Compositional Uncertainties in Laboratory PVT Data

Abstract

Abstract Accuracy of phase behavior and volumetric calculations from a cubic equation of state (EOS) depends on the accuracy of the molar compositions used as input to the model. Lab-reported compositions have uncertainty, like all other measured PVT data. This paper discusses different sources of uncertainty in lab-reported compositions, the magnitude of uncertainty, and we propose methods to correct for uncertainty that improve PVT calculations of individual samples. Lab-reported molar compositions can have uncertainty due to (a) baseline shift and (b) internal standard used in gas chromatography, (c) component molecular weights used to convert measured mass fractions to mole fractions, and (d) the gas-oil molar ratio (i.e., gas-oil ratio) used in recombination. A molar distribution model is used to assess and quantify uncertainty in chromatographic measurements of heptanes and heavier (C7+) fractions, also providing a method to correct for possible errors. As a theoretical basis, synthetic examples are used to demonstrate the application of the gamma molar distribution model to quantify and correct compositional uncertainty in C7+ mass fractions due to baseline shift and internal standard. The workflow includes use of a distribution model that describes more than 50 PVT samples with widely varying gas-oil ratios and API densities, all from the same basin / field, and analyzed by several PVT laboratories over an entire decade. Examples show that a common distribution model reliably corrects for compositional uncertainty from baseline shift and internal standard errors. The model also provides consistent and representative estimates of C7+ component molecular weights that are used to convert masses to moles. The same model provides consistent sample-specific average C7+ molecular weights that are used in correlating property variations across the basin. Most engineers use the lab-reported molar composition "as is" from a PVT report, often directly as input to an EOS model. We show quantitatively the four reasons why a lab composition may have systematic error. We also provide methods to quality check and correct lab-reported compositions. A molar distribution model is used to model heavier (C7+) components quantified by gas chromatography, where the model can be used to identify errors introduced by internal standard and baseline shift issues. The proposed methods are illustrated for an entire basin where more than 50 samples have been used, covering a wide range of GOR and API. To our knowledge, this is the first attempt to identify and deal with composition errors with a systematic and comprehensive workflow.