With the popularity of deep learning (DL), more and more studies are focusing on replacing time-consuming numerical simulations with efficient surrogate models to predict the production of multi-stage fractured horizontal wells. Previous studies on constructing surrogate models for the prediction of the production of fractured horizontal wells often relied on directly applying existing deep learning architectures without incorporating physical constraints into the model. When dealing with the large number of variables necessary for characterizing the properties of fractures, the input variables of proxy models are often oversimplified; meanwhile, lots of physical information is lost. Consequently, predictions are sometimes physically inconsistent with the underlying principles of the domain. In this study, by modifying the traditional Seq2Seq (LSTM–LSTM) deep learning architecture, a physics-informed encoder–decoder (PIED) architecture was developed to surrogate the numerical simulation codes for predicting the production of horizontal wells with unequal-length intersecting hydraulic fractures on a 2D plane. The encoder is a LSTM network, and the decoder consists of LSTM and fully connected layers. The attention algorithm is also applied in the Seq2Seq architecture. The PIED model’s encoder is capable of extracting the physical information related to fractures. And the attention module effectively passes on the most relevant physical information related to production to the decoder during the training process. By modifying Seq2Seq architecture, the decoder of the PIED incorporates the intermediate input, which is the constant production time, along with the extracted physical information to predict production values. The PIED model excels in extracting sufficient physical information from high-dimensional inputs while ensuring the integrity of the production time information. By considering the physical constraints, the model predicts production values with improved accuracy and generalization capabilities. In addition, a multi-layer perceptron (MLP) which is broadly used as a proxy model; a regular Seq2Seq model (LSTM–Attention–LSTM); and the PIED were compared via a case study, and their MAE values were shown to be 241.76, 184.07, 168.81, respectively. Therefore, the proposed model has higher accuracy and better generalization ability. In the case study, a comparative experiment was conducted by comparing LSTM–MLP (with an MAE of 221.50) and LSTM–LSTM to demonstrate that using LSTM as the decoder structure is better for predicting production series. Moreover, in the task of predicting production sequences, LSTM outperforms MLP. The Seq2Seq architecture demonstrated excellent performance in this problem, and it achieved a 48.4% reduction in MSE compared to MLP. Meanwhile, the time cost for build datasets was considered, and the proposed model was found to be capable of training in a small dataset (e.g., in the case study, 3 days were used to generate 450 samples for training.); thus, the proposed model has a certain degree of practicality.

Hydraulic fracturing is a complex nonlinear hydro-mechanical coupled process. Accurate numerical simulation is of great significance for reducing fracturing costs and improving reservoir development benefits. The aim of this paper is to propose an efficient numerical simulation method for the fracturing-to-production problem under a unified framework that has good convergence and accuracy. A hydro-mechanical coupled fracturing model (HMFM) is established for poroelastic media saturated with a compressible fluid, and the local characteristics of the physical field are fully considered. Each fracture is explicitly characterized using the discrete fracture model (DFM), which can better reflect the physical characteristics near fractures. Based on the extended finite element method (XFEM) and the Newton–Raphson method, a fully coupled approach named Unified Extended Finite Element (UXFEM) is developed, which can solve the nonlinear system of equations that describe the solution under a unified framework. UXFEM can accurately capture the local physical characteristics of different physical fields on the orthogonal structured grids. It realizes the grid-fracture decoupling, and fractures can propagate in any direction, which shows greater flexibility in simulating fracture propagation. The fully coupled approach can better reflect the essential relationship between pressure, stress, and fracture, which is beneficial to studying hydro-mechanical coupled problems. To validate the UXFEM, UXFEM is compared with the classical KGD model, analytic solution, and COMSOL solution. Finally, based on UXFEM, the interference phenomenon and fracturing-to-production study are carried out to prove the broad practical application prospect of this new fully coupled approach.

