Accurate production prediction is crucial in the field of reservoir management and production optimization. Traditional models often struggle with the complexities of nonlinear relationships and high-dimensional data, which hinders their ability to capture the variability of the production process efficiently and results in time-consuming calculations. To overcome these limitations, this paper introduces an innovative proxy modeling technique employing a convolutional spatiotemporal neural network. This method utilizes convolutional neural networks to extract spatial features from high-dimensional data, while the Transformer is used to model and predict complex temporal dynamics in production. To validate the effectiveness of the proposed proxy model, two case studies involving four injection and nine production wells within two-dimensional (2D) and three-dimensional (3D) non-homogeneous reservoirs were conducted, with the R2 coefficient serving as the primary evaluation metric. As the number of training iterations and data volume increase, the proxy model demonstrates rapid convergence. In tests conducted on the 2D and 3D datasets, the average R2 value exceeded 0.96 and 0.94. These results confirm the accuracy and stability of the proxy model. It also shows that the proxy model can accurately describe the geological and fluid seepage characteristics of the reservoir, which in turn can achieve a highly accurate match with the real data. In addition, the computational time is reduced by two orders of magnitude compared to traditional models. Compared with the long short-term memory method, the accuracy of the prediction results is increased by 30%, which greatly enhances efficiency and accuracy. To some extent, the presented proxy model can provide some guidance for the efficient history match of production data.

To obtain the maximum field oil recovery (FOR) and CO₂ sequestration ratio (CSR), it is imperative to optimize the CO₂ injection and liquid production rate. However, previous studies ignore the geomechanical risks indeed. Therefore, a hybrid optimization framework was designed that combines artificial intelligence methods (Support Vector Regression with the Gaussian kernel, Gaussian-SVR or Long Short-Term Memory, LSTM) and multi-objective optimization algorithms (multiple objective particle swarm optimization, MOPSO or Non-dominated Sorting Genetic Algorithm II, NSGA-II) to find the optimal CO₂ injection and production strategies under different water cut. With this framework, the largest oil recovery and CO₂ storage under the lowest fault slip displacement (FSD) can be obtained simultaneously. In this framework, Latin hypercube sampling (LHS) is used to produce the samples for training and testing for cases with water cut 0.7, 0.8, 0.9 and 0.95, and the corresponding results are obtained from numerical simulations. Thus, Gaussian-SVR and LSTM are trained as the proxy model to substitute the numerical simulator. Thus, the MOPSO and NSGA-II are utilized to determine the Pareto Front of the optimum result and work schedules. A synthetic case reservoir model with high-water cut and one fault is employed to test the robustness of this framework. The results show that compared with FOR and CSR, due to the serious nonlinearity, the training and prediction of FSD with the proxy model are not very good. The prediction errors increase with the water cut, and when the field water cut is larger than 0.9, the practical requirements (± 20% errors) are not yet met. In general, the performance of proxy model with LSTM is superior to the Gaussian-SVR. The solutions obtained from the Pareto optimal set for the NSGA-II algorithm exhibit faster convergence, better superiority and reliability than MOPSO. As the rise of water cut, the optimal average field gas injection rate (FGIR) decreases, while the average field liquid production rate (FLPR) increases. The novelty of this work mainly lies in the consideration of fault slip during CO₂ injection for multi-objective optimization in high-water cut oil reservoirs, which can provide some guidance for the design of schemes.

With the deepening of exploration and development of tight sandstone gas reservoirs, the remaining recoverable reservoirs gradually become thinner with the vertical stratigraphic structure. The geomechanical properties become complex, and development based on conventional hydraulic fracturing methods often leads to serious problems, such as difficult control of fracture height, penetrating interlayers, too short fracture length, and inadequate proppant filling. In view of the above problems, we conducted a numerical investigation on a hydraulic fracturing scheme in a multi-layered and thin tight sandstone gas reservoir. According to the dataset from wells in a real gas reservoir in China’s Ordos Basin, the relevant geomechanical characteristics of the gas layers, together with the interlayers in the main production interval, were obtained, based on which, a fine numerical model was developed. By using the PL3D fracture propagation algorithm, a 3D hydraulic fracture propagation model was produced, and then using microseismic monitoring and production data matching, a high-precision hydraulic fracture model of the multi-layered and thin tight sandstone gas reservoir was obtained. On this basis, the influence of different geomechanical parameters and fracturing operational parameters on hydraulic fracture propagation was analyzed. Finally, an optimized hydraulic fracturing scheme that fitted the characteristics of the multi-layered and thin tight sandstone gas reservoir was proposed. Using a typical reservoir example, the optimized scheme enabled control of the fracture height in thin layers and the creation of long fractures with better proppant filling, so that the productivity of the fracture was significantly improved.

