Data-Driven Surrogate Modelling for Industrial Scale-Up of Packed Bed Columns Using Residual Biomass for Pb(II) Removal

Pb(II) is a potentially toxic element that can cause problems in the respiratory, digestive, and nervous systems; it can also cause cancer. Most studies have been conducted primarily in the laboratory; therefore, it is necessary to find a way to predict performance on a large scale. Taking this into account, the objective of this study was to simulate an industrial-scale packed adsorption column for the removal of Pb(II) using Dioscorea rotundata-based biomass as an adsorbent, employing the Aspen Adsorption software. To this end, a parametric analysis was conducted to evaluate the performance of the bed under different operating conditions, using the Langmuir and Freundlich isotherms in conjunction with Linear Driving Force (LDF) kinetics. Removal efficiencies of up to 92.4% were observed for the Freundlich–LDF model and 92.4% for the Langmuir–LDF model. Additionally, machine learning algorithms were combined with statistical indicators (R2, RMSE) to analyze the models’ effectiveness. R2 values of up to 0.999 were obtained in the validation and testing phases. This research demonstrates a novel approach to predicting the potential performance of adsorption columns packed with biomass derived from organic waste in the field of engineering, using computational tools with machine learning algorithms.