New fully implicit formulations for the multicomponent surfactant-polymer flooding reservoir simulation

Abstract

The injection of chemicals such as polymers, surfactants, and alkali into oil reservoirs are referred to chemical enhanced oil recovery (CEOR) and can significantly increase the recovery of oil. Polymer increases water viscosity, yielding better mobility control, which reduces the amount of oil bypassed. Surfactants reduce capillary forces by lowering the interfacial tension, allowing the recovery of trapped oil that is normally left behind in conventional waterflooding. Despite the strong appeal that these two techniques offer, the injection of chemicals is expensive, and proper field optimization is required to ensure a robust recovery design. Such optimization can be carried out with the help of a reservoir simulator, a numerical tool capable of emulating the multiphase flow in the reservoir rock. Most attempts to develop chemical flooding reservoir simulators have faced challenges due to the particular nonlinearities of the physicochemistry for modeling CEOR. Therefore, most simulators reported in the literature use the IMPEC (implicit pressure, explicit concentrations) approach, which suffers from time-step restrictions, or fully implicit approaches that do not cover the whole range of the microemulsion (ME) phase behavior. This work presents new models and fully implicit approaches for surfactant-polymer flooding considering all important phase behavior types and dependency on salinity, assuming no gas is present and that the brine, oil, and microemulsion phases are slightly compressible. The oil phase is modeled as a single component phase but multiple components (i. e. surfactant, anions, polymer) are transported in aqueous and surfactant-rich microemulsion phases, therefore making this a multicomponent model. This paper presents three different fully implicit approaches to solve the flow and transport equations, along with new relative permeability models that smooth and minimize discontinuities at phase changes, thereby improving the convergence of the Newton scheme. The first scheme uses pressure and concentrations as primary variables, while the second is based on natural variables, and the third one considers a mixed variable scheme. These schemes are compared with an IMPEC approach, and the different numerical models are compared in terms of oil recovery, production rates, saturation profiles, CPU time, Newton iterations, and linear solver iterations. The new formulations were implemented into the University of Texas at Austin Chemical Flooding Reservoir Simulator (UTCHEM-RS), a variation of the well-established UTCHEM. It was observed that the fully implicit (FI) approaches have a better computational performance than the IMPEC approach for polymer flooding, while the FI global variables presented a better computational performance for field-scale surfactant-polymer flooding through the increase of the time-step. Accuracy was observed to deteriorate with the increasing of time-step, although it had minor effects on the oil recovery. However, the results suggests that caution must be taken when selecting the time-step.