To initially improve the gas production rate and shorten the payback period for tight gas reservoirs, the multiple-fractured horizontal well (MFHW) model is always applied. However, in the late stages of exploitation, it is difficult to adopt reasonable measures for enhanced gas recovery (EGR), particular for continental sedimentary formation with multiple layers, and efficient strategies for EGR in this type of gas field have not yet been presented. Therefore, in this paper, a typical tight gas reservoir in the late stages of exploitation, the Denglouku gas reservoir in Changling gas field, in which MFHWs were utilized and contributed to the communication of the higher Denglouku formation (0.34 mol% CO₂) and lower Yingcheng formation (27 mol% CO₂) during hydraulic fracturing, is studied comprehensively. Firstly, alongside the seismic, logging, drilling and experimental data, 3D geological and numerical simulation models are developed. According to the differences in CO₂ mole fractions for different formations, the gas production rate of MFHWs produced from Denglouku formation is accurately calculated. Then, the well gas production rate (WGPR) and the well bottom-hole pressure (WBHP) history are matched with the calculated values, and thus the types of remaining gas are provided through the fine reservoir description. Finally, in a combination of gas recovery and economics, the optimal infill well type and the adjustment scheme are determined. The results show that there are three main categories of remaining gas, which are areal distribution, abundant points, and marginal dispersion, and the ratios of reaming gas reserve for these three types are 80.3%, 4.2%, and 15.5%, respectively. For the tight gas reservoir developed by MFHWs with parallel and zipper patterns, the best infilling well type is the vertical well. The combination of patching holes, sidetracking, infilling and boosting can obtain the highest gas recovery, while the scheme with patching holes and sidetracking has the best economic benefits. To balance the gas recovery and economics, the measurement of patching holes, sidetracking and infilling with vertical wells is utilized. In the final production period, compared with the basic schemes, the gas recovery can increase by 5.5%. The primary novelty of this paper lies in the determination of the optimal infilling well types and its presentation of a comprehensive adjustment workflow for EGR in tight gas reservoirs. The conclusions in this paper can provide some guidance for other similar tight gas reservoirs developed with MFHWs in the later period.

CO₂ flooding technology is an effective technical means for tertiary oil recovery in high water cut reservoirs. Using an existing well pattern to inject CO₂ is an economical and feasible development method under low contemporary oil prices. Although the particle migration caused during the displacement process will block the pore throat of the rock, the injection of CO₂ can effectively reduce the CO₂ injection pressure, and the dissolution that occurs during the injection process improves the seepage capacity of the fluid as a whole. In this study, assessments of particle migration and plugging after CO₂ flooding and evaluations of rock dissolution by CO₂-aqueous solution were carried out, and the variation characteristics of the relative permeability curve before and after displacement were evaluated, further explaining the influence of CO₂ flooding on reservoir seepage capacity. The results show that, in the process of CO₂ injection, the injection pressure decreases under different injection rates. After CO₂ flooding under oil conditions, the core permeability loss was 43.8%, and after cleaning with toluene and anhydrous ethanol, the core permeability recovery was 24.3%, indicating that the particles were bound by crude oil during the migration process and accumulated into clusters, resulting in blockages. Although there was a certain degree of blockage, the CO₂-aqueous solution mainly reacted with chlorite and released Ca²⁺, Mg²⁺, and Fe²⁺ ions. The concentration of Ca²⁺ increased by 157.43%, the concentration of Mg²⁺ decreased by 43.33%, and the concentration of Fe²⁺ decreased by 50.47%, indicating that, although the generated MgCO₃ and Fe₂O₃ blocked the pore throat of the rock, the overall dissolution effect of CO₂ on CaCO₃ was stronger, and the overall seepage capacity of the fluid was improved. To a certain extent, this technique can improve the water injection capacity and the effect of subsequent water flooding.

The prediction of production decline rate in the development of offshore high water-cut reservoirs predominantly relies on the traditional Arps decline curves. However, the solution process is complex, and the interpretation efficiency is low, making it difficult to meet the demand for rapid prediction of production decline rates. To address this, this paper first identifies the key influencing factors of production decline rate through comprehensive feature engineering. Subsequently, it proposes a novel prediction method for the production decline rate in offshore high water-cut reservoirs by integrating Moth–Flame Optimization with Extreme Gradient Boosting Tree (MFO-XGBoost). This method utilizes seven dynamic and static influencing factors, namely vertical thickness, perforated thickness, shale content, permeability, crude oil viscosity, formation flow coefficient, and well deviation angle, to predict the production decline rate. The forecasting outcomes of the MFO-XGBoost method are then compared with those of standard RF, standard DT, the standalone XGBoost model, and the calculated results from the exponential decline model. Additionally, the forecasting capability of the MFO-XGBoost method is benchmarked against Particle Swarm Optimization–XGBoost (PSO-XGBoost) and Bayesian Optimization–XGBoost methods for predicting the production decline rate in offshore high water-cut reservoirs. The findings from the experiments show that the MFO-XGBoost method can achieve accurate prediction of the production decline rate in offshore high water-cut reservoirs, with a coefficient of determination (R²) reaching 0.9128, thereby providing a basis for strategies to mitigate the production decline rate.