A Beta-Type Reservoir Simulator for Approximating Compositional Effects During Gas Injection

Publisher Website
Google Scholar
Abstract
This paper presents the formulation and applications of a two-dimensional two-phase beta-type numerical model for simulating oil and gas reservoir performance where fluid compositional effects are significant. The model utilizes PVT data as functions of pressure and a compositional parameter to reflect changes in fluid composition parameter to reflect changes in fluid composition resulting from dry gas injection. The model differs from previous beta-type simulators for approximating compositional effects in that it accounts for the reduced tendency of oil to vaporize as light ends are removed by continued contact with dry gas. A simple linear compositional model is used to compute the changing fluid properties in each cell as injection gas mixes with in-place fluid during a series of constant pressure displacements. The simple-model data are correlated against the cumulative pore volumes of injected gas contacting each cell at various points in time for each pressure level. By tracking this parameter for each cell in the two-dimensional beta-type model the spatial and time variations of fluid properties with pressure and injection gas throughput are computed in the model from the correlations. Special provisions are included for gas condensate systems above their dew-point pressure and for reservoir oil systems above their bubble-point pressure. A comparison is made with results from a previously published fully compositional simulator. It is shown that, in addition to yielding quite similar computed compositional effects, the beta-type simulator attains an advantage in computing speed of at least 3 to 1 over a July compositional simulator. Example applications are presented for gas injection in an oil reservoir and for retrograde liquid recovery by dry gas sweep at a Pressure considerably below the dew point of a gas-condensate reservoir. Introduction Conventional beta-type numerical simulators have been used for many years to simulate the performance of so-called "black oil" reservoirs. The term "black oil" denotes oil of medium to heavy gravity at moderate temperature and pressure. Such oils can be reasonably approximated as binary fluid systems where the amount of gas dissolved in the oil is merely a function of reservoir pressure and temperature. For these systems the reservoir gas phase is assumed to contain no recoverable liquids phase is assumed to contain no recoverable liquids when flashed through surface separates. Moreover, the further assumption is made that injected gas combines with reservoir oil exactly as does in-place reservoir gas, disregarding compositional differences between the gas phases. For oils of higher gravity, existing at higher temperatures and pressures, the preceding assumptions become no longer valid. Not only does the reservoir gas phase contain a significant amount of vaporized stock-tank liquid, but injected gas can have a significant effect on the phase behavior of the reservoir hydrocarbon system. At the far end of the compositional spectrum, gas - condensate reservoirs contain the total stock-tank liquid in the vapor phase. Moreover, injected gas has a diluting effect on the liquid content as it mixes with reservoir gas. To account for the collects of mass transfer between the vapor and liquid phases, and the composition changes resulting from gas injection, several recent authors have developed fully compositional reservoir simulators. These simulators actually track individual components of a hydrocarbon fluid system, using equilibrium constants representative of the fluid compositions being studied. These fully compositional simulators are applicable to a much broader range of reservoir fluid systems than the conventional beta-type reservoir simulator. However, since the computing requirements are generally directly proportional to the number of hydrocarbon components used in the fluid system, they can be quite expensive to employ as a general purpose simulator. purpose simulator. SPEJ P. 471

This publication has 0 references indexed in Scilit: