A gas–water two-phase fluid is present in a reservoir before a water-flooded sandstone gas reservoir is rebuilt. Therefore, in the process of injection and production of the rebuilt underground gas storage, the injected gas is easily blocked by the water in the pores, and the efficiency is low, resulting in a significant loss of gas. The study completely utilizes the geological data and dynamic operation monitoring data of a water-flooded sandstone underground gas storage and clarifies the rule of the gas–water three-phase seepage in a high-intensity injection–production process. Moreover, the main control factors of the low efficiency of this type of underground gas storage are clarified. The lost gas generated in the injection–production process is described from two aspects: microcosmic experiment and macroscopic law analysis. The type, mechanism, and occurrence state of the loss gas are clearly defined, its main type is “water trapped gas”, it formed when the gas rushing into the water area under high pressure and surrounded by water, and its occurrence of this kind of lost gas is mainly sporadic or continuous free gas. A gas–water two-phase mathematical model that can simulate the high-intensity injection–production process is set up according to the experimental result, this model is used to simulate the operation process of the Ban 876 underground gas storage. Based on the simulation results, the gas–water macroscopic movement rule and macroscopic accumulation mode of the lost gas are defined, and then the collection area of the lost gas is predicted and quantitatively described. The calculation results show that the lost gas in one cycle is about 775 × 10⁴ m³, which are mainly concentrated in the inner of the gas-water transition zone. According to the numerical simulation result, six new wells have been designed to develop its internal lost gas, they all have good predictions, can increase the working gas volume of 3000 × 10⁴ m³ and reduce the single cycle lost gas by 50%, which is only 326 × 10⁴ m³. This provides guidance for the expansion and exploitation of the same type of water-flooded sandstone underground gas storage.

This study uses advanced deep learning techniques to present an efficient and data-driven method for predicting the Hydrocarbon Reservoir Quality Index (HRQI) in deepwater carbonate systems. Traditional approaches like core sampling and Nuclear Magnetic Resonance logging are often costly and limited in data coverage, particularly in older wells. To overcome these limitations, the study utilizes 5112 petrophysical log samples from three wells located in a Middle Eastern deepwater carbonate reservoir. Three sequential deep learning models—Recurrent Neural Network and Gated Recurrent Unit—were developed and optimized using the Adam algorithm. The Adam-LSTM model outperformed the others, achieving a Root Mean Square Error of 0.009 and a correlation coefficient (R²) of 0.9995, indicating excellent predictive performance. Feature analysis revealed RHOB, DT, NPHI, and Vp as the most significant inputs for predicting HRQI. The model enhances reservoir characterization by capturing complex lithological patterns, supporting better zonation and well placement. Its scalability and cost-effectiveness make it especially useful in settings with limited data availability. This research highlights the practical value of integrating artificial intelligence into reservoir evaluation, offering a powerful tool for improving decision-making in deepwater hydrocarbon exploration and development.

Understanding water saturation levels in tight gas carbonate reservoirs is vital for optimizing hydrocarbon production and mitigating challenges such as reduced permeability due to water saturation (Sw) and pore throat blockages, given its critical role in managing capillary pressure in water drive mechanisms reservoirs. Traditional sediment characterization methods such as core analysis, are often costly, invasive, and lack comprehensive spatial information. In recent years, several classical machine learning models have been developed to address these shortcomings. Traditional machine learning methods utilized in reservoir characterization encounter various challenges, including the ability to capture intricate relationships, potential overfitting, and handling extensive, multi-dimensional datasets. Moreover, these methods often face difficulties in dealing with temporal dependencies and subtle patterns within geological formations, particularly evident in heterogeneous carbonate reservoirs. Consequently, despite technological advancements, enhancing the reliability, interpretability, and applicability of predictive models remains imperative for effectively characterizing tight gas carbonate reservoirs. This study employs a novel data-driven strategy to prediction of water saturation in tight gas reservoir powered by three recurrent neural network type deep/shallow learning algorithms—Gated Recurrent Unit (GRU), Recurrent Neural Networks (RNN), Long Short-Term Memory (LSTM), Support Vector Machine (SVM), K-nearest neighbor (KNN) and Decision tree (DT)—customized to accurately forecast sequential sedimentary structure data. These models, optimized using Adam's optimizer algorithm, demonstrated impressive performance in predicting water saturation levels using conventional petrophysical data. Particularly, the GRU model stood out, achieving remarkable accuracy (an R-squared value of 0.9973) with minimal errors (RMSE of 0.0198) compared to LSTM, RNN, SVM, KNN and, DT algorithms, thus showcasing its proficiency in processing extensive datasets and effectively identifying patterns. By achieving unprecedented accuracy levels, this study not only enhances the understanding of sediment properties and fluid saturation dynamics but also offers practical implications for reservoir management and hydrocarbon exploration in complex geological settings. These insights pave the way for more reliable and efficient decision-making processes, thereby advancing the forefront of reservoir engineering and petroleum geoscience.

The strata of the Shan 2 Member of the Shanxi Formation of Permian(S2)in the LQ Gas Field in the northeast of the Ordos Basin restrict the study of sand body distribution and lateral distribution. The accuracy of sand body prediction is reduced, which greatly affects the exploration and development of oil and gas. In order to accurately characterize the transverse distribution of sand bodies and improve the prediction accuracy of sand bodies, S2 is divided by the sequence stratigraphy method. Research based on the traditional stratigraphic division method includes seismic, logging, and core data. The logging curve is processed by INPEFA and the wavelet transform to reduce artificial division interference. The range of the S2 section is limited by the seismic interpretation surface. The maximum entropy spectral analysis and the wavelet transform are used for fine division, and S2 is divided into 1 long-term base level cycle, 6 medium-term base level cycles, and 11 short-term base level cycles. The fine stratigraphic division improves the exploration accuracy of oil and gas reservoirs.

Water alternating gas injection is considered as efficient enhanced oil recovery techniques to increase oil cumulative production rate. The objective of this comprehensive study is to measure the profound impact of different permeability heterogeneities on the ultimate recovery factor of miscible alternative injection of formation brine–carbon dioxide processes. To evaluate the results of each heterogeneity types, three heterogeneity levels of homogeneous, layered, and composite were taken into consideration, and the ultimate recovery factor was explicitly investigated for each case. The ultimate oil recovery factor for the homogeneous sample had witnessed the highest value at the permeability ratio of 1 (about 86.4%). According to the results of this experiment, miscible water alternating gas (henceforth; WAG) injectivity as a secondary enhanced oil recovery (henceforth; EOR) mode provided higher recovery factor due to the higher interfacial contact among in-situ oil, carbon dioxide and higher values of permeability ratios have led to lower ultimate recovery factor. After the breakthrough, the impact of permeability sequence and heterogeneity had caused to differ the results of recovery factor for each scenario. It was about 90.4% for the homogeneous sample, 86.5% and 83.8% for the low-high-low and high-low-high respectively. Consequently, in the absence of crossflow, the recovery factor had provided higher values rather than crossflow presence.