Creation of geological and simulation models is the necessary condition for decision making towards current development status, planning of well interventions, field development planning and forecasting. In case of isothermal process, for proper phase behavior and phase transitions two key approaches are used: a) simplified model of non-volatile oil, so called “black oil” model, in which each phase – oil, water and gas, are represented by respective component, and solution to fiow equations is based on finding the saturations and pressures in each numerical cell, and change of reservoir fiuid properties is defined in table form as a function of pressure; b) compositional model, in which based on equation of state, phase equilibrium is calculated for hydrocarbon and non-hydrocarbon components, and during fiow calculations, apart from saturations and pressures, oil and gas mixture is brought to phase equilibrium, and material balance is calculated for each component in gas and liquid phase. To account for components volatility, the classic black oil model was improved by adding to the formulation gas solubility and vaporized oil content. This allows its application for the majority of oil and gas reservoirs, which are far from critical point and in which the phase transitions are insignificant. Due to smaller number of variables, numerical solution is simpler and faster. But, considering the importance and relevance of increasing the production of Ukrainian gas and optimization of gas-condensate fields development, the issue of simplified black oil PVT-model application for phase behavior characterization of gas-condensate reservoirs produced under natural depletion depending on the liquid hydrocarbon’s potential yield. Comparative study results on evaluation of production performance of synthetic reservoir for different synthetically-generated reservoir fiuids with different С5+ potential yield is provided as plots and tables. Based on the results the limit of simplified black oil PVT-model application and the moment of transition to compositional model for more precise results could be defined.
Bayesian Estimation of Linear Mixtures Using the Normal Compositional Model. Application to Hyperspectral Imagery
