Correlation for fitting multicomponent vapor-liquid equilibria data and prediction of azeotropic behavior

Abstract

Correlation equations for expressing the boiling temperature as direct function of liquid composition have been tested successfully and applied for predicting azeotropic behavior of multicomponent mixtures and the kind of azeotrope (minimum, maximum and saddle type) using modified correlation of Gibbs-Konovalov theorem. Also, the binary and ternary azeotropic point have been detected experimentally using graphical determination on the basis of experimental binary and ternary vapor-liquid equilibrium data. In this study, isobaric vapor-liquid equilibrium for two ternary systems: “1-Propanol – Hexane – Benzene” and its binaries “1-Propanol – Hexane, Hexane – Benzene and 1-Propanol – Benzene” and the other ternary system is “Toluene – Cyclohexane – iso-Octane (2,2,4-Trimethyl-Pentane)” and its binaries “Toluene – Cyclohexane, Cyclohexane – iso-Octane and Toluene – iso-Octane” have been measured at 101.325 KPa. The measurements were made in recirculating equilibrium still with circulation of both the vapor and liquid phases. The ternary system “1-Propanol – Hexane – Benzene” which contains polar compound (1-Propanol) and the two binary systems “1-Propanol – Hexane and 1-Propanol – Benzene” form a minimum azeotrope, the other ternary system and the other binary systems do not form azeotrope. All the data passed successfully the test for thermodynamic consistency using McDermott-Ellis test method (McDermott and Ellis, 1965). The maximum likelihood principle is developed for the determination of correlations parameters from binary and ternary vapor-liquid experimental data which provides a mathematical and computational guarantee of global optimality in parameters estimation for the case where all the measured variables are subject to errors and the non ideality of both vapor and liquid phases for the experimental data for the ternary and binary systems have been accounted. The agreement between prediction and experimental data is good. The exact value should be determined experimentally by exploring the concentration region indicated by the computed values.