The poroelastodynamic failure of a wellbore due to periodic loading during drilling is an unsolved problem. The conventional poroelastic method to calculate the stress distribution around wellbore is for static loading cases and cannot be used for short-time dynamic-loading cases which result in wave propagation in the formation. This paper formulates a poroelastodynamic model to characterize dynamic stress and pressure wave due to periodic loadings and to analyze the transient failure of the suddenly drilled wellbore in a non-hydrostatic stress field. The fully coupled poroelastodynamic model was developed based on the equations of motion, fluid flow and constitutive equations to reflect stress and pressure waves that resulted from a periodic stress perturbation at the wellbore surface. The model was analytically solved by means of field expansions of the solutions, by performing a Laplace transform as well as some special techniques. Simulation results show that the pressure and stress responses inside the formation resemble a damped oscillator where the amplitude decays as the distance to wellbore increases. Especially the potential shear failure zone around the wellbore was computed and plotted. Influences of poroelastic parameters, in-situ stress and periodic load parameters on the shear failure responses were analyzed in a detailed parametric study, and the results provide fundamental insights into wellbore stability maintenance in different reservoirs.

As shale gas development gradually advances to a deeper level, the economic exploitation of deep shale gas has become one of the key technologies for sustainable development. Large-scale, long-term and effective hydraulic fracturing fracture networks are the core technology for achieving economic exploitation of deep shale gas. Due to the high-pressure and high-temperature characteristics of deep shale gas reservoirs, traditional seepage models cannot effectively simulate gas flow in such environments. Therefore, this paper constructs a fluid–solid–thermal coupling model, considering the creep characteristics of deep shale, the effects of proppant embedment and deformation on fracture closure, and deeply analyzes the effects of proppant parameters on the shale gas production process. The results show that factors such as proppant concentration, placement, mechanical properties and particle size have a significant effect on fracture width, fracture surface seepage characteristics and final gas production. Specifically, an increase in proppant concentration can expand the fracture width but has limited effect on increasing gas production; uneven proppant placement will significantly reduce the fracture conductivity, resulting in a significant decrease in gas production; proppants with smaller sizes are more suitable for deep shale gas fracturing construction, which not only reduces construction costs but also improves gas seepage capacity. This study provides theoretical guidance for proppant optimization in deep shale gas fracturing construction.

This study investigates Cambrian and Sinian carbonate outcrops in the Tarim Basin using 19 stratigraphically diverse rock samples. Through integrated X-ray diffraction mineralogical analysis, triaxial compression testing, and Brazilian splitting experiments, we systematically characterized rock mechanical properties and their correlations with microscopic mineral constituents. Key findings demonstrate remarkably distinct mechanical properties across formations: vuggy dolomites from the Xiaqiulitage formation exhibit the lowest compressive strength (minimum 200.0 MPa) and tensile strength (3.85 MPa), while the Yuertusi formation’s Y5 layer dolomites achieve exceptional tensile strength (21.69 MPa). Mineral composition fundamentally controls rock strength: dolomite or quartz concentrations exceeding 90% significantly enhance strength, whereas calcareous minerals (calcite, fluorapatite) degrade mechanical integrity. Most specimens display pronounced brittle failure characteristics; uniquely, basal dolostones of the Awatage formation exhibit distinctive plastic deformation. This research elucidates the synergistic effects of tectonic history, mineral assemblages, and microtextural attributes on rock mechanical behavior, providing critical theoretical underpinnings for deep carbonate reservoir development in overpressured basins.

Ultra-deep dolomite formations exhibit pronounced mechanical heterogeneity gov-erned by mineral composition, inter-particle cementation, and observation scale, com-plicating their mechanical characterization. Because intact downhole cores are diffi-cult to obtain, a mesoscopic numerical framework that reproduces and classifies such behavior is needed. A two-dimensional mesoscopic model of Cambrian ultra-deep do-lomite from the Tarim Basin was developed using the combined finite-discrete element method, incorporating mineral composition from X-ray diffraction and scanning elec-tron microscopy with energy-dispersive spectroscopy. A Weibull-distribution discreti-zation scheme was introduced to represent matrix and cementation heterogeneity separately. Sensitivity analyses examined three factor groups: mineral composition and spatial distribution, matrix and cementation shape parameters, and mesh size as the observation scale. Accordingly, a three-level graded characterization framework was proposed, in which the compositional level governs strength magnitude, the structural level governs the elastic modulus, pre-peak nonlinearity and failure mode, and the scale level governs response representativeness and convergence. Quantitative grading indicators and threshold ranges were extracted for each level. Triaxial valida-tion against laboratory experiments at confining pressures of 0 to 220 MPa confirmed strength and stress-strain response consistent with the proposed grading. The model and framework offer a transferable methodological reference for the mesoscopic me-chanical evaluation of ultra-deep heterogeneous dolomite.