【作者】 宋静；

【导师】 宋海华；

【作者基本信息】 天津大学 ， 化工过程机械， 2008， 博士

【摘要】 化工生产中常遇到欲分离组分之间的相对挥发度接近1或形成共沸物的难分离物系,萃取精馏和液液萃取是常用且有效的方法,其中萃取剂的选择是关键。随着全球性污染问题的日益严峻,选择环境友好的萃取剂是今后溶剂选择发展的大趋势。本文提出了一种新的基于模拟退火算法和模糊综合评价的多目标优化方法,用于萃取精馏和液液萃取分离过程环境友好的溶剂的计算机辅助分子设计。首先针对萃取剂的评价具有多指标性和模糊性的特点,将模糊综合评价的方法用于萃取精馏溶剂和液液萃取萃取剂的评价,基于萃取剂的分离性能和环境性能,建立了的模糊综合评价隶属函数。其次,提出了改进的模拟退火算法,相对与传统的模拟退火算法,增加了记忆功能,能将搜索过程中最优的解保存下来,将所得的最终解与记忆器中的解比较取较优的作为最后结果,提高了算法的解的质量;使用加权后的综合目标函数进行多目标优化的考核,多次运行或并行该算法可得到Pareto边界上不同方向的Pareto最优解。再次,基于Mod.UNIFAC基团贡献法,对基团进行了预选及分类,有效克服了传统分子设计方法的“组合爆炸”,提出了实数编码的基团编码策略和二行矩阵的分子表达方法,第一行表示分子中包含的基团的种类,第二行表示分子中包含的这些基团的数目。将提出的多目标优化方法用于萃取精馏和液液萃取过程萃取剂的计算机辅助分子设计,生成策略包括引入基团、删除基团、替换基团、修复操作四种,避免了垃圾分子的生成,确保产生的分子满足结构可行性及化学可行性准则。最后将提出了方法用于实际分离物系的萃取剂选择和设计,对甲苯-甲基环己烷、甲醇-丙酮、乙醇-乙酸乙酯、甲醇-乙酸甲酯四个萃取精馏体系进行了环境友好溶剂的计算机辅助分子设计,编制了相应的Matlab源程序,得到一组满足分离要求和环境要求的溶剂,并针对甲醇-丙酮体系,采用PROII软件对设计的溶剂进行了模拟验证及比较;对醋酸-水、糠醛-水、正丁醇-水、和苯酚-水四个液液萃取体系进行了环境友好溶剂的计算机辅助分子设计,编制了相应的Matlab源程序,得到一组满足分离要求和环境要求的溶剂,并针对苯酚-水体系,选择了四种设计的溶剂采用PROII软件进行了模拟验证,证明了设计结果的准确性,且模拟结果表明本文设计的溶剂性能更优。由此表明采用本文提出的结合模拟退火算法和模糊综合评价的多目标优化方法进行计算机辅助分子设计具有较高的准确性和很好的实用性,选择的溶剂不仅满足分离过程的要求,且考虑到了设计的溶剂对环境的影响,为工业生产提供了更优的选择。更多还原

【Abstract】 Extractive distillation and liquid-liquid extraction are effective separation processes to deal with the separation of close boiling compounds or azeotropes systems. The selection of appropriate solvents is the key. As global environment problem increasing severely, the selection of environmental friendly solvents is the trend. A new multi-objective optimization approch, based on simulated annealing algorithms and fuzzy comprehensive evaluation method, is presented and applied to the computer-aided molecular design of environmentally benign solvents for separation processes.Firstly, based on the characteristic of fuzziness and involvement of multiple factors in selection of solvents, the fuzzy comprehensive evaluation method is used to evaluate the design solvents, the membership functions are established according to the separation and environmental performance.Secondly, the modified simulated annealing approach for multi-objective optimization problem is proposed,which can record the optimal solution in the process of annealing, improving the quality of solution; evaluating the solution by the synthetical value of target functions, the algorithm can obtain a set of Pareto optimum by running the algorithm multiple times.Based on Modfied UNIFAC method, classifying and pre-selection the functional groups systematically are described according to some rules. The functional groups are coded according to the real-coded technique. The solvent molecules are represented by a matrix, including a group vector and a composition vector.The multi-objective optimization model of CAMD for separation process is formulated, and the multi-objective optimization approach combined simulated annealing approach and fuzzy comprehensive evaluation is applied to CAMD. The steps and flowsheet of multi-objective optimization approach are addressed, the parmeters are suitably designed. The new molecules can be generated by addition, deletion, replacement and restoration, which can ensure the molecular structures satisfy the construction and chemical feasible criteria.Finally, the CAMD multi-objective optimization approach is illustrated with four extractive distillation examples and four liquid-liquid extraction examples, the program is programmed using Matlab language, the design results are examined by the process simulation software PROII, confirm the validity of the proposed multi-objective optimization method. The solvents not only satisfy the requirement of the separation process, furthermore, but satisfy the requirement of environment.更多还原