A New Adaptive Implicit Method for Multicomponent Surfactant-Polymer Flooding Reservoir Simulation

Abstract

Summary In the oil industry, chemicals can improve oil production by mobilizing trapped and bypassed oil. Such processes are known as chemical-enhanced oil recovery (CEOR). Surfactants and polymers are important chemicals used in CEOR with different mechanisms to improve oil recoveries, such as reduction in residual saturation, oil solubilization, and mobility control. However, both surfactant and polymer may increase the cost of oil production, making optimizing these processes essential. Reservoir simulators are tools commonly used when performing such field optimization. The simulation of surfactant flooding processes has been historically performed with the implicit pressure explicit composition (IMPEC) approach. The injection of surfactants requires modeling the brine/oil/microemulsion phase behavior along with other processes, such as capillary desaturation and retention. The microemulsion phase behavior and the complex relative permeability behavior can lead to convergence issues when using fully implicit (FI) schemes. Only recently, the FI approach has been efficiently applied to simulate this process using new modeling. The adaptive implicit method (AIM) can combine the benefits of the FI and IMPEC approaches by dynamically selecting the implicitness level of gridblocks in the domain. This work presents a new AIM in conjunction with recently developed models to mitigate discontinuities in the microemulsion relative permeabilities and phase behavior. The approach presented here considers the stability analysis method as a switching criterion between IMPEC and FI. To the best of our knowledge, the approach presented here is the first AIM to consider the brine/oil/microemulsion three-phase flow in its conception. The new approach uses the finite volume method in conjunction with Cartesian grids as spatial discretization and is applied here for field-scale problems. The new approach is tested for polymer flooding and surfactant-polymer (SP) flooding for problems with several active cells ranging from about a hundred thousand to almost a million. The AIM approach was compared with the FI and IMPEC approaches and displayed little variation in the computational performance despite changes in the timestep size. The AIM also obtained the fastest performance for all cases, especially for SP flooding cases. Furthermore, the results here suggest that the gap in performance between the AIM and FI seems to increase as the number of gridblocks increases.