Characterizing Hydrocarbon Plus Fractions

Abstract

Whitson, Curtis H., SPE, U. of Trondheim Abstract Methods are developed for characterizing the molar distribution (mole fraction/molecular weight relation) and physical properties of petroleum fractions such as heptanes-plus (C7 +). These methods should enhance equation-of-state (EOS) predictions when experimental data are lacking. predictions when experimental data are lacking. The three-parameter gamma probability function is used to characterize the molar distribution, as well as to fit experimental weight and molar distributions and to generate synthetic distributions of heptanes-plus fractions. Equations are provided for calculating physical properties such as critical pressure and temperature properties such as critical pressure and temperature of single-carbon-number (SCN) groups. A simple three-parameter equation is also presented for calculating the Watson characterization factor from molecular weight and specific gravity. Finally, a regrouping scheme is developed to reduce extended analyses to only a few multiple-carbon-number (MCN) groups. Two sets of mixing rules are considered, giving essentially the same results when used with the proposed regrouping procedure. Introduction During the development of the application of EOS's to naturally occurring hydrocarbon mixtures, it has become clear that insufficient description of heavier hydrocarbons (e.g., heptanes and heavier) reduces the accuracy of PVT predictions. Volatile oil and gas-condensate volumetric phase behavior is particularly sensitive to composition and properties of the heaviest components. properties of the heaviest components. Until recently there has not been published in technical journals a comprehensive method for characterizing compositional variation, which we call "molar distribution." Several authors have given lucid descriptions of petroleum fraction characterization, though they deal mainly with physical property estimation. Usually, only physical property estimation. Usually, only a single heptanes-plus (C7 + ) fraction lumps together thousands of compounds with a carbon number higher than six. Molecular weight and specific gravity (or density) of the C7 + fraction may be the only measured data available. Preferably, a complete true-boiling-point (TBP) analysis should be performed on fluids to be matched by an EOS. Distillation experiments yield boiling points, specific gravities, and molecular weights, from which molar distribution is found directly. Special analyses of TBP data can also provide estimates of the paraffin/napthene/ aromatic (PNA) content of SCN groups, which are useful in some property correlations. Unfortunately, such high-quality data are seldom available for fluids being matched or predicted by an EOS. If data other than lumped C7+ properties are available, they might include a partial component analysis (weight distribution) from chromatographic measurements. In this case. only weight fractions of SCN groups are reported; normal boiling points, specific gravities, and molecular weights (needed to convert to a molar basis) simply are not available. Compositional simulation based on an EOS involves two major problems:how to "split" a C7 + fraction into SCN groups with mole fractions. molecular weights, and specific gravities that match measured C7+ properties, andif a partial extended analysis (e.g., C 11 + ) is available, how to extend it to higher carbon numbers. The first step in addressing these problems is to find a versatile, easy-to-use probability function for describing molar distribution. The distribution function should allow consistent matching and reasonable extension of partial analyses. Also, it should not contain too many unknown or difficult-to-determine parameters. This paper presents such a probabilistic model and describes its application to several reservoir fluids under "Molar Distribution."The second step in characterizing plus fractions involves estimating SCN group specific gravities, which, together with estimated molecular weights (from the probabilistic model), could be used to estimate critical properties required by EOS's. We address this problem and suggest a simple method for specific gravity estimation under "Physical Properties Estimation." SPEJ p. 683