Synthesis Of Separation Process For Multicomponent Products Via Genetic Programming

Separation process that deals with multicomponent products produced from mixture materials through various operations is called nonsharp separation and used widely in the chemical industry and the petroleum industry. Nonsharp separation process synthesis is always the focus in the process synthesis domain. The amount of the possible separation plans, however, will increase exponentially due to the strong interrelation between the splitter, mixer and the distillation column in the flowsheet which makes the problem being extremely combinatorial complex. The current research work for this kind of problem is to the decision of the simple separation plan, but the importance of this kind of problem lets us study the synthesis method and strategy in depth from the flowsheet level. This paper addresses a sort of united synthesis method that based on Genetic Programming(GP)and complex method. The proposed method regards minimizing the annually total cost as optimization objective and seeks the heat integration nonsharp separation optimal flowsheet to give the major technological parameters of the important equipments which provide the accurate theoretical basis for actual production. The main content of this paper is as follows:1. The united synthesis method that based on GP and complex method to solve the sharp separation problem for pure component products is firstly proposed in order to probe into the theoretical basis solving the nonsharp separation synthesis problem using GP. The algorithm, under the guide of chemical engineering separation professional requirements, utilizes GP to solve the sharp separation process structural variables that expressed by the complex distillation column configurations combing with the heat integration method. The encoding model and evolutionary scheme that applicable to this problem are proposed and the heat integration strategy that matches system condenser and reboiler randomly is built to produce complex match scheme between various distillation structures. At the same time, the algorithm adopts complex method to optimize the entire continuous variable in the separation system which improves the computational efficiency obviously.2. The united synthesis method that based on GP and complex method to solve the nonsharp separation problem is proposed based on the sharp separation process research. The chemical engineering node model is proposed to denote distillation separation, splitting and mixing task as distillation column node, splitter node and mixer node separately. Both the distillation column node and splitter node generate left and right branch to denote separation and splitting operation for all streams and the corresponding node property is set to denote their operation specialty. As for mixer node, unique multilayer mixer grid structure that made up of many mixers is established to denote arbitrary mixing operation. To clearly denote the location of every mixer in the grid structure, a pair of property parameters is set in the mixer node to record the number of the row and column where the mixer located in separately.3. Multilayer cross tree encoding structure is established in this paper to accurately express the GP code that produced from the above three kind of nodes to denote nonsharp separation flowsheet. A GP sub-tree is randomly produced from every mixer as sub-root node and it is made up of distillation column nodes and splitter nodes that randomly connected each other. The branch of distillation column node or splitter node of every sub-tree can joint to any mixer of the inferior mixer grid at random in this paper. Many GP sub-trees at different layer may have stream cross and the streams can pass through all layers of the mixer grid to arrive at terminal product mixer directly. So, the overall GP code tree is made up of layer upon layer sub-trees and forms complex multi-cross tree structure to describe all sorts of complex jointed flowsheet possibly produced between splitter, mixer and distillation column in nonsharp separation flowsheet. This enable the algorithm to discovery better flowsheet structure that is difficult found by other algorithms. Besides, modified GP evolutionary strategy that is applicable for nonsharp separation problem is proposed in this paper. Different operable node classes are defined in various operation steps, especially for cross and mutation operation, to insure the professional validity of all GP codes, which improves the evolutionary efficiency evidently.4. Synthesis strategy that adopts simultaneously optimization of structural variables and continuous variables is proposed in this paper. Complex method is used to optimize system continuous variables when using the GP to search the optimal flowsheet structure. As for the distribution ratio of the splitter, light and heavy key component recovery ratio of nonsharp distillation column, complex method is used to optimize them thoroughly to ensure generating valid codes that suitble to the required mixture products. As for the operation pressure and reflux ratio of the distillation column, complex method is used to search the optimum within the limited times. The above optimization method that using different strategy for various types of continuous variables not only accelerates the process of structural variable optimization effectively but also improves the computational efficiency.5. To the various objective functions, the united synthesis method that based on GP and complex method to solve the nonsharp separation problem is used to study the problem from different level. A kind of simple linear function is used as the objective firstly and the better solutions are found through optimization of the problems at the stream decision flowsheet level which demonstrate the validity of the algorithm.In this paper, annually total cost, for the first time, is used as optimization objective and the equipment design and cost model of nonsharp separation system is given. Heat integration is proposed simultaneously to generate complex heat exchange scheme that can express the condenser and reboiler of one distillation column matching with some condensers and reboilers of other distillation columns simultaneously. Typical nonsharp separation problems that include ester, hydrocarbon and alcohol mixtures, for the first time, are solved using above strategy. Bigger size of example, as much as 11 components, is solved comparing with the reported 5 components example. The computational results demonstrate that, based on the synthesis algorithm proposed in this paper, the optimal structural decision flowsheet can be obtained by using nonsharp distillation column and heat integration in the flowsheet. The proposed algorithm features simple to use, small amount of calculation and fast and accurate. The superstructure of separation system is not required in solving process and the solution bound is far and wide that shows the united synthesis algorithm proposed in this paper is an effective method for nonsharp separation system synthesis.