Shaohua Li

Evaluating the productivity contribution of individual fracturing stages in deep coalbed methane (CBM) horizontal wells remains a critical challenge, hindering the optimization of stimulation designs. This study systematically integrates dual-phase (aqueous and gaseous) fracturing tracer monitoring with machine learning algorithms to address this issue. Based on large-scale field applications across ten deep CBM horizontal wells in the Changqing mining area of the Ordos Basin, comprising 132 monitored stages, quantitative production profile data were interpreted. Three distinct gas production archetypes—Homogeneous, Heel-Dominated, and Heterogeneous—were identified, each governed by specific geomechanical and stratigraphic controls. Pearson correlation analysis and Random Forest feature importance ranking were employed to decouple the hierarchical influence of geological parameters (Class I coal intersection length, trajectory position, coal thickness) and engineering parameters (proppant volume, pumping rate, fluid volume). A power-law correlation between Class I coal length and initial gas productivity was quantified (R² = 0.71). For the first time, an economically viable “differentiated fracturing scale window” tailored to coal petrophysical classes and wellbore trajectory positions was defined. Subsequently, a machine learning-assisted geology-engineering closed-loop optimization methodology was established, using tracer data as a dynamic feedback bridge to iteratively refine fracturing designs. This research provides a reliable technical approach and practical template for enhancing single-well productivity and recovery efficiency in deep unconventional gas reservoirs.

In response to the challenges in monitoring the production profile during the development of the Qingcheng shale oil field in the Changqing Oilfield, this study systematically investigates the application mechanism and practical effectiveness of Distributed Temperature Sensing (DTS) technology for dynamic monitoring in horizontal wells. By establishing a coupled model of fracture–matrix dual-porosity media flow and wellbore thermodynamics, which integrates mass, momentum, and energy conservation equations solved via the finite difference method, an interpretation method for the production profile based on the Joule–Thomson effect is proposed. The model was calibrated using shut-in temperature data and validated by comparing simulated temperature profiles with DTS measurements under constant-rate production. Field tests conducted in six horizontal wells in the Qingcheng oil field enabled the quantitative analysis of cluster-level production contributions along the horizontal section, with a water-producing zone localization accuracy of ±3.5 m. The results indicate that shale oil wells exhibit a non-uniform production characteristic of “high at the front and low at the rear” during the early production stage, where the production contribution from fully fractured segments can be up to 2.8 times that of adjacent segments. Inversion of the fiber-optic monitoring data reveals that differences in the conductivity of hydraulic fractures are the primary cause of flow heterogeneity. This research provides a theoretical foundation and technical support for the efficient development of shale oil, contributing to the transition of China’s continental shale oil development from “experience-driven” to “data-driven.”

An integrated analysis incorporating total organic carbon (TOC) content measurement, X-ray diffraction (XRD), scanning electron microscopy (SEM), and gas adsorption experiments was performed on core samples from Well FY1-4 of the upper-fourth Shahejie Formation (Es₄) in the Minfeng Sag. To address the lack of systematic research on the pore and fractal characteristics of organic-rich low-maturity shales in the Minfeng Sag (against the preponderance of studies on high-maturity shales), this study characterized the lithofacies, reservoir space and pore fractal features of the target low-maturity shale interval and clarified the sedimentary controls on lithofacies and key factors regulating pore fractal heterogeneity. The results reveal that the shale in the Es₄ of the study area exhibits low thermal maturity, with six distinct lithofacies identified. Organic-rich laminated calcareous shale lithofacies (RL-1) and organic-rich laminated calcareous/argillaceous mixed shale lithofacies (RL-2) represent the most favorable lithofacies, which are dominated by large mesopores and macropores. Their reservoir spaces were primarily composed of intergranular pores, intragranular pores, and organic pores, whereas the other lithofacies are dominated by small mesopores. The pore surface fractal dimension (D) was calculated using the Frenkel–Halsey–Hill (FHH) model based on low-temperature N₂ adsorption (LTNA) data. The meso-macropore system shows higher heterogeneity than the micropore system (D₂ > D₁). Both D₁ and D₂ exhibit a weak negative correlation with TOC and carbonate content and a positive correlation with clay content. In the initial depositional stage of the Es₄, the arid climate, weak terrigenous input, shallow lake depth, and high salinity resulted in the strongly reducing saline depositional environment with relatively low organic matter enrichment. As the climate became progressively humid in the middle and late stages, hydrodynamic conditions intensified, leading to a lithofacies transition from mixed shales to argillaceous calcareous shales. Increased TOC and carbonate contents reduce the pore fractal dimension of shale. Smaller fractal dimensions directly indicate a simple pore structure and regular pore surface in the shale oil reservoir of the Minfeng Sag, where reservoir space is dominated by large pores such as intercrystalline pores and dissolved pores. Such pore characteristics are more favorable for the enrichment of shale oil.

High sand-ratio and low-connectivity reservoirs are commonly developed in deep-water depositional environments. Well-developed muddy interlayers reduce reservoir connectivity and form multiple discrete sandbody units, thereby offering good potential for layered development. However, due to limited research and insufficient data, modeling such complex reservoir structures remains challenging. The existing multiple-point statistics (MPS) method can utilize limited training images for geological modeling, but under high sand-ratio conditions, it often produces models with excessively high connectivity, failing to accurately represent reservoir characteristics. To address this issue, this study proposes a self-attention-enhanced single-image generative adversarial network (SA-SinGAN). Based on the original 2D SinGAN, the method was extended to 3D modeling by incorporating a self-attention mechanism and trained layer by layer to capture multi-scale geological features. Experimental results show that the FID score of SA-SinGAN is 143.75, compared with 175.61 for the MPS method. In terms of average connectivity error, MPS yields 0.42, which is significantly higher than 0.13 for SA-SinGAN, while the average NTG error is similar. SA-SinGAN can more accurately reproduce the low-connectivity characteristics of reservoirs while maintaining randomness, outperforming MPS in modeling performance. This demonstrates the applicability of SA-SinGAN for modeling complex reservoirs.

High-resolution LiDAR-derived three-dimensional (3D) digital outcrop models are crucial for detailed geological analysis. However, their massive data volumes often exceed the rendering and memory capacities of standard computer systems, posing significant visualization challenges. Although Level of Detail (LOD) techniques are well-established in Geographic Information Systems (GISs) and computer graphics, they still require customized design to address the unique characteristics of geological outcrops. This paper presents an automated method for constructing and visualizing LOD models specifically tailored to high-resolution LiDAR outcrops. The workflow begins with segmenting the single-body model based on texture coverage, followed by building an adaptive LOD tile pyramid for each segment using a pseudo-quadtree approach. The proposed LOD construction method incorporates several innovative components: segmentation based on texture coverage, an adaptive LOD tile pyramid using a pseudo-quadtree, and a feature-preserving mesh simplification algorithm that includes vertex sharpness constraint and boundary freezing strategy to maintain critical geological features. For visualization, a dynamic multi-scale loading and rendering mechanism is implemented using an LOD index with the OpenSceneGraph (OSG) engine. The results demonstrate that the proposed method effectively addresses the bottleneck of rendering massive outcrop models. The models loading time and average memory usage were reduced by more than 90%, while the average display frame rate reached around 60 FPS. It enables smooth, interactive visualization and provides a robust foundation for multi-scale geological interpretation.

High-resolution LiDAR-derived 3D digital outcrop models are essential for detailed geological analysis, but their massive data volumes often exceed the rendering and memory capacities of standard computer systems, posing significant visualization challenges. Although Level of Detail (LOD) techniques are well-established in GIS and computer graphics, they are not directly suitable for the elongated, cliff-face morphology of typical outcrops. This paper presents an automated method for constructing and visualizing LOD models specifically tailored to high-resolution LiDAR outcrops. The workflow begins by segmenting the monolithic model based on texture coverage, followed by building an adaptive LOD tile pyramid for each segment using a pseudo-quadtree approach. A feature-preserving Quadric Error Metric (QEM) simplification algorithm, incorporating boundary freezing and fallback strategies, is employed to retain critical geological structures while significantly reducing geometric complexity. For visualization, a dynamic multi-scale loading and rendering mechanism is implemented using an LOD index with the OpenSceneGraph (OSG) engine. Results demonstrate that the proposed method effectively overcomes the bottleneck of rendering massive outcrop models, enabling smooth, interactive visualization and providing a robust foundation for multi-scale geological interpretation.

