Tight reservoirs have poor physical properties: low permeability and strong heterogeneity, which makes it difficult to predict productivity. Accurate prediction of oil well production plays a very important role in the exploration and development of oil and gas reservoirs, and improving the accuracy of production prediction has always been a key issue in reservoir characterization. With the development of artificial intelligence, high-performance algorithms make reliable production prediction possible from the perspective of data. Due to the high cost and large error of traditional seepage theory formulas in predicting oil well production, this paper establishes a horizontal well productivity prediction model based on a hybrid neural network method (CNN-LSTM), which solves the limitations of traditional methods and produces accurate predictions of horizontal wells’ daily oil production. In order to prove the effectiveness of the model, compared with the prediction results of BPNN, RBF, RNN and LSTM, it is concluded that the error results of the CNN-LSTM prediction model are 67%, 60%, 51.3% and 28% less than those of the four models, respectively, and the determination coefficient exceeds 0.95. The results show that the prediction model based on a hybrid neural network can accurately reflect the dynamic change law of production, which marks this study as a preliminary attempt of the application of this neural network method in petroleum engineering, and also provides a new method for the application of artificial intelligence in oil and gas field development.

Porous carbon nanofibers doped with nickel (Ni) were successfully fabricated through electrospinning, carbonization, and CO₂ activation techniques using polyacrylonitrile (PAN) and petroleum pitch as carbon sources and nickel acetate as the dopant. During the activation process, Ni was reduced and dispersed in situ on the carbon matrix. The effects of Ni doping content on the morphology and structure of the carbon nanofibers were systematically investigated using SEM, TEM, XPS, XRD, Raman, and BET analyses. The experimental results revealed that the prepared materials had a hierarchically porous structure and that Ni nanoparticles played multiple roles in the preparation process, including catalyzing pore expansion and catalytic graphitization. However, particle agglomeration and fiber fracture occurred when the Ni content was high. In the adsorption/desorption experiments, the sample with 10 wt% Ni doping exhibited the highest specific surface area and micropore volume of 750.7 m²/g and 0.258 cm³/g, respectively, and had the maximum hydrogen storage capacity of 1.39 wt% at 298 K and 10 MPa. The analyses suggested that the hydrogen adsorption mechanism contributed to enhanced H₂ adsorption by the spillover effect in addition to physisorption.

In heavy oil thermal recovery processes, higher pressure usually leads to low dryness and expansion difficulty for the injected steam in thermal recovery processes, which will result in lower oil recovery and more carbon emissions. This paper proposed a new CO₂-inducing method to accelerate the steam chamber expansion, based on a core flooding experiment and numerical simulation. First, the CO₂ showed significant viscosity reduction at high pressure in the PVT test. In the core flooding experiment, the CO₂ provided strong flow conductivity in porous media for the thermal flooding, as the CO₂ pre-injection restrained the steam condensation. Using the CO₂-inducing method, CO₂ pre-injection before steam built a fast flow channel in a relatively higher permeability layer and reduced the thermal injection pressure by about 1.0~2.4 MPa. As a result, the steam overlap around the injection wells became slower and the gravity drainage process was able to heat and displace the heavy oil above the channel. Furthermore, the CO₂ gas trapped at the top reduced heat loss by about 12.4%. The field numerical simulation showed that this new method improved thermal recovery by 7.5% and reduced CO₂ emissions by about 18 million kg/unit for the whole process. This method changes the conventional thermal expansion direction by CO₂ inducing effect and fundamentally reduces heat loss, which provides significant advantages in low-carbon EOR.

