Health & Environmental Research Online (HERO)


Print Feedback Export to File
6943933 
Journal Article 
Olefin separation from FCC naphtha by dimethyl sulfoxide: Effect of structure on the separation and correlation of thermodynamic models 
Zhang, Y; Wang, Hui; Zhao, L; Gao, J; Xu, C; , 
2019 
1 
Separation and Purification Technology
ISSN: 1383-5866 
ELSEVIER 
AMSTERDAM 
228 
English 
WOS:000478709000008 
The performance of dimethyl sulfoxide (DMSO) solvent in separating olefins from aromatics and thiophene sulfides was evaluated through a series of ternary liquid-liquid equilibrium (LLE) experiments at 101.325 kPa and 303.15 K. The influences of the structures of olefins, aromatics, and thiophene sulfides on the separation efficiency were analyzed. Normal 1-hexene and isomer 2,4,4-trimethyl-1-pentene were easily separated from aromatics or thiophene sulfides, which only contain aromatic ring structures, such as benzene and thiophene. The reason is that the induced dipole on the aromatic ring structure introduced by the DMSO solvent significantly improved the polarity of aromatics and thiophene sulfide molecules, thereby enhancing the interaction with the ring structure. In addition, olefins were easier to separate from thiophene sulfides than from aromatics. The olefin separation factor of separating 2,4,4-trimethyl-1-pentene from thiophene was three times greater than that of separating the same olefin from benzene. The binary parameters between DMSO and olefins, aromatics, and thiophene sulfides for NRTL and UNIQUAC models were obtained through ternary LLE experiments. The maximum value of the root-mean-square deviation of the simulated compositions of phases compared with experimental data was 0.0052 in the two adopted models. Hence, the regressed binary parameters could efficiently describe the experimental LLE process. On the basis of the obtained binary parameters, the NRTL and UNIQUAC models are suitable for predicting the performance of olefin separation from fluid catalytic cracking naphtha. However, the NRTL model is more precise than the UNIQUAC model.