A Novel Procedure for Addressing Compositional Uncertainty in Volatile Oils, Critical Fluids, and Gas Condensates

Abstract

Abstract Accurate fluid compositions are prerequisite to the development of reservoir fluid Equation of State (EoS) models; however, previous work shows that measured plus fractions can have relative deviations exceeding 25%. This paper addresses the issue of uncertainty in reservoir fluid compositions with focus on the plus fraction amount in fluids with a GOR greater than 600 Sm3/Sm3. A new modeling workflow handling uncertainty in the plus fraction amount to improve the accuracy of predicted EoS simulations is presented. The paper presents a comprehensive review of previous studies on composition measurement and their limitations. Based on this review, a new modeling workflow is proposed that can identify and correct potential problems with the measured fluid composition when a composition issue is identified. The workflow involves an automated procedure that determines the optimum carbon number to lump back to during fluid characterization such that the extrapolated fluid composition gives the best match to the measured fluid composition. As opposed to methodologies that rely on artificial intelligence, the proposed fluid characterization procedure is fully transparent and interpretable. The workflow is found to be suitable for volatile oils, critical fluids, and gas condensates and is tested on several samples with plus fraction uncertainties identified in their composition measurements. Previous literature has shown that characterization methods can typically predict saturation pressures with relative deviations below 10%. However, we find that uncertainties in plus fraction amounts can potentially cause simulated saturation pressures to deviate by 30% or more from experimental measurements. When applying the new methodology to a variety of measured experimental reservoir fluid composition data, we demonstrate that the experimental match of saturation pressures can be brought back within the expected deviation of 10%. In this way, the characterization model becomes predictive, requiring only minor tuning of EoS parameters. The workflow identifies the potential problem of inaccurate plus fraction measurements and provides an automated method for handling this uncertainty in compositional analysis. By doing so, the proposed workflow can provide more accurate and reliable predictions of the phase behavior of light reservoir fluids before any tuning, which can be of great benefit to the petroleum industry. This paper presents novel information on how to handle the uncertainty in the plus fraction amount for reservoir fluid compositions, which is a critical issue that has not been fully addressed in previous literature. The proposed workflow offers a practical solution to this problem, providing a new approach to correcting reservoir fluid compositions. The paper can be of significant benefit to the petroleum industry by improving the reliability of fluid characterizations for lighter reservoir fluids.