...

Ex situ aqueous-based CO₂ mineralization of magnesium-bearing mine wastes presents a promising pathway for carbon sequestration and resource recovery, relying on both direct and indirect carbonation approaches. However, their efficiency depends on optimizing key process parameters, such as solid/liquid particle size, pretreatment, temperature, pressure, pH, and their interactions. This study provides a comprehensive analysis of CO₂ mineralization in magnesium-based mine wastes, utilizing an extensive data library from existing literature. Both direct and indirect mineralization approaches, including extraction and carbonation, were assessed to understand key process parameter interactions. Utilizing artificial neural networks (ANN) and a 3^k full factorial design coupled with Analysis of Variance (ANOVA), the study investigated nonlinear relationships and the statistical significance of influencing factors for these processes. Key findings indicated that for the extraction process, optimization is driven by feedstock material pretreatment, extraction agent, temperature, and particle size. For direct CO₂ carbonation, prioritization of pretreatment, CO₂ concentration, and particle size reduction are most important, with a secondary focus on solution chemistry. However, for the indirect carbonation approach, the dominance of carbonation assisting agents and solution pH highlights the importance of solution chemistry in aqueous carbonation rather than the physical properties of the feedstocks. These insights provide a robust framework for understanding the complex relationships between different process variables that play a pronounced role in the CO₂ mineralization of magnesium-rich mining materials. Understanding influential factors enables better design and optimization of processes for enhanced efficiency and sustainability.