An Expert System to Aid in Compositional Simulation of Miscible Gas Flooding
An expert system (UTINPUT) has been developed for preparing appropriate input data sets to design miscible gas floods using The University of Texas Compositional Simulator (UTCOMP). The input data values for UTCOMP must be carefully assigned to represent the petroleum reservoir with an appropriate model and to make accurate and efficient simulations. Using emerging expert system technology, this expertise has been implemented on a workstation using an expert system shell. The resulting expert system is interactive, user-friendly, efficient, and provides the necessary expertise to run compositional simulations. A major part of this research has been the accumulation of expertise through literature, simulation experts, and to a much greater extent through conducting detailed simulation studies with emphasis on reservoir fluid characterization, selection of timestep size, and the design of miscible gas floods. A reservoir fluid characterization scheme has been developed to characterize C02/oil mixtures forming up to three phases. A new correlation for COr hydrocarbon binary interaction coefficients was required to make accurate predictions of three-phase liquid-liquid-vapor region as well as the liquid-liquid region at higher pressures. The characterization scheme has been validated for four West Texas oils. Comparisons between actual three-phase solutions and the corresponding nonphysical two-phase solutions are also shown. Selection of appropriate timestep size is very important in compositional simulation to avoid numerical instabilities and to save computer time. A wide variety of cases have been investigated to understand timestep size control using the method of relative changes. The governing control criteria were identified and appropriate ranges were determined. Simulation design studies were conducted to develop an insight into the miscible gas flooding design procedure. The effects of WAG ratio, solvent slug size per WAG cycle, and the total solvent slug size on oil recovery were investigated under various C02 flooding conditions using field data. Sensitivities of oil recovery to three-phase relative permeability models, gravity number, and effective length-to-thickness ratio were investigated. Comparisons between two-and three-dimensional simulations were also made. Finally, a systematic procedure was developed for designing miscible gas floods using a compositional simulator.