Rapid Uncertainty Quantification and Forecasting in Mature Floods Using Streamline Surveillance Models

Abstract Effective management of mature reservoir flooding operations, which include water, CO2, polymer, and solvent injections, relies heavily on reservoir flow models. These models are essential for evaluating flood performance and identifying optimization opportunities. Our study enhances the streamline-based reservoir surveillance model (SLSV)—a much more efficient and effective approach than traditional reservoir simulation—with data-driven techniques to quantify forecast production uncertainty due to explicit geological variability and original fluids in place (OFIP). In this work, we explore the use of probabilistic fluids in place (FIP) maps generated from SLSV models to facilitate quick yet informed well intervention decisions. These interventions are aimed at targeting bypassed oil and achieving incremental production compared to a no-further-action (NFA) scenario. Probabilistic FIP maps are derived by running an exhaustive set of SLSV models across an ensemble of geological models and variability in original fluids in place (OFIP). Unlike traditional reservoir simulation, the surveillance approach requires none or minimal well-level history matching to generate remaining FIP maps allowing a much larger ensemble of models to be used because SLSV models are numerically more efficient than simulation models. Surveillance models leverage historical injected and produced surface well volumes and streamlines to define control volumes (CV) to construct remaining FIP maps using material balance.