The convolutional neural network (CNN) algorithm in deep learning has been widely applied in petroleum geology research both domestically and internationally. Automated and accurate segmentation of thin-section images of rocks is foundational for in-depth analysis. However, traditional segmentation methods for reservoir rock thin sections often suffer from low accuracy and high cost. To address these issues, this paper proposes a novel segmentation algorithm based on an improved UNet network, integrating residual networks and the CBAM attention mechanism. By incorporating residual modules, the network depth is expanded, and the CBAM attention mechanism enhances the feature weighting capability during learning. Experimental results demonstrate that this method outperforms traditional approaches in both segmentation accuracy and efficiency, representing significant advancements in reservoir rock thin-section image segmentation.
In the process of oilfield development, the surfactant–polymer (SP) composite system has shown significant effects in enhancing oil recovery (EOR) due to its excellent interfacial activity and viscoelastic properties. However, with the continuous increase in the volume of composite flooding injection, a decline in injection–production capacity (I/P capacity) has been observed. Through the observation of frozen core slices, it was found that during the secondary composite flooding (SCF) process, a large amount of residual oil in the form of intergranular adsorption remained in the core pores. This phenomenon suggests that the displacement efficiency of the composite flooding may be affected. Research has shown that polymers undergo flocculation reactions with clay minerals (such as kaolinite, Kln) in the reservoir, leading to the formation of high-viscosity mixtures of migrating particles and crude oil (CO). These high-viscosity mixtures accumulate in local pores, making it difficult to further displace them, which causes oil trapping and negatively affects the overall displacement efficiency of secondary composite flooding (SCF). To explore this mechanism, this study used a microscopic visualization displacement model (MVDM) and microscopy techniques to observe the migration of particles during secondary composite flooding. By using kaolinite water suspension (Kln-WS) to simulate migrating particles in the reservoir, the displacement effects of the composite flooding system on the kaolinite water suspension, crude oil, and their mixtures were observed. Experimental results showed that the polymer, acting as a flocculant, promoted the flocculation of kaolinite during the displacement process, thereby increasing the viscosity of crude oil and affecting the displacement efficiency of secondary composite flooding.
In the process of oilfield development, the surfactant-polymer (SP) composite system demonstrates significant effectiveness in enhancing oil recovery. However, as the injection volume of the composite flooding increases, a decline in injectivity and productivity is observed. Frozen core slicing reveals that binary composite flooding results in the formation of residual oil adhering between particles. Research indicates that the polymer induces a flocculation reaction among clay and other migrating particles in the reservoir, which subsequently mix with crude oil to form a high-viscosity mixture. This mixture tends to accumulate locally, making it difficult to displace. In this study, a microvisual displacement model and microscopy were employed to analyse and observe these phenomena. Kaolinite aqueous suspensions were used to simulate migrating particles in the reservoir. Comparative experiments were conducted using the composite system to displace kaolinite suspensions, crude oil, and mixtures of kaolinite suspensions and crude oil. The findings ultimately confirm that the polymer acts as a flocculant, causing kaolinite particles to aggregate. This, in turn, increases the viscosity of the crude oil and negatively impacts the displacement efficiency of the binary composite flooding system.
The timing and geodynamic processes of the transition from oceanic subduction to continental collision along the Heihe–Hegenshan Suture remain debated, particularly the relationship between initial collision and final amalgamation of the Xing’an and Songliao-Xilinhot blocks. This study presents new zircon U-Pb age determinations and comprehensive whole-rock geochemical analyses for Late Paleozoic magmatic and metamorphic rocks from the Xinkailing Complex, situated on the northeastern margin of the Xing’an Block in the Nenjiang area. These findings provide new constraints on the Late Paleozoic tectonic evolution of the Heihe-Hegenshan Suture. Our new U-Pb data indicate a maximum depositional age of 311 Ma for the sedimentary protolith of biotite plagioclase gneiss. The magmatic rocks, including quartz diorites, gneissic granodiorite and the magmatic protolith of amphibolite, were emplaced from the Late Carboniferous to the Early Permian, at 318–319 Ma, 288 Ma, and 278 Ma, respectively. Metamorphic zircon populations record subsequent metamorphism during the Middle Jurassic, at approximately 175–169 Ma, and 164 Ma. Geochemical evidence suggests that the protolith of the biotite-plagioclase gneiss was a Late Carboniferous ferruginous-aluminous sedimentary deposit formed in a continental island arc setting. The Late Carboniferous quartz diorites are characterized by high Sr/Y ratios (30–113), elevated Mg# values (43–54) and high MgO contents (1.80–5.67 wt%), coupled with depleted heavy rare earth elements (HREE), indicative of adakitic affinities. The Early Permian gneissic granodiorite also shows elevated Sr/Y ratios (71–93) and depleted HREE, suggesting adakitic affinities, but with lower MgO contents (1.41–1.65 wt%) than the quartz diorites. The protolith of the amphibolite is interpreted as basaltic volcanic rock formed during the Early Permian. The geochemical compositions of the quartz diorites point to derivation from magmas generated by partial melting of a subducted oceanic slab. In contrast, the low-MgO gneissic granodiorite is likely derived from magmas produced by partial melting of a thickened mafic lower continental crust. The basaltic protolith of the amphibolite is interpreted to have formed from the mantle wedge in a subduction setting. By integrating the new data with previous regional research, it is concluded that the initial collision between the Xing’an Block and the Songliao-Xilinhot Block occurred during the Early Permian, whereas the final amalgamation of the two blocks did not take place until after the Early Permian.