Modeling And Optimization Of Xylene Adsorption Separation Process

P-xylene (PX) is an important organic chemical raw material and the head of polyester chain. The separation of PX from mixed xylenes is the most difficult step in the process of PX production, and the molecular sieve adsorption separation which based on simulated moving bed (SMB) is the mainstream technology of the production of PX. Considering the SMB mechanism is complicated, and the operation variables are many and strong coupling. There is a certain difficulty in its modeling and optimization. So, the scheme to establish the SMB mechanism model and optimize its operation has been a hot area of research recently. The work are summarized as follows:Firstly, Aspen Chromatography is used as a developmental platform. On the basis of the TMB modeling method and industry data. Then select the proper mass transfer model, the adsorption equilibrium model, and the model solving methods. The SMB mechanism model is eatablished. By comparing the calculated component concentration with the actual component concentration of each single bed, it is found that the modeling performs great in describing the industrial SMB. Furthermore, the influences of the switching time and the zone reflux ratio on SMB separation performance were analyzed. The results showed that switching time between110and112s lead to PX purity above99.7%; the reflux ratio of zone two has a significant influence on both PX purity and recovery; the reflux ratio of zone three has a greater influence on PX recovery but little influence on purity. What’s more, it can obtain the operation interval under different product quality requirements and feed composition.Secondly, on the basis of the established model, multi-objective teaching-learning-base optimization (MOTLBO) algorithm is used to solve two typical multi-objective optimization problems of SMB. Comparing with NSGA-Ⅱ, the MOTLBO algorithm has been verified to be more efficient. In addition, The influences of the extract flow rate, the raffinate flow rate and the switching time on the pareto optimal solutions were also analyzed. The optimization can facilitate the design and operation of SMB.Finally, considering the poor local search capability of MOTLBO and easy loss of population diversity caused by unitary teaching methods, and the operation optimization of SMB is a complex constrained multi-objective optimization problem. So, this paper proposes a hybrid MOTLBO with alpha constraints technique. This constrained technique can utilize the useful information carried by some infeasible solutions, and then increase the diversity of population. Meanwhile, considering the good local search ability of differential algorithm, the proposed method introduce the differential mutation operator to teacher phase, which improves the local search capability and increase the algorithm’s accuracy. Experimental results show that the proposed algorithm has a good convergence and distribution on handling with these several constrained multi-objective optimization problems.