基于混整空间粒子群算法的精馏塔优化设计

发布时间：2018-01-27 13:09

本文关键词： 混合整数非线性规划粒子群算法共沸精馏优化设计流体力学　出处：《青岛科技大学》2015年硕士论文　论文类型：学位论文

【摘要】：精馏塔是化工与石化行业中最重要的分离单元设备之一。在化工企业中,约90%的分离与提纯操作为精馏操作,约占分离总能耗的90%。因此,对精馏塔进行优化设计,减少设备投资与能耗,具有重大的经济效益。由于精馏塔优化设计的变量既包括了塔板数、进料位置等整数变量,又包括了回流比等实数变量,因此精馏塔优化是一个混合整数非线性规划( Mixed integer nonlinear programming, MINLP)问题。本文研究了MINLP问题求解的各算法的优、缺点之后,最终选择粒子群优化(Particle swarm optimization, PSO)算法进行精馏塔的优化设计。为提高PSO算法求解MINLP的效率,本文研究和分析了现有的PSO算法整数编码和约束处理技术之后,提出了新的整数变量处理策略和流体力学约束条件,并以Deb方法处理约束,提出了混合整数空间粒子群优化(Mixed integer space particle swarm optimization, MIS-PSO)算法。采用C#编程语言开发了精馏塔MINLP界面平台,采用C++编程语言开发了MIS-PSO算法的动态链接文件,在MINLP平台中调用MIS-PSO算法进行粒子位置的更新。通过ActiveX控件调用Aspen Plus,将Aspen Plus中建立的精馏模型作为适应度函数,计算粒子所对应的塔结构和操作参数而获得粒子的适应度函数值,充分利用了Aspen的强大计算能力和成熟的精馏计算模块,并使得寻优过程更易收敛。之后将MIS-PSO算法用于一个共沸精馏塔的设计,MIS-PS O算法计算得出的最优年度总费用(Total annual cost, TAC)为7.08×105$/year,该结果不仅满足所有约束,而且比Aspen Consep的计算结果降低了19.64%,并且优于文献结果。由Aspen Consep所设计的塔不仅TAC最优值不及MIS-PSO法得出的最优值,而且空塔气速低于0.5m/s的下限,说明MIS-PSO算法可以更好地应用于精馏塔的优化设计。为考察多塔优化的有效性,将MIS-PSO法用于一个异丙醇脱水过程的共沸精馏双塔序列。计算结果表明,TAC 1.218×106$/year比文献值减少16.96%,符合流体力学约束条件。文献值经过计算后TAC高于MIS-PSO法,并且C-1塔的空塔气速0.851 m/s大于0.8m/s的上限,不满足流体力学约束。该计算结果进一步说明MIS-PSO法计算多塔得出的结果也优于文献结果。上述工作表明,MIS-PSO法直接在混合整数空间进行寻优,为PSO算法求解混合整数规划问题提供了新的思路。MIS-PSO法可以用于精馏序列的最优设计,并且所得的结果优于以往的文献值。未来可将本文成果应用于更为复杂的精馏塔设计优化,以满足大规模化工流程优化的需要。
[Abstract]:Distillation column is one of the most important separation units in chemical and petrochemical industries. In chemical enterprises, about 90% of separation and purification operations are distillation operations, accounting for about 90 percent of the total separation energy consumption. The optimization design of distillation column to reduce equipment investment and energy consumption has great economic benefits. Because the variables of optimization design of distillation column include integral variables such as tray number feed position and so on. Reflux ratio and other real variables are also included, so the distillation column optimization is a Mixed integer nonlinear programming. This paper studies the advantages and disadvantages of the algorithms for solving the MINLP problem. Particle Swarm Optimization (PSO) is used to optimize particle swarm optimization. In order to improve the efficiency of PSO algorithm to solve MINLP, this paper studies and analyzes the existing PSO algorithm integer coding and constraint processing technology. A new integer variable processing strategy and fluid mechanics constraint conditions are proposed, and Deb method is used to deal with the constraints. Mixed integer space particle swarm optimization is proposed for mixed integer space particle swarm optimization. MIS-PSO algorithm. The MINLP interface platform of distillation column is developed by C # programming language, and the dynamic link file of MIS-PSO algorithm is developed by C programming language. Call the MIS-PSO algorithm in the MINLP platform to update the position of the particle. Call Aspen Plus through the ActiveX control. The distillation model established in Aspen Plus is taken as fitness function, and the corresponding tower structure and operation parameters of particles are calculated to obtain the fitness function of particles. The powerful computing power of Aspen and the mature distillation calculation module are fully utilized, and the optimization process is more easily converged. Then, the MIS-PSO algorithm is used in the design of an azeotropic distillation column. Total annual cost (TAC) calculated by MIS-PS O algorithm is 7.08 脳 10 ~ 5 / year. The results not only satisfy all the constraints, but also decrease by 19.64% compared with the results of Aspen Consep. The column designed by Aspen Consep is not only less than the optimal value obtained by MIS-PSO method, but also lower than the lower limit of 0.5 m / s. It shows that MIS-PSO algorithm can be applied to the optimization design of distillation column better. The MIS-PSO method was applied to an azeotropic distillation two-column sequence in the dehydration process of isopropanol. The calculated results show that the value of TAC 1.218 脳 106 / year is 16.96% less than that in the literature. The TAC value is higher than that of MIS-PSO method, and the upper limit of the empty tower gas velocity of C-1 tower is 0.851 m / s > 0.8 m / s. The results show that the MIS-PSO method is also superior to the results of the literature. The MIS-PSO method is directly applied to the optimization of mixed integer space, which provides a new idea for the PSO algorithm to solve the mixed integer programming problem. MIS-PSO method can be used in the optimal design of distillation sequences. The results obtained are better than the previous ones. In the future, the results of this paper can be applied to the design optimization of more complex distillation column to meet the needs of large-scale chemical process optimization.
【学位授予单位】：青岛科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TQ053.5;TP18

【相似文献】