Abstract: In the industrial wastewater treatment sector of chemical engineering, the separation of multicomponent azeotropic mixtures remains a persistent challenge. This complexity arises from the intricate phase equilibrium behavior of these systems, which involves minimum/maximum boiling azeotropes and liquid-liquid phase separation. Pharmaceutical wastewater, often contains a ternary mixture of acetonitrile, n-propanol, and water. This mixture exhibits significant non-ideality and multiple azeotropic points, including binary azeotropes for acetonitrile-water, n-propanol-water, and acetonitrile-n-propanol pairs, as well as a ternary azeotrope. Thus, developing an energy-efficient separation process is essential. To address this issue, this study conducted a comprehensive investigation encompassing thermodynamic modeling, solvent screening, process design, and multi-objective optimization. A reliable thermodynamic framework was established using an activity coefficient model, and its validity was confirmed through experimental data verification, thereby ensuring the reliability of subsequent process simulations. Based on systematic analysis of vapor-liquid equilibrium diagrams, ethylene glycol was finally identified as the optimal extractant due to its optimal selectivity. Three distinct separation processes were developed: a conventional three-column distillation sequence, a four-column pre-concentration extractive distillation configuration, and an innovative three-column integrated pre-concentration extractive distillation system incorporating a thermally coupled column. Quantitative process evaluation was performed through multi-objective optimization using the improved nondominated sorting genetic algorithm (NSGA-II), with optimization targets comprising total annualized cost (TAC), CO2 emissions (E_(CO_2 )), and thermodynamic efficiency (η), subject to stringent purity constraints (≥99.9wt% for product components and ≥99.99wt% for extractant recycle). Optimization results showed the superior performance of the integrated three-column configuration, achieving 41.2% lower in TAC, 50.4% fewer CO2 emissions, and 102% higher thermodynamic efficiency than conventional approaches. This integrated preconcentration-extractive distillation process is established as an industrially viable, energy-efficient solution for acetonitrile-n-propanol-water separation that aligns with green chemistry principles.

Key words: ternary multi-azeotropic mixtures, separation, pre-concentration, multi-objective, genetic algorithm

摘要： 工业废水中常存在含多种共沸物的混合物，其相平衡的复杂性给分离带来了挑战。本工作对医药行业生产废水中的乙腈、正丙醇、水混合物进行分离研究。筛选出乙二醇为萃取剂，基于三元相图和剩余曲线设计了分离序列，进而搭建常规三塔精馏工艺、四塔预浓缩萃取精馏工艺和带有集成蒸馏塔的三塔预浓缩萃取精馏工艺。借助改进非支配排序遗传算法(NSGA-II)，以年总费用(TAC)、二氧化碳排放量(E_(CO_2 ))及热力学效率(η)为目标，对各工艺进行多目标优化，要求各组分质量分数达到99.9wt%，循环萃取剂质量纯度为99.99%。选择TAC最小的一组Pareto前沿解为最优解，与常规三塔流程相比，带有集成蒸馏塔的三塔预浓缩萃取精馏工艺TAC降低了41.2%，二氧化碳排放量降低了50.4%，热力学效率提升约102%，表明带有集成的预浓缩萃取精馏工艺是分离乙腈-正丙醇-水体系的高效节能、降本增效工艺。

关键词: 三元多共沸混合物, 分离, 预浓缩, 多目标, 遗传算法