This study quantifies morphological features of global braided river deltas using Google Earth imagery, analyzing eight systems (e.g., Yukon–Kuskokwim, Poyang Lake, Lena River deltas). Methods include listwise deletion for missing data (retaining 87% of Poyang Lake delta samples) and sensitivity analysis (threshold changes ≤2.4%). Nonparametric tests (Kruskal–Wallis, H = 12.73, p = 0.005) show significant differences in bifurcation angles across deltas, with the wave-dominated Po River (59.2°) having an 18% higher 80% threshold the than tide-dominated Poyang Lake (50.1°, p = 0.003). Key quantitative results include the following: 1.65% of bifurcation angles cluster at 30–60°, differing from fan deltas (p < 0.01); wavelength–amplitude relationships are nonlinear (R² = 0.537–0.913), with positive slopes indicating a high sediment supply (e.g., Yukon–Kuskokwim) and negative slope channel avulsion (e.g., Poyang Lake); bifurcation spacing correlates with the sediment supply—54% of Poyang Lake spacings < 2000 m (dense networks) vs. 80% of Lena River spacings < 15,000 m (stable channels). The resulting dataset enables global, remote-sensing-based comparisons, providing thresholds for sedimentary modeling and reservoir prediction. Moderate missing data (≤13%) minimally affect results, though high-missingness cases need further analysis. This study replaces empirical rules with statistical validation, showing that morphometric differences reflect depositional dynamics, which are critical for reservoir heterogeneity assessments.

The variogram function is a crucial tool for quantifying spatial correlation and a key component of Kriging interpolation, directly influencing the accuracy of interpolation results. However, traditional variogram fitting methods, such as least squares and weighted polynomial regression, often face limitations when handling complex or non-stationary spatial structures, which can lead to unstable fitting outcomes. To address these challenges, this paper proposes an automatic fitting method for experimental variogram functions based on deep learning. The variogram fitting process is inherently a nonlinear optimization problem, where the goal is to optimize the alignment between experimental and theoretical variogram functions. In this study, we construct a deep learning model designed to capture the nonlinear characteristics of experimental variograms with the objective of minimizing fitting errors and improving both the accuracy and stability of the fitting process. Extensive experimental results demonstrate that the proposed neural network model can effectively learn to fit a variety of experimental variograms and accurately represent different theoretical variogram functions. Compared to traditional methods, this approach consistently achieves superior fitting performance across multiple datasets, validating its effectiveness and robustness under diverse conditions. The stronger generalization capability and improved accuracy of the proposed method significantly enhance the automatic fitting of experimental variogram functions.

The target reservoir is a typical blocky bottom water reservoir. There are several interlayers of varying scales inside it, which impact the characteristics of the oil production and water breakthrough curves of the wells in the water flooding development oilfield, resulting in strong heterogeneity within the reservoir. The Sangtamu Oilfield has an average well spacing of approximately 600 m, causing a challenge in accurately identifying the range of small-scale interlayer spreading. This, in addition, challenges the subsequent 3D geological modeling process. This paper proposes a method to accurately characterize the size of various interlayers in a reservoir by using production dynamics data from bottom water reservoirs to invert the interlayer scale. By classifying the water breakthrough curve styles derived from theoretical testing, the range of reservoir internal interlayer scales is compared and inverted based on the actual water breakthrough styles of wells in the Sangtamu oil field. In the process, many interlayer scales derived from various types of geologic data acquisition were combined to form a quantitative geologic knowledge database of interlayers that synthesizes both dynamic and static data. This method has been applied in the Tarim Basin’s Sangtamu Oilfield, where the reservoir numerical simulation history was used to fit the model. This resulted in an overall fitting coincidence rate of 96% in the Sangtamu Oilfield and a single-well fitting coincidence rate of over 90% in well LN14. This method offers a new viewpoint on characterizing the interlayer’s scale in the area of the non-dense well network.