No abstract

Phase equilibrium calculations based on 150 n-alkane + aromatic and n-alkane + naphthenic hydrocarbon binary mixtures were performed. These calculations were compared with experimental measurements whenever possible, and additional measurements were made as part of this work. The widely used Peng–Robinson (PR) and Soave–Redlich–Kwong (SRK) equations of state are shown to predict nonphysical liquid–liquid phase behavior for long-chain n-alkane + aromatic and long-chain n-alkane + naphthenic hydrocarbon binary mixtures with standard pure-component parameters (T c, P c, ω). Incorrect phase behavior prediction is shown to be insensitive to the selection of correlations for estimating pure-component properties for n-alkanes that are not available from experimental data. For cubic equations of state, correct phase behaviors are obtained only when negative values of the binary interaction parameters (k ij ) are used. For PC-SAFT, a noncubic equation of state (with standard parameter values defining molecules and with binary interaction parameters set to zero), phase behaviors that are consistent with observed phase behaviors are obtained. However, below the melting temperature of at least one of the components, liquid–liquid phase behavior is predicted for some binary mixtures. The quality of liquid/vapor phase composition and dew and bubble pressure predictions from the cubic and PC-SAFT models was not evaluated. Measurement and phase behavior modeling outcomes are discussed.

Experimental vapor−liquid equilibrium (VLE) data for long-chain n-alkane + naphthenic mixtures are scarce in the open literature. In this study, VLE data for representative binary mixtures of naphthenes with long-chain n-alkanes are presented. The selected compounds include two n-alkanes (n-hexadecane and n-eicosane) and three naphthenes (cyclohexane, methylcyclohexane, and ethylcyclohexane). The experimental data are compared with computed bubble pressures using the Peng−Robinson and PC-SAFT equations of state in order to evaluate the accuracy of predictions and to obtain regressed k ij values. Expected applications of these contributions include improved phase behavior model accuracy for hydrocarbon production, transport, and refining applications.

No abstract