"A New Approach For Characterizing Oil Fractions, And For Selecting Pseudo-Components Of Hydrocarbons"

Abstract

E. GONZALES P. COLONOMOS I. RUSINEKthis article begins on the next page F F JCPT86-02-07 SIMULATION A new approach for characterizing oil fractions, and for selecting pseudo-components of hydrocarbons E.GONZALEZ, P. COLONOMOS; and I. RUSINEK INTEVEP S.A. ABSTRACT The non-uniqueness of the inversion of physicochemical parameters to obtain molecular weight distributions for heavy oil fractions is discussed. A method to obtain bounds for th-e cumulatives of these distributions from the usually available experimental information is presented. The method can be applied systematically using linear programming techniques, and the bounds obtained are mathematically rigorous due to the properties of Tschebychev systems. The method is also used to calculate bounds for the pseudo-critical properties needed to characterize pseudo-components for equation-of-state use. These bounds provide a way to select optimal lumping schemes (for a given number of pseudo-components) in a systematic manner. Numerical examples for the bounds obtained and for the optimized lumping procedure are presented. Introduction Equations-of-state are used in reservoir engineering to predict the PVT behaviour of crudes when complete experimental measurements are not available and in compositional numerical simulation.Generally, with a sufficiently large number of pseudo-components, a satisfactory description of the heavy fraction of the fluid of interest can be obtained. However, there are strict limitations on the maximum number of components that can be used in numerical simulation work and the original composents have to be lumped into a smaller number of new pseudo-components. One way of determining a characterization for a heavy frac-tion is to take each distillation cut or identified Cn group on a partially extended analysis as a pseudo-component, determining its pseudo-critical parameter via correlations and completing the match by fitting the known experimental data using the binary interaction coefficients as adjustable parameters. Whitson(l) has presented a review of the many correlations available for this purpose.An alternate approach proposed by Whitson(2) assumes a given functional form with adjustable parameters for the dis-tribution of C. groups (Cn molecular weights) in the heavy frac-I . The.-adjustable parameters are used to fit the known average properties for the heavy fraction and for the distillation cuts or identified C. groups. The adjusted distribution together with a set of generalized properties for C. groups up to C,, com-pletes the characterization. The problem of lumping these pseudo-components into a smaller number without losing the predicting power of the equa-tion of state has been addressed by many authors(2-5). This prob-lem has several stages: into how many components should one lump'?, which components should be lumped together?, and how, does one determine the EOS constants for the new lumped pseudo-components? In this paper we address the latter two points-Lee e. t 6!/.(3) have proposed a method in which thermody...