Effective Phase Equilibrium Calculation for Equation of State Compositional Reservoir Simulation

• Published in: Journal of The Japan Petroleum Institute Vol. 44; no. 4; pp. 244 - 251
• Main Authors: ARIHARA, Norio, YAZAWA, Nintoku, SATO, Kozo, TAKEDA, Hideaki, HONAMI, Yukiko
• Format: Journal Article
• Language: English; Japanese
• Published: Tokyo The Japan Petroleum Institute 01.07.2001; Sekiyu Gakkai
• Subjects: Compositional model; Equation of state; Reduction method
• ISSN: 0582-4664
• DOI: 10.1627/jpi1958.44.244

Field studies based on compositional simulations require a huge number of phase equilibrium calculations resulting in costly computing times. The most common method for reducing the computing time is to simplify the fluid description by combining its components into several pseudo-components. The main drawback of such pseudoization is the loss of detailed compositional information about the reservoir fluids. This study evaluated the accuracy and efficiency of different computational methods for the equation of state incorporated in a compositional simulation model. The compositional model was formulated by the IMPECS approach. The iterative EOS flash calculations were performed by the successive substitution iteration (SSI) method, a combination of the SSI and Minimum Variable Newton-Raphson methods (SSI+MVNR), and a combination of the direct flash calculation and MVNR methods (DFC+MVNR). These three flash algorithms were implemented in a generalized Michelsen method, in which the number of the independent variables was optionally reduced. The flash calculation methods were evaluated by simulating the behavior of a single well in a gas condensate reservoir, in which the reservoir fluid was grouped into 5, 10, and 16 pseudo-components. The simulation results demonstrated high effectiveness for the SSI+MVNR and DFC+MVNR combination methods with parameter reduction, resulting in significant decreases in the flash computing time compared to the conventional method, particularly when the number of pseudo-components was increased. The computing time decreased as the number of parameters is decreased. However, setting all BIP to zero is not recommended, because the simulation results were quite different for the case of all zero BIP and the cases of non-zero BIP assigned to more than one component.