document

Summary

Compositional reservoir simulators that are based on equation-of-state (EOS) formulations typically do not handle the modeling of aqueous phase behavior, and those that are designed for modeling chemical processes typically assume simplified hydrocarbon phase behavior. There is a need to have a single reservoir simulator capable of combining both approaches to benefit from the advantages of both aqueous and hydrocarbons models. Developing and implementing fully implicit procedures for modeling both hydrocarbon and aqueous phase behavior simultaneously is a complex process. An approach to integrate a surfactant phase behavior model into an existing fully implicit, parallel, EOS compositional simulator is presented in this paper. Physical property models describing the flow and transport of surfactant and polymer species have been implemented. These properties include surfactant phase behavior, interfacial tension, capillary desaturation, viscosity, adsorption, and relative permeability as a function of trapping number. Polymer properties include viscosity, permeability reduction, inaccessible pore volume, and adsorption. The simulation results were validated by comparison with the explicit chemical-flooding simulator UTCHEM and are shown in this paper. Test runs were performed with high-resolution models in a parallel environment, with results indicating a good scalability of the simulator.

Introduction

Increased oil production using improved oil recovery processes requires numerical modeling of such processes to minimize the risk involved in development decisions. The oil industry is requiring much more detailed analyses with a greater demand for reservoir simulation with geological, physical, and chemical models of much more detail than the past. Reservoir simulation has become an increasingly widespread and important tool for analyzing and optimizing oil recovery projects.

Numerical simulation of large petroleum reservoirs with complex recovery processes is computationally challenging because of the problem size and detailed property calculations involved. This problem is compounded by the finer resolution needed to model such processes accurately. Traditionally, such simulations have been performed on workstations or high-end desktop computers. These computers restrict the problem size because of their address- able memory limit, and simulation studies of the entire project life become time-consuming. Parallel reservoir simulation, especially on low-cost, high-performance computing clusters, has alleviated these issues to a certain extent. Recent publications describe the development of such approaches and emphasize the necessity and advantages of using parallel processing. 1--4

Compositional reservoir simulators that are based on EOS formulations do not handle the modeling of aqueous phase behavior and those that are designed for chemical-flood modeling typically assume simplified hydrocarbon phase behavior. There is need to have a single reservoir simulator capable of combining both approaches to benefit from the advantages of both models. The overall objective of this research is to develop such technology using a computational framework that also allows parallel processing. The initial stage of development involved the formulation of a fully implicit, parallel, EOS compositional simulator. 5 The description of the framework approach used for modular code development and the application to gas injection is in Wang et al. 6

In this paper, we focus on the implementation of the chemical module to the existing EOS simulator, its validation, and its application to large-scale chemical-flooding simulations. The formulation of the compositional model is briefly described. The assumptions for the chemical model and its formulation are described next. We use Hand's rule 7 to describe surfactant/oil/brine Type II(--) phase behavior. The trapping number model for relative permeability is implemented to capture the changes in residual saturations caused by the lowered interfacial tension. The validation of the implementation against the explicit chemical flooding simulator UTCHEM is shown. Application to large-scale problems and tests showing the parallel performance of the simulator are described. The approach we used to couple the models is easy to implement, computationally efficient, and extendable to many other interesting reservoir problems involving aqueous chemistry. With the capability of parallel processing, the general purpose adaptive simulator (GPAS) can now be used to simulate chemical flooding on a larger scale than before.