| Products, process and separation engineering design needs the phase equilibrium data of complex fluids. From the molecular structure, intermolecular forces and microaggregate state structure, building model to predict the thermodynamic properties and phase behavior of the complex fluid mixtures is a great challenge.Molecular property depends on the property, number of the elements and interaction between elements(such as chemical bonds). The mixture behavior depends on the molecular property and intermolecular forces. Based on the above-mentioned idea, a new activity coefficient prediction model was proposed from molecular structure and intermolecular forces. New elements and chemical bonds groups were defined by considering the structure of organic compounds. Based on 1085 binary systems vapor-liquid equilibrium data including alkanes, olefins, aromatics, alcohols, acids, esters ketones and halogenated hydrocarbons(14323 data points), the interaction energy parameters of 10 elements as well as 33 chemical bonds were regressed. The objective function, which was minimized by the parameter estimation program in software Matlab, was a sum of squared deviations between experimental and calculated activity coefficients. The optimization algorithm applied in the parameter estimation program was the quasi-Newton method.Vapor-liquid equilibrium of 87 groups of binary system that were not included in data base and 13 groups of the ternary system were predicted by the new model. The results were compared with existing group contribution methods, such as UNIFAC(2003), UNIFAC(Dortmund), UNIFAC(Lyngby) and ASOG(2011). For the binary systems, the average relative error of vapor composition and the temperature or pressure of the bubble point were 7.46%, 3.63% respectively. The results were better than that of UNIFAC(2003), UNIFAC(Lyngby) and ASOG(2011) and were slightly worse than the UNIFAC(Dortmund).For ternary system, average relative error of vapor composition for compound 1, compound 2 and the temperature or pressure of the bubble point were 10.91%, 6.95%,1.52% respectively. The results were slightly worse than the other group contribution model.Combining SRK with UNICAC, a new GE-EoS model called SRK-UNICAC was proposed. Interaction parameter table of UNICAC model was used. In order to verify the prediction accuracy and the applicable scope in pressure of SRK-UNICAC model, phase equilibrium were forecasted for 87 groups of binary system under depression, 12 groups of binary system under high pressure and 13 groups of the ternary system. A detailed comparison was made about predictions of SRK-UNICAC model and the other activity coefficient models. For the binary systems under depression, the results were better than that of UNIFAC(Lyngby), UNIFAC(2003) and ASOG(2011) and were slightly worse than the UNICAC and UNIFAC(Dortmund). For the binary systems under high pressure and the ternary systems, the results were better than the other group contribution models including UNICAC.The vapor-liquid equilibrium of 14 binary systems was predicted by UNICAC. The missing parameters in NRTL for simulation of the rectification Process in Propylene Oxide Plant were obtained. Rigorous simulation of the rectification Process in Propylene Oxide Plant was made with NRTL and UNIFAC by Aspen Plus. The results of the simulation were compared with design data. The results of NRTL were better than the results of UNIFAC.In summary, the accuracy for the vapor-liquid equilibrium prediction method based on elements and chemical bonds was good. Compared with the existing group contribution models, the new method which had advantages of fewer parameters and the group split more convenient, can provide quantitative estimate of the vapor-liquid equilibrium for the design of chemical process separation unit. The new method is easier, broader and more accurate. Therefore, the results of this research have important scientific significance and wide application prospects. |
PDF Full Text Request