高明杰,李开宇,崔哲,田文德

• 1:青岛科技大学化工学院
• 2:青岛科技大学环境与安全工程学院

摘要(Abstract)：

针对甲基叔丁基醚(MTBE)合成工艺中流程复杂、优化难度大等问题,文中提出了一种基于机器学习代理模型与多目标进化算法的工艺设计与优化方法。研究选取了3种典型工艺结构,即传统反应-精馏串联系统(TR-TCD)、反应精馏塔(RD)和隔壁反应精馏塔(RDWC),建立代理模型,并结合非支配排序遗传算法(NSGA-Ⅱ)实现多目标优化。优化目标包括年化总成本(TAC)、二氧化碳(CO_2)排放量和烟效率(EE)。结果表明,RDWC工艺在3种工艺中综合性能最佳,其TAC为4783.83×10~3 USD·a~(-1),CO_2排放量为38 370.65 kg·a~(-1),EE达到12.44%。相比之下,TR-TCD和RD工艺在经济性或能效方面均存在一定劣势。本研究提出的设计与优化方法不仅提高了工艺的经济性和环境友好性,也为复杂反应-分离耦合过程的能效优化与低碳化设计提供了新的研究思路。

关键词(KeyWords)： 甲基叔丁基醚;反应精馏;隔壁塔;代理模型;XGBoost;NSGA-Ⅱ;多目标优化

Abstract:

Keywords:

基金项目(Foundation): 国家自然科学基金(22578239);; 山东省自然科学基金(ZR2022QB020,ZR2021MB113)

作者(Author): 高明杰,李开宇,崔哲,田文德

参考文献(References)：

• [1]谭捷.我国甲基叔丁基醚合成技术研究进展[J].精细与专用化学品，2024,32(1):47-49.TAN J, Research progress on the synthesis technology of methyl tert-butyl ether in China[J]. Fine and Specialty Chemicals, 2024,32(1):47-49.
• [2]MURAVYOVA A E. Development of a system for control of synthesis of methyl tertiary butyl ether[J]. Journal of Physics:Conference Series,2020,1679(4):042090-042092.
• [3]DUAN Z, WANG J, WU C. Comparative study of MTBE production routes via reactive distillation and fixed-bed reactor+distillation[J]. Chemical Engineering Transactions,2020,82:541-546.
• [4]XIAO W, ZHANG Y, WANG X, et al. Parameters optimization of MTBE reactive distillation process with response surface methodology[J]. Chemical Engineering Transactions,2019,76:541-546.
• [5]BARRIENTOS A, FERNANDEZ B, MORANTE R, et al. Recent advances in reactive distillation[J]. Engineering Proceedings, 2023,56(1):99-102.
• [6]PAZMI??O-MAYORGA I, JOBSON M, KISS A. A systematic methodology for the synthesis of advanced reactive distillation technologies[J]. Industrial&Engineering Chemistry Research,2023,62(14):5907-5928.
• [7]王超，马延朋，张杨，等.基于Aspen Plus模拟优化高纯MTBE合成装置[J].山东化工，2023,52(19):186-189,198.WANG C, MA Y, ZHANG Y, et al. Simulation and optimization of high-purity MTBE synthesis unit based on Aspen Plus[J].Shandong Chemical Industry,2023,52(19):186-189,198.
• [8]GONCALVES G, BATCHVAROV A, LIU Y, et al. Data-driven surrogate modelling and benchmarking for process equipment[EB/OL]. arXiv:2003.07701,2020.
• [9]STEVANOVIC S,DASHTI H,MILOSEVIC M,et al. Comparison of ANN and XGBoost surrogate models trained on small numbers of building energy simulations[J]. PLoS ONE,2024,19(10):e0312573.
• [10]CHENG L,ZHANG W,FENG Z. Evaluation of CO2 emissions from chemical processes based on utility consumption:A case study of MTBE[J]. Journal of Cleaner Production,2023,398:136720.
• [11]FENG Z,ZHANG J,LENG J,et al. Enhancing economic,exergy efficiency and environment of extractive distillation processes via integrating two-enthalpy-feed and vapor recompression heat pump[J]. Separation and Purification Technology, 2025, 360(P1):130950.
• [12]SHAO Y, BAI J, LI K, et al. Surrogate-model-assisted multi-objective optimization of dividing wall columns in Aspen Plus[J].Separation and Purification Technology,2022,288:120600.
• [13]QIAN Y,VADDIRAJU S,KHAN F. Inherent Process Risk Index(IPRI)–A tool for analyzing inherently safer design using Aspen Plus simulation[J]. Process Safety and Environmental Protection,2024,183:399-416.
• [14]DEB K,PRATAP A,AGARWAL S,et al. A fast and elitist multi-objective genetic algorithm:NSGA-II[J]. IEEE Transactions on Evolutionary Computation,2002,6(2):182-197.
• [15]XIAO W, ZHANG Y, JIANG X, et al. Multi-objective optimization of MTBE reactive distillation process parameters based on NSGA-II[J]. Chemical Engineering Transactions,2018,70:1621-1626.
• [16]SONG C, LENG J, DONG L, et al. Design and optimization of reactive dividing wall distillation column process for methyl methacrylate production[J]. Separation and Purification Technology,2025,360(P2):130992.
• [17]WHITE E,WITING F,WITTEKIND I,et al. Distilling the Pareto optimal front into actionable insights[J]. Environmental Modelling and Software,2025,191:106508.
• [18]HE R, JU K, LI L, et al. Multi-objective collaborative optimization of distillation column group based on system identification[J].Processes,2023,11(2):436.