A New Approach to Limited Compositional Simulation: Direct Solution of the Phase Equilibrium Equations

Abstract
Summary A new method for solving the phase equilibrium in limited compositional models is presented. It is a direct solution algorithm based on simplified thermodynamics and can be easily implemented in existing simulators to model such mechanisms as retrograde condensation, oil stripping by gas cycling, and miscible and immiscible gas drive. In field-scale studies, the simplification in phase behavior is more than compensated for by the increase in simulator efficiency. Examples are given to illustrate the method's use in large-scale gas cycling. Introduction Many reservoir simulation studies require the inclusion of certain compositional effects, such as retrograde condensation, stripping of volatile oil by gas cycling, and the salting out of solution gas by injected gas in an immiscible gas drive. The numerically efficient black-oil simulators are incapable of describing such mechanisms. Compositional simulators will model the physics but may require impractical amounts of human and computing resources for large-scale, full-field studies. Some tool of efficiency comparable to the black-oil simulator that is capable of modeling the essential features of the key compositional effects is therefore desirable. Preferably, this tool also takes very little time to develop. Considerable investment has gone into developing an existing black-oil, compositional, or general-purpose simulator, so it is advantageous to start with that framework, and customize it to a specific application. We shall discuss a methodology for effecting such modifications. This new algorithm can be easily implemented in an existing implicit pressure, explicit saturation (IMPES) simulator to include some of these compositional effects without adding substantial computations. Coats1 and Watts2 proposed a single general-purpose simulator that is extremely desirable from the maintenance viewpoint. This new formulation can also be used effectively inn such a simulator for certain otherwise prohibitively expensive full-field studies. This direct solution algorithm first locates the state of the fluid mixture in any cell on the phase diagram through a sequence of logical tests. The phase saturation and volume fractions in each phase can then be computed. The appearance and disappearance of phases are handled automatically. For tracer studies, each component can be subdivided into a number of subcomponents with similar dynamic properties. The phase equilibrium module is easily interfaced with the pressure solution in either type of base simulator. Example applications are presented to demonstrate that a complex seven-component system can be modeled with practically the same computational effort required of a three-component black-oil system. Equations for an Isothermal Reservoir Simulator The standard reservoir simulator is a discretized version of the set of conservation equations (one for each component in each cell) of the formEquation 1 whereEquation 2 coupled with some auxiliary thermodynamic relations,Equation 3 andEquation 4 and consistency relations,Equation 5 andEquation 6 The two common classes of simulators differ in their treatment of the phase behavior. The classic black-oil simulators use pressure-dependent solubilities (Eq. 4) of each component in the different phases and require only the pressure and phase saturations to specify a physical system completely. The compositional simulators, on the other hand, describe the phase behavior by use of vapor/liquid equilibrium relations that depend on phase compositions as well as pressure. In either case, the phase behavior is strongly coupled to the material-balance equations, and they must be solved simultaneously or sequentially. The following is a typical solution procedure.Combine all the material-balance equations into a single equation in the unknown pressure at the new timestep with saturations and compositions from the previous timestep.Solve for the new pressure, ignoring the saturation and compositional dependence in the system compressibility during the iterations.(a) For the black-oil model, solve the individual material-balance operations for the new phase saturations and therefore the amount of each component at the new timestep. (b) For the compositional model, solve the individual material-balance equations for the amount of each component at the end of the timestep, then flash the mixture at the new pressure to give the new phase saturations.Optionally update the system compressibility by use of the new saturations and compositions, and repeat Steps 2 and 3.Repeat for new timestep.