Application of the FlowNet model for history matching and production optimization in water alternating gas enhanced oil recovery

Abstract

Enhanced oil recovery (EOR) strategies, particularly CO2 flooding, play a crucial role in optimizing oil reservoir exploitation while addressing carbon sequestration. Despite their effectiveness, the application of these techniques is often hindered by complex reservoir dynamics and the computational intensity of traditional simulation models. This study introduces a novel approach utilizing the FlowNet model, which combines data-driven analytics and physics-based modeling, aimed at expediting history matching and production optimization processes. The FlowNet model simplifies the representation of reservoirs by using virtual well points along flow paths and employs a non-linear solver for quick resolution of flow equations. Our method significantly enhances the efficiency of history matching by reducing computational overheads and leveraging streamlined network structures, thereby facilitating faster and more accurate production forecasts. We implement the model in several case studies involving CO2 and water alternating gas flooding, which demonstrate an 11% increase in the economic net present value compared to traditional methods. These findings highlight the potential of integrating data-driven techniques with physical modeling to improve EOR performance predictions and optimize production strategies, ultimately promoting more sustainable and economically viable oil recovery practices.