Summary
A fully implicit, parallel, compositional reservoir simulator has been developed that includes both a cubic equation of state model for the hydrocarbon phase behavior and Hand's rule for the surfactant/oil/brine phase behavior. The aqueous species in the chemical model include surfactant, polymer, and salt. The physical property models include surfactant/oil/brine phase behavior, interfacial tension, viscosity, adsorption, and relative permeability as a function of trapping number. The fully implicit simulation results were validated by comparison with results from our IMPEC chemical flooding simulator (UTCHEM). The results indicate that the simulator scales well using clusters of workstations. Also, simulation results from parallel runs are identical to those using a single processor. Field-scale surfactant/polymer flood simulations were successfully performed with over 1,000,000 gridblocks using multiple processors.
Introduction
Chemical flooding is a method to improve oil recovery that involves the injection of a solution of surfactant and polymer followed by a polymer solution. The surfactant causes the mobilization of oil by decreasing interfacial tension, whereas the polymer increases the sweep efficiency by lowering the mobility ratio. Chemical flooding has the potential to recover a very high fraction of the remaining oil in a reservoir, but the process needs to be designed to be both cost effective and robust, which requires careful optimization. Several reservoir simulators with chemical flooding features have been developed as a tool for optimizing the design (Delshad et al. 1996; Schlumberger 2004; Computer Modeling 2004). The University of Texas chemical flooding simulator, UTCHEM (Delshad et al. 1996) is an example of a simulator that has been used for this purpose. However, because UTCHEM is an Implicit Pressure and Explicit Concentration (IMPEC) formulation and in its current form cannot run on parallel computers, realistic surfactant/polymer flooding simulations are limited to around 100,000 gridblocks because of small timestep restrictions and insufficient memory.
Recently, the appropriate chemical module was added to the fully implicit, parallel, EOS compositional simulator called GPAS (General Purpose Adaptive Simulator) based on a hybrid approach (John et al. 2005). GPAS uses a cubic equation of state model for the hydrocarbon phase behavior and the parallel and object-based Fortran 95 framework for managing memory, input/output, and the necessary communication between processors (Wang et al. 1999; Parashar et al. 1997). In the hybrid approach implemented in GPAS, the material balance equations for hydrocarbon and water components are solved implicitly first. Then, the material balance equations for the aqueous components such as surfactant, polymer, and electrolytes are solved explicitly using the updated phase fluxes, saturations, and densities.