Subsurface-Guided Production Surveillance for High-Confidence Operational Decisions

Abstract

Abstract The surveillance team in an oilfield has the difficult task of maximizing hydrocarbon production while delaying water production to achieve optimum profitability. For instance, in a waterflooded asset, it needs to intelligently allocate the available injection water to achieve a balanced sweep of oil across the reservoir. A sound understanding of the subsurface flow and inter-well communication is essential here, but the team rarely has access to high-fidelity tools that can help them understand the reservoir behavior. Reservoir simulation models encapsulate all the acquired data along with the interpretations of the subsurface teams and are thus ideal tools to base such decisions on but are seldom used in operations as the associated workflows do not conform to the fast decision-making timeframe. This paper presents a system that leverages cloud scalability, automation, and data analytics to extract insights from subsurface models and generate timely operational advice. The solution connects subsurface models with real-time production data through a cloud-based data platform to automate the update of models with the latest production data. An optimizer is employed that uses streamline-based properties to determine the optimum operating settings for the injection and production wells. The optimization objective can be tailored to align with the asset management goals, such as reducing water recycling and balancing recovery or voidage across the field. The outputs from the subsurface model are translated into actionable insights through a dashboard of fit-for-purpose analytics that presents operational recommendations along with the forecasted outcomes. The system also performs a series of domain-derived confidence checks on the model to quantify the reliability of the recommendations generated. A virtual field management framework is used that captures all the field operating constraints. The entire workflow is automated and can be scheduled to run at a defined frequency so that the surveillance team always has access to proposed actions based on the latest production conditions. To further accelerate the time to decision, machine learning-based avatars of the full subsurface model and reduced-order representations can be integrated into the framework. A case study is presented that describes the application of this subsurface model-driven operational optimization system to a field in the Amazon basin, South America. Using the solution, the subsurface modeling, production surveillance, and operations teams were able to work together to identify opportunities for reducing water recycling and increasing oil production while considerably accelerating the decision-making process due to automation and focused analytics. The paper demonstrates how the latest digital technologies have removed the barriers to the use of detailed subsurface models in guiding operations. The framework described can be used to improve the operational decision-making in any hydrocarbon asset regardless of the recovery mechanism.