A New Algorithm for Multiphase-Fluid Characterization for Solvent Injection

Abstract

Summary Compositional simulation of solvent injection requires reliable characterization of reservoir fluids by use of an equation of state (EOS). Under the uncertainty associated with nonidentifiable components, reservoir fluids are conventionally characterized in the absence of universal methodology. This is true even for relatively simple fluids involving only the gaseous (V) and oleic (L1) phases. No systematic method has been presented for characterization of more-complex fluids, exhibiting three hydrocarbon phases: the V, L1, and solvent-rich-liquid (L2) phases. This paper presents a new algorithm for systematic characterization of multiphase behavior for solvent-injection simulation. The reliability of the method comes mainly from the binary-interaction parameters (BIPs) newly developed for the Peng-Robinson (PR) (Peng and Robinson 1976, 1978) EOS to represent three-phase behavior, including upper critical endpoints, for n-alkane and carbon dioxide (CO2)/n-alkane binaries. The regression part in fluid characterization broadly follows the concept of perturbation from n-alkanes, which was successfully applied for simpler two-phase fluids in our prior research. The algorithm, in its simplest form, uses only the saturation pressure and liquid density at a given composition and reservoir temperature. Case studies are presented to demonstrate the reliability of the algorithm for 90 reservoir fluids and their mixtures with solvents. Predictions are compared with experimental data for up to three phases. Results show that the simple algorithm developed in this research enables the PR-EOS to predict multiphase behavior in spite of the limited data used in the regression. Without the use of the BIPs developed in this research, the PR-EOS may fail to predict three phases, or may provide erroneous three-phase predictions.