Polymer flooding is one of the most commonly used chemical enhanced oil recovery (EOR) methods. Conventionally, this technique was believed to improve macroscopic sweep efficiency by sweeping only bypassed oil. Nevertheless, recently it has been found that polymers exhibiting viscoelastic behavior in the porous medium can also improve microscopic displacement efficiency resulting in higher additional oil recovery. Therefore, an accurate prediction of the complex rheological response of polymers in porous media is crucial to obtain a proper estimation of incremental oil to polymer flooding. In this paper, a novel viscoelastic model is proposed to comprehensively analyze the polymer rheological behavior in porous media. This proposed model was developed and validated using 30 coreflooding tests obtained from the literature and further verified against a few existing viscoelastic models.
The proposed viscoelastic model is considered an extension of the unified apparent viscosity model provided in the literature and is termed as extended unified viscoelastic model (E-UVM). The main advantage of the proposed model is its ability to capture the polymer mechanical degradation at ultimate shear rates primarily observed near wellbores. Moreover, the fitting parameters used in the model were correlated to rock and polymer properties using machine learning technique, significantly reducing the need for time-consuming coreflooding tests for future polymer screening works. Furthermore, the E-UVM was implemented in MATLAB Reservoir Simulation Toolbox (MRST) and verified against the original shear model existing in the simulator. It is worth mentioning that the irreversible viscosity drop for mechanical degradation regime was captured during implementing our model in the simulator. It was found that implementing the E-UVM in MRST for polymer non-Newtonian behavior might be more practical than the original method. In addition, the comparison between various viscosity models proposed earlier and E-UVM in the reservoir simulator showed that the latter model could yield more reliable oil recovery predictions as the apparent viscosity is modeled properly in the mechanical degradation regime, unlike UVM or Carreau models.
This study presents a novel viscoelastic model that is more comprehensive and representative as opposed to other models in the literature. Furthermore, the need to conduct an extensive coreflooding experiment can be reduced by virtue of developed correlations that may be used to estimate model fitting parameters accounting for shear-thickening and mechanical degradation.