Experimental analysis, modelling, and optimisation of alkaline leaching in coal fly ash treatment

The present study introduces a novel integration of Gaussian Process Regression (GPR) modelling and Particle Swarm Optimisation (PSO) to improve the efficiency of alkaline leaching of coal fly ash (CFA). The selected operating variables for the alkali leaching process include temperature, leaching time, concentration of the alkalis (<i>NaOH</i> and <i>KOH</i>), and the liquid-to-solid ratio. A GPR model was employed for data fitting of the leaching process, yielding high predictive accuracy with <i>R</i>² values of 0.9978 for <i>SiO₂</i> dissolution, 0.9742 for <i>Al₂O₃</i> dissolution, and 0.9945 for Al/Si ratio in the <i>NaOH</i>-treated CFA process. In the <i>KOH</i>-treated CFA process, the GPR model achieved <i>R</i>² values of 0.9645 for <i>SiO₂</i> dissolution, 0.9873 for <i>Al₂O₃</i> dissolution, and 0.9960 for Al/Si ratio. Under optimised conditions, both <i>NaOH</i>- and <i>KOH</i>-treated leaching processes demonstrated an effective desilication of CFA, with <i>NaOH</i> showing higher silica dissolution and <i>KOH</i> yielding greater alumina recovery. The resulting <i>Al/Si</i> ratios further confirmed the efficiency of treatment, with the higher ratio in the <i>NaOH</i> process reflecting more effective silica removal. These findings demonstrate the efficacy of using PSO in conjunction with GPR models to optimise leaching processes, offering a significant advancement in the efficient processing of CFA through precise control of operational parameters.