Experimental Design-Guided Optimization of Pervaporative Dehydration of an Esterification Mixture

This study investigates the pervaporation dehydration of a quaternary esterification mixture containing water, 2-ethylhexyl acrylate, 2-ethylhexanol, and propionic acid using a pilot-scale HybSi membrane (BTESE on Al₂O₃). A design of experiments was implemented to evaluate the influence of mixture composition on water content in the retentate, permeation flux, and water removal efficiency. Descriptive analysis revealed that the initial water content is the dominant factor governing both permeation flux and dehydration performance, whereas acid, alcohol, and ester have secondary but interactive effects. Reduced cubic polynomial models including linear, binary, and ternary interactions were developed, showing good agreement with experimental data. Ternary diagrams highlighted composition regions where molecular interactions significantly affect separation performance. Multi-response optimization based on desirability functions identified optimal operating conditions at high initial water content (1.146 wt.%), yielding a permeation flux of 0.144 kg·m^-2·h^-1, a final water content close to the industrial target (0.2 wt.%), and a water removal efficiency of 86%. Experimental validation confirmed the reliability of the predictive model. The results provide insights into composition-performance relationships and demonstrate the suitability of BTESE membranes for low-water-content esterification systems.