Prediction of the Density of Two-Phase Hydrocarbon Systems Particularly Near the Critical Region

Abstract

Abstract Molecular refraction is introduced as a new and improved third parameter for prediction of the PVT behavior of hydrocarbon systems. This parameter, characterizing the complex as well as the pure hydrocarbon systems is a direct measure of London dispersion forces which affect the PVT behavior; can be measured directly, accurately and easily for complex as well as pure hydrocarbon systems; varies significantly between the reference substances thus allowing accurate interpolation of PVT properties; and has a fundamentally sound basis. Molal average molecular refraction was used with the pseudocritical mixing rule proposed by Leland and Mueller to predict the molal volume of liquid and vapor for 84 published data points on nC4-nC10, C1-nC5, C1-nC4-nC10 systems in the region of coexisting vapor-liquid and at the critical point. The same technique was applied to predict the densities of liquid hydrocarbon mixtures in the two-phase and single-phase regions near the critical point for seven different systems each having different properties of the heptanes-plus fraction, but all simulating the fluids of either gas condensate or volatile oil reservoirs. A modification of the Leland-Mueller mixing rule for the determination of pseudocritical properties is also presented which yields more accurate liquid density predictions than the original form when used wish the proposed third parameter. Introduction The PVT behavior of multicomponent systems has received widespread attention in the literature in recent years. A large portion of the reported work has stemmed from attempts to overcome the inherent difficulties encountered when using the standard compressibility-factor chart which shows z as a function of reduced temperature Tr and reduced pressure Pr. This chart assumes that zc is constant for all systems. The error in this assumption is not pronounced for pipeline-type gases high in methane and low in heavy ends, particularly when contained at moderate pressures. For such gases the error in the calculated z is usually less than 2 to 3 per cent. However, the reliability of this approach becomes somewhat nebulous for gas condensate and volatile oil systems at high pressure. One logical approach has been to recognize the variation of Zc. This idea was first suggested by Meissner and Seferian and used by Lyderson, Greenkorn and Hougen to prepare extensive tables of thermodynamic properties. Other third parameters have been proposed by Stockmeyer, Kihara, Riedel, Lightfoot, Bloomer and Pitzer. These were tested by Satter and Campbell for many gas mixtures and it was shown that they were inter-related. All of these third parameters have a common fault - they cannot be directly evaluated for the heptanes-plus fraction. Consequently, one necessary requisite of an improved method for predicting the density of multicomponent hydrocarbon systems, was development of a third parameter that could be evaluated for that fraction. PREVIOUS MULTICOMPONENT DENSITY PREDICTION METHODS Sage and Lacey suggested a method of using partial molal volumes for computing the density of hydrocarbon gas mixture for pressures to 3,000 psi. In this technique, they approximate the complex multicomponent gaseous mixtures by a quarternary system of methane, ethane, propane and butane. Sage, Hicks and Lacey suggest a method of using partial molal volumes for computing the density of hydrocarbon liquids. The results agree within 3 per cent of the experimental values. However, the composition range is limited to about 10 per cent by weight of methane. Consequently, this correlation does not cover the low-molecular-weight liquid similar to natural gasoline and high-pressure gas condensates. SPEJ P. 78ˆ