G-2 and G-3 Reservoirs, Delta South Field, Nigeria: Part 2 - Simulation of Water Injection

Abstract

Summary To date, laboratory experiments to evaluate a fieldprospect for CO2 flooding have concentrated on determining minimum miscibility pressure. Other experiments include core floods and single- and multiple-contact phase behavior studies. This paper discusses relationshipsbetween the various experiments in the light of current understanding of the mechanisms operating in the CO2-flooding process. Objectives and limitations ofeach experiment are discussed. Limitations on the use of the resulting data to estimate recovery from field CO2 floods also are reviewed. A new experimental procedure is described that permits rapid determination of phasebehavior and fluid properties of CO2 /crude-oil mixtures at reservoir conditions. The experimental procedure, which involves continuous contacting of the reservoirfluid with CO2, can be used efficiently to obtainphase-behavior and fluid-property data that would require difficult and time-consuming experiments if obtained bypreviously available techniques. Introduction Although CO2 flooding has been investigated since theearly 1950's, consensus has not developed as to thelaboratory experiments required for the evaluation of a CO2-flooding prospect. Laboratory experimentsperformed most often, however, fall into three generalareas:slim-tube displacements,core displacements, andhigh-pressure volumetric (PVT)and vapor/liquid equilibrium (VLE) experiments. Thethree types of experiments yield different information about a potential CO2 flood. This paper reviews brieflythe uses and limitations of the information obtained inthe various laboratory experiments. Design of any large-scale CO2 flood inevitably involves the use of numerical reservoir simulators. Those applicable to CO2 flooding also fall into three general groups:miscible simulators,compositional simulators, andhybrid miscible/compositional simulators. The type of data required to support a simulation effort depends (in part, at least) on the choice of the simulator. Miscible simulators based on the work of Todd and Longst afftreat CO2 as completely miscible with theoil, and require little more than the permeability, relative-permeability, capillary-pressure, viscosity, anddensity data common to black-oil simulators. Routine reservoir-fluid volumetric and fluid-property measurements, in addition to standard core characterizations (permeability, relative permeability, and capillary pressure), are adequateto support such simulations. JPT P. 888^