Abstract

Membrane distillation-crystallization (MDCr) has emerged as a promising alternative for the production and purification of different products; however, its operational performance is strongly limited by membrane fouling associated with crystal nucleation, growth, and deposition. In this work, fouling behavior in MDCr applied to aqueous LiOH solutions, an industrially relevant system for battery-grade lithium production, is analyzed under different operating conditions, building on a previous study due to the availability of detailed experimental data. Given the increasing demand for lithium and the need to develop more sustainable production routes, improving the understanding of such systems is essential for advancing emerging technologies in lithium processing. Two modeling approaches are compared: conventional filtration models (CFM) based on Hermia's laws, and integrated filtration models (IFM), which explicitly incorporate pore-blocking mechanisms within a phenomenological framework of mass and heat transfer. Both approaches were applied at different process scales: membrane distillation stage (MD-s), the membrane distillation-crystallization stage (MDCr-s), and the whole process without differentiation (full-MDCr). Both modeling approaches adequately reproduced the decline in transmembrane flux. However, the IFM framework demonstrated superior physical consistency with SEM-EDX observations and enhanced predictive capability, particularly under conditions that promote accelerated supersaturation. For the full-MDCr stage, R2 up to 0.97 were obtained for IFM compared to 0.92 for CFM, together with significantly lower errors. Similar trends were observed during the MD-s. In contrast, comparable performance was observed for both approaches during MDCr-s, attributed to the strong flux fluctuations induced by crystal formation. SEM-EDX analysis confirmed the coexistence and transition of standard, intermediate, and complete pore-blocking mechanisms depending on operating conditions. Overall, the results demonstrate that embedding pore-blocking laws within a phenomenological modeling framework improves both predictive accuracy and physical consistency in describing fouling in MDCr processes. The proposed approach provides a robust basis for interpreting fouling mechanisms and can support future studies aimed at process design and control.