Numerical simulations of surfactant flooding in carbonate reservoirs

Abstract

We demonstrate how geological heterogeneity impacts the effectiveness of surfactant-based enhanced oil recovery (EOR) at larger (inter-well and sector) scales when upscaling small (core) scale heterogeneity and physicochemical processes. We used two experimental datasets of surfactant-based EOR where spontaneous imbibition and viscous displacement, respectively dominate recovery. We built 3D core-scale simulation models to match the data and parameterize surfactant models. The results were deployed in high-resolution models that preserve the complexity and heterogeneity of carbonate formations in the inter-well and sector scale. These larger-scale models were based on two outcrop analogues from France and Morroco, respectively, which capture the reservoir architectures inherent to the productive carbonate reservoir systems in the Middle East. We then assessed and quantified the error in production forecast that arises due to upscaling, upgridding, and simplification of geological heterogeneity. Simulation results showed a broad range of recovery predictions. The variability arises from the choice of surfactant model parameterization (i.e., spontaneous imbibition vs viscous displacement) and the way the heterogeneity in the inter-well and sector models was upscaled and simplified. We found that the parameterization of surfactant models has a significant impact on recovery predictions. Oil recovery at the larger scale was observed to be higher when using the parametrization derived from viscous displacement experiments compared to parameterization from spontaneous imbibition experiments. This observation clearly demonstrated how core-scale processes impact recovery predictions at the larger scales. Also, the variability in recovery prediction due to the choice of surfactant model was as large as the variability arising from upscaling and upgridding. Upscaled and upgridded models overestimated recovery because of the simplified geology. Grid coarsening exacerbated this effect because of the increased numerical dispersion. These results emphasize the need to use correctly configured surfactant models, appropriate grid resolution that minimizes numerical dispersion, and properly upscaled reservoir models to accurately forecast surfactant floods. Our findings present new insights into how the uncertainty in production forecasts during surfactant flooding depends on the way surfactant models are parameterized, how the reservoir geology is upscaled, and how numerical dispersion is impacted by grid coarsening.