Well-logging is an important formation characterization and resource evaluation method in oil and gas exploration and development. However, there has been a shortage of well-logging data because Well-logging can only be measured by expensive and time-consuming field tests. In this study, we aimed to find effective machine learning techniques for well-logging data prediction, considering the temporal and spatial characteristics of well-logging data. To achieve this goal, the convolutional neural network (CNN) and the long short-term memory (LSTM) neural networks were combined to extract the spatial and temporal features of well-logging data, and the particle swarm optimization (PSO) algorithm was used to determine hyperparameters of the optimal CNN-LSTM architecture to predict logging curves in this study. We applied the proposed CNN-LSTM-PSO model, along with support vector regression, gradient-boosting regression, CNN-PSO, and LSTM-PSO models, to forecast photoelectric effect (PE) logs from other logs of the target well, and from logs of adjacent wells. Among the applied algorithms, the proposed CNN-LSTM-PSO model generated the best prediction of PE logs because it fully considers the spatio-temporal information of other well-logging curves. The prediction accuracy of the PE log using logs of the adjacent wells was not as good as that using the other well-logging data of the target well itself, due to geological uncertainties between the target well and adjacent wells. The results also show that the prediction accuracy of the models can be significantly improved with the PSO algorithm. The proposed CNN-LSTM-PSO model was found to enable reliable and efficient Well-logging prediction for existing and new drilled wells; further, as the reservoir complexity increases, the proxy model should be able to reduce the optimization time dramatically.

The efficient development of tight gas reservoirs is significantly enhanced by multi-stage and multi-cluster fracturing techniques in conjunction with horizontal well technology, leading to substantial increases in reservoir drainage volume and individual well productivity. This study presents a tailored fine-fracturing approach for horizontal wells in tight gas reservoirs, supported by a gas–water two-phase numerical simulation model. Utilizing the orthogonal experimental design method, we simulated and optimized various fracture distribution schemes to refine fracturing parameters for maximum efficiency. The optimization was further validated through a comparison with actual well completion and development dynamics. The quantitative results highlight the optimal fracture distribution for horizontal wells, with a horizontal section length of 1400 to 1600 m and 14 to 16 fracturing stages. The pattern features a “dense at both ends and sparse in the middle” strategy, with stage spacing of 80 to 110 m, and a “longer in the middle and shorter at both ends” fracture half-length of 100 to 140 m, achieving a fracture conductivity of 30 μm²·cm. To ensure the economic feasibility of the proposed fracturing strategy, we conducted an economic evaluation using the net present value (NPV) method, which confirmed the robustness of the optimization outcomes in terms of both technical performance and economic viability. The reliability of these optimization outcomes has been confirmed through practical application in the development of horizontal wells in the study area. This research approach and methodology can provide theoretical guidance for the design of hydraulic fracturing operations and the integration of geological and engineering practices in similar unconventional oil and gas reservoirs.

The fracture-cavity combination structure between wells in fracture-cavity reservoirs is complex and changeable. Reliably identifying and quantitatively characterizing the fracture-cavity combination structure between wells has become an important prerequisite for flow channel adjustment in fracture-cavity reservoirs after water channeling and flooding. Aiming at the problems that it is difficult for the existing carving technology to characterize the flow characteristics of the injected fluid in the interwell fracture-cavity composite structure during the production process, and it is difficult for the existing interwell tracer proxy model to consider the specific fracture-cavity composite structure, this paper proposes a quantitative interpretation model for interwell tracers in fracture-cavity reservoirs with different architectures. Taking the Tahe fracture-cavity reservoir as the object, the matching relationship between the interwell fracture-cavity structure and the tracer curve was analyzed, and the tracer curve characteristics of five types of fracture-cavity structures were clarified. Considering the basic idea of tracing, a unified quantitative interpretation model of tracers under different fracture-cavity configurations based on branched flow channels and karst caves was deduced and established, and the input parameters required to apply the model, the parameters obtained directly by fitting, and further expandable calculated parameters were clarified. The interpretation model was used to fit, quantitatively interpret, and verify the reliability of the tracer curves of three wells in group TK411 of fracture-cavity unit S48 in the fourth area of Tahe Oilfield. The results show that the tracer curve fitting effect of each well was good, and the average relative error between the total flow rate explained by the tracer and the daily water production during the tracer monitoring period in the mine was only 3.02%, which effectively shows that the applicability and reliability of the quantitative interpretation model are established. The research results provide an effective way to apply tracer data in deep mining while improving the quantitative characterization ability of interwell tracer monitoring in fracture-cavity reservoirs.

Rapid and effective identification of interwell fracture-cavity composite structures is a necessary prerequisite for a detailed and in-depth understanding of interwell connectivity in fracture-cavity reservoirs. Current identification methods and technologies have the problems of being large-scale and low-resolution; in view of these problems, a method is proposed for rapidly identifying interwell fracture-cavity combination structures using tracer-curve morphological characteristics (peak number and characteristics of two wings). Based on concentration models of tracer curves for an interwell single fracture/pipe/cavity, the morphological characteristics of tracer curves were researched in five different series-parallel combination modes consisting of fractures, pipes, and cavities. The tracer curves of fracture-cavity reservoirs are categorized into three types: single sharp peak, single slow peak, and multipeak. Furthermore, a matching relationship between different fracture-cavity combination structures and the morphological characteristics of tracer curves is clarified. The single-sharp-peak curve with basically symmetrical wings reflects that of an interwell single fracture/pipe; the single-slow-peak curve with a steep ascending branch and a slow descending branch (obvious trailing phenomenon) reflects that of an interwell single cavity or fracture/pipe series cavity; the multipeak curve reflects that of an interwell multifracture/pipe/cavity in parallel; according to the flow difference of each branch flow channel, they can be divided into independent multipeak and continuous multipeak forms. Taking tracer monitoring results from a well group in the Tahe oilfield as an example, field application analysis and verification were carried out. The results show that this method is simple and reliable and can provide a fast and effective means for identifying interwell fracture-cavity combination structures. Meanwhile, the research results can lay a foundation for quantitative interpretation modeling of interwell tracers in fracture-cavity reservoirs considering fracture-cavity configuration.

This study addresses the unclear mechanisms by which preferential flow channels (PFCs), formed during long-term waterflooding, affect nano-gel microsphere (NGM) flooding efficiency, utilizing CMG reservoir numerical simulation software. A dynamic evolution model of PFCs was established by coupling CROCKTAB (stress–porosity hysteresis) and CROCKTABW (water saturation-driven permeability evolution), and the deep flooding mechanism of NGMs (based on their gel properties such as swelling, elastic deformation, and adsorption, and characterized by a “plugging-migration-replugging” process) was integrated. The results demonstrate that neglecting PFCs overestimates recovery by 8.7%, while NGMs reduce permeability by 33% (from 12 to 8 mD) in high-conductivity zones via “bridge-plug-filter cake” structures, diverting flow to low-permeability layers (+33% permeability, from 4.5 to 6 mD). Field application in a Chang 6 tight reservoir (permeability variation coefficient 0.82) confirms a >10-year effective period with 0.84% incremental recovery (from 7.31% to 8.15%) and favorable economics (ROI ≈ 10:1), providing a theoretical and engineering framework for gel-based conformance control in analogous reservoirs.