Cross-flow area recognition in fractured oil reservoirs and foam flow characteristics

Fracture-cavity reservoirs are prone to frequent channeling pathways during foam flooding due to their complex fracture networks, which severely restricts the enhancement of oil recovery and CO₂ storage efficiency. To address this critical issue, this study proposes a dynamic path identification method that integrates an improved Dijkstra algorithm with multiphysics simulation, aiming to quantitatively reveal channeling mechanisms and optimize mitigation strategies. By establishing a fracture network model, the traditional Dijkstra algorithm is enhanced into a dynamic weight assignment model, where weights are dynamically calculated based on fracture geometric parameters (e.g., tortuosity, roughness) and fluid properties (e.g., viscosity, density). Coupled with the Darcy-Weisbach equation and mass-momentum conservation equations, the model achieves precise predictions of foam fluid density (error < 0.01 g/m³) and pressure drop (error < 0.01 Pa). Simulation results demonstrate that doubling fracture tortuosity leads to a 50% reduction in flow rate, while increased surface roughness reduces flow rate by 46.2%. By classifying cumulative flow values (≥ 0.08 m³/s as dominant channels), the model successfully identifies high-conductivity zones. Compared to existing static models, this study pioneers the coupling of graph theory algorithms with dynamic fluid behavior, overcoming the limitations of traditional methods in adapting to complex fracture networks. The findings provide a robust theoretical foundation for the efficient development of fractured reservoirs and the synergistic integration of CO₂-enhanced oil recovery with carbon sequestration under carbon neutrality goals.