As the development of oilfields in China enters its middle-to-late stage, the old oilfields still occupy a dominant position in the production structure. The seepage process of reservoirs in the high Water Content Period (WCP) presents significant nonlinear and non-homogeneous evolution characteristics, and the traditional seepage-modeling methods are facing the double challenges of accuracy and adaptability when dealing with complex dynamic scenarios. In recent years, Deep Learning technology has gradually become an important tool for reservoir seepage field prediction by virtue of its powerful feature extraction and nonlinear modeling capabilities. This paper systematically reviews the development history of seepage field prediction methods and focuses on the typical models and application paths of Deep Learning in this field, including FeedForward Neural networks, Convolutional Neural Networks, temporal networks, Graphical Neural Networks, and Physical Information Neural Networks (PINNs). Key processes based on Deep Learning, such as feature engineering, network structure design, and physical constraint integration mechanisms, are further explored. Based on the summary of the existing results, this paper proposes future development directions including real-time prediction and closed-loop optimization, multi-source data fusion, physical consistency modeling and interpretability enhancement, model migration, and online updating capability. The research aims to provide theoretical support and technical reference for the intelligent development of old oilfields, the construction of digital twin reservoirs, and the prediction of seepage behavior in complex reservoirs.

The transport behavior of multiphase flow in porous media is governed by the cross-scale coupling between fluid properties and pore structure, and serves as the theoretical foundation for core processes in fields such as energy development, underground carbon storage, and environmental remediation. Accurately characterizing the intrinsic relationship between physical properties and seepage responses is crucial for enhancing engineering prediction capabilities and optimizing operational strategies. However, the inherent heterogeneity and multiscale nature of natural reservoirs, coupled with the limitations of traditional experimental methods in terms of optical opacity and spatiotemporal resolution, severely hinder a deep understanding of the mechanisms of multiphase flow at the pore-scale. This paper systematically reviews the methodological framework for characterizing physical properties and seepage mechanisms in multiphase flow systems, with a focus on cutting-edge breakthroughs in experimental measurement and visualization technologies over the past decade. Starting with classical and emerging testing methods for key physical properties such as saturation, relative permeability, capillary pressure, and interfacial tension, the paper evaluates the applicability, accuracy advantages, and inherent limitations of different techniques. The paper focuses on the latest advancements in pore-scale visualization technologies, covering microfluidic models, high-resolution X-ray CT, synchrotron rapid dynamic imaging, and multimodal, multiscale imaging fusion strategies; it also explores AI-enabled image processing and data analysis methods, as well as the application potential of cross-scale numerical coupling models in revealing transient seepage mechanisms and correlating them with macroscopic responses. Based on this, an integrated analytical framework of “physical property measurement—visualization characterization—theoretical modeling—engineering application” is established, and the core challenges and future pathways for advancing multiphase flow and seepage research toward “quantification of mechanisms, cross-scale correlation, and adaptation to in situ real-world conditions” are identified.

Tight sandstone reservoirs are characterized by complex pore structures and ultra-low permeability, resulting in low recovery efficiency when relying solely on formation elastic energy. Consequently, water flooding is commonly employed for reservoir development. To investigate the two-phase flow behavior and pressure response during water flooding in tight sandstone reservoirs, long-core displacement experiments were conducted using outcrop tight sandstone cores. Thirteen pressure monitoring points were distributed along the core to continuously record the pressure response during water flooding, while cumulative oil and water production were measured to characterize the displacement performance. The temporal evolution of pressure and productivity index was systematically analyzed, and numerical simulation was further performed to interpret the relationship between pressure variations and water-front propagation. The results show that the cumulative oil production reached 50.95 mL, corresponding to a final oil recovery of 31.26%. No water production was observed during the initial stage of displacement, and water production began at approximately 8000 min, after which both water production and oil production gradually approached stable levels. The pressures at different monitoring locations exhibited a characteristic increase–decrease–increase pattern during the displacement process, whereas the productivity index decreased progressively before approaching a relatively stable value. Based on the experimental observations and numerical simulation results, the pressure response characteristics were closely associated with the propagation of the water front, with distinct pressure inflection points observed at different locations along the core. The mechanisms underlying pressure fluctuations and the decline in productivity index were analyzed from two perspectives: the extensive formation of dispersed phases and the dynamic evolution of interfacial flow resistance. These results provide experimental and numerical evidence for understanding pressure propagation and productivity evolution during water flooding in tight sandstone reservoirs.

Chemical profile control has long been an effective method for reducing water cut and increasing oil production in heterogeneous reservoirs. However, its application has been limited due to high costs, economic constraints, and difficulties in adapting to complex reservoir environments characterized by high temperatures and high salinity. To address this, this paper proposes a multiphase mixed fluid system in which the aqueous phase serves as the continuous phase and high-viscosity oil droplets as the dispersed phase. Through parallel sand pack and transparent visualization plate displacement experiments, the mechanisms by which this multiphase mixed fluid system enhances oil recovery in heterogeneous reservoirs were systematically investigated. Experimental results indicate that the multiphase mixed fluid system forms dynamic blockages in high-permeability channels via the Jamin effect. In the plate experiments, the permeability of the high-permeability layer decreased from 2227 to 178 mD, effectively achieving fluid redirection and increasing the recovery factor in the low-permeability zone by 8.5%; the system continues to function during the subsequent accelerated displacement phase; visualization experiments intuitively revealed the dynamic process of “water channeling-blockage-redirection,” while sand pack experiments confirmed the significant increase in recovery factor through precise measurement. The study confirms that multiphase mixed fluid technology, through intelligent blockage and energy redistribution, can effectively address the challenge of low sweep efficiency in heterogeneous reservoirs, providing a new technical direction for the development of high-water-cut oilfields.