The water drive reservoir in Shengli Oilfield has entered a stage of ultra-high water cut development, forming an advantageous flow channel for the water drive, resulting in the inefficient and ineffective circulation of injected water. Therefore, the distribution characteristics of water drive flow channels and their controlled residual oil in ultra-high water cut reservoirs are of great significance for treating water drive dominant flow channels and utilizing discontinuous residual oil. Through microscopic physical simulation of water flooding, color mixing recognition and image analysis technology were used to visualize the evolution characteristics of water flooding seepage channels and their changes during the control process. Research has shown that during the ultra-high water content period, the shrinkage of the water drive seepage channel forms a dominant seepage channel, forming a “seepage barrier” at the boundary of the dominant seepage channel, and dividing the affected area into the water drive dominant seepage zone and the seepage stagnation zone. The advantage of water flooding is that the oil displacement efficiency in the permeable zone is as high as 80.5%, and the remaining oil is highly dispersed. The water phase is almost a single-phase flow, revealing the reason for high water consumption in this stage. The remaining oil outside the affected area and within the stagnant flow zone accounts for 89.8% of the remaining oil, which has the potential to further improve oil recovery in the later stage of ultra-high water cut. For the first time, the redundancy index was proposed to quantitatively evaluate the control effect of liquid extraction and liquid flow direction on the dominant flow channels in water flooding. Experimental data showed that both liquid extraction and liquid flow direction can regulate the dominant flow channels in water flooding and improve oil recovery under certain conditions. Microscopic physical simulation experiments were conducted through the transformation of well network form in the later stage of ultra-high water content, which showed that the synergistic effect of liquid extraction and liquid flow direction can significantly improve the oil recovery effect, with an oil recovery rate of 68.02%, deepening the understanding of improving oil recovery rate in the later stage of ultra-high water content.

Non-Newtonian fluids may cause nonlinear seepage even for a single-phase flow. Through digital rock technologies, the upscaling of this non-Darcy flow can be studied; however, the requirements for scanning resolution and sample size need to be clarified very carefully. This work focuses on Bingham fluid flow in tight porous media by a pore-scale simulation on CT-scanned microstructures of tight sandstones. A bi-viscous model is used to depict the Bingham fluid. The results show that when the Bingham fluid flows through a rock sample, the flowrate increases at a parabolic rate when the pressure gradient is small and then increases linearly with the pressure gradient. As a result, an effective permeability and a start-up pressure gradient can be used to characterize this flow behavior. By conducting flow simulations at varying sample sizes, we obtain the representative element volume (REV) for effective permeability and start-up pressure gradient. It is found that the REV size for the effective permeability is almost the same as that for the absolute permeability of Newtonian fluid. The interesting result is that the REV size for the start-up pressure gradient is much smaller than that for the effective permeability. The results imply that the sample size, which is large enough to reach the REV size for Newtonian fluids, can be used to investigate the Bingham fluids flow through porous media as well.

Accurate characterization of the micro- and nano-pore radius values in a tight sandstone reservoir is the key work to reasonably evaluate reservoir properties. The previous exploration of pore-stitching methods is mainly based on the morphological extension of similar segments. However, few scholars compare and verify the image and non-image stitching methods, so they cannot clarify the application scope of different pore-stitching methods. In this study, the pore structures of eight selected tight sandstone samples were evaluated using high-pressure mercury injection, nuclear magnetic resonance, scanning electron microscope, and the helium porosity test. Then, the C-value fitting, interpolation fitting, and morphological fitting were used to establish high-pressure mercury injection and Nuclear Magnetic Resonance (NMR) pore distribution curves to evaluate the differences among the micro-nano-scale pore radius values determined by the three fitting methods. Finally, the pore radius distribution is extracted from the binary image of Scanning Electron Microscope (SEM). After correcting the helium porosity data, the application scope of different fitting methods is evaluated by using the mean standard deviation verification method, and the optimal solution of the stitching method of pore radius distribution in each application scope is found. Compared to other studies, this research demonstrated three relatively simple methods for the determination of the full range of pore size distributions, providing a reliable method to evaluate the prerequisites of the range of application. This study provides a new idea for the micro-nano-scale pore radius splicing method of a tight sandstone reservoir, and the research results can provide a reference for the actual reservoir evaluation of oil and gas fields.