Modelagem de separação seletiva de compostos químicos baseada em sistema aquoso bifásico: abordagens mecanísticas e híbrida

Abstract

This work presents the mathematical modelling of the selective separation of high-value compounds in alcohol-salt type Aqueous Two-Phase Systems (ATPS). It represents the third stage of a continuous research effort, which previously involved the evaluation of phase diagrams and batch separations of vanillin and ascorbic acid, followed by the experimental study of a continuous extractor. Three mechanistic models of progressive complexity were developed to represent mass transfer and phase thermodynamic equilibrium. However, their limitations in predicting dynamics at low concentrations highlighted shortcomings in traditional approaches. In response, a new approach was proposed, combining mechanistic equations with phenomenological correlations derived from data-driven analysis. This hybrid model increased accuracy by depicting non-linear behaviours neglected by purely mechanistic models. Validation through confidence interval analysis under different operating protocols, including countercurrent operation, confirmed the predictive capability of the hybrid model, reducing residual standard deviations by up to 50% compared to the best mechanistic results. Challenges remain in predicting the partitioning of ascorbic acid in systems with real food samples, due to pH variations and the presence of biomolecules with unforeseen intermolecular interactions. This study deepens the understanding of partitioning mechanisms in ATPS and contributes to the advancement of modelling such systems, providing a solid foundation for future applications in more complex continuous operations. The critical evaluation of the proposed models reinforces the relevance of integrated approaches in developing solutions applicable to the selective separation of biocompounds at laboratory scale and with potential industrial implementation.