After a shale gas reservoir is fractured, hydraulic fractures interweave with natural fractures to form a fracture network. Numerical simulation based on the continuous fracture network model is a relatively economical and convenient method to predict fracture network morphology and size in the field application. However, some important factors, such as fracture height variation and filtration loss, have not been considered in the past continuous fracture network models. Therefore, this paper is aimed at establishing a novel continuous fracture network model to improve simulation accuracy. Firstly, this paper established a method to judge whether natural fractures develop or not. Then, a novel continuous fracture network model considering fracture height variation and asymmetry, filtration loss, fluid flow, and other key factors was established, and the forward algorithm and inverse algorithm of the model were proposed. At last, this model was applied in a field case to verify accuracy, and the average accuracy is more than 90%. Compared with the traditional Meyer software, the average error of prediction was reduced by 7.86%.

Proppant is one of the key materials for hydraulic fracturing. For special situations, such as middle-deep reservoirs and closure pressures ranging from 40 MPa to 60 MPa, using a single proppant cannot solve the contradiction between performance, which means crushing rate and fracture conductivity, and cost. However, using combined proppants is an economically effective method for hydraulic fracturing of such special reservoirs. Firstly, for different types, particle sizes, and proportions of combined proppants, various contact relationships between proppant particles are considered. The random phenomenon of proppant particle arrangement is described using the Monte Carlo method, and the deterministic phenomenon of proppant particles is processed using an optimization model, achieving computer simulation of the microscopic arrangement of proppant particles. Secondly, a mathematical model for the force analysis of combined proppant particles is established, and an improved singular value decomposition method is used for numerical solution. A computational model for the crushing rate and fracture conductivity of combined proppants is proposed. Thirdly, the numerical calculation results are compared and discussed with the test values, verifying the accuracy of the computational model. Finally, the application of combined proppants is discussed, and a model for optimizing the proportion of combined proppants is proposed. The onsite construction technology is introduced, and the cost and economic benefits of combined proppants are compared with those of all ceramic particles and excessive all-quartz sand. It is proved that combined proppants can balance performance and price, and are an economically effective method for hydraulic fracturing of special reservoirs. The research results can select the optimal proppant material and optimize the combination of different proppant types, which can help achieve cost reduction and efficiency increase in oil and gas development.

Proppants are one of the key materials for hydraulic fracturing, whose main role is to support fractures and create a channel through which oil and gas can flow. The nature of proppants is the most talked about feature besides their cost, for example, their sphericity, turbidity, particle size, or strength. The porosity, permeability, and fracture conductivity of proppants in fractures are also the main indicators to measure the performance of them. These indicators are usually obtained through physical experiments. However, experimental results often differ depending on the experimental scheme. Different stacking methods of proppant particles lead to this phenomenon. The nature of proppant particles in fractures varies with the way they accumulate. This paper will start with the microscopic arrangement of proppant particles. Considering the randomness and certainty of three-dimensional particle stacking and arrangement, the Monte Carlo stochastic method and an optimization model were used to conduct three-dimensional computer simulation of proppant particles. This lays an important foundation for revealing the randomness and regularity of the micro arrangement of proppant particles.

After hydraulic fracturing, some treatments intended for production enhancement fail to yield predetermined effects. The main reason is the insufficient research about the fracture propagation mechanism. There is compelling evidence that I-shaped fracture, two horizontal fractures at the junction of coalbed and cover/bottom layer, and one vertical fracture in the coalbed have formed in part of the coalbed after hydraulic fracturing. Therefore, this paper aims at I-shaped fracture propagation simulation. A novel propagation model is derived on the basis of a three-dimensional (3D) model, and the coupling conditions of vertical fracture and horizontal fractures are established based on the flow rate distribution and the bottom-hole pressure equality, respectively. Moreover, an associated PDA (pressure decline analysis of post-fracturing) model is established. Both models complement with each other and work together to guide fracturing treatment. Finally, a field case is studied to show that the proposed models can effectively investigate and simulate fracture initiation/propagation and pressure decline.

Deep coalbed methane reservoirs are characterized by complex geological conditions, strong heterogeneity, and significant variations in in situ stress, posing challenges for accurate three-dimensional in situ stress prediction. To address the issues of strong dependence on rock mechanical parameters in traditional physical models, as well as the lack of physical constraints and poor generalization capability under small-sample conditions in purely data-driven methods, this paper proposes a LightGBM prediction model that integrates physical information and data clustering. A total of 1289 fracturing clusters in the DJ block are selected as the research objects. First, the K-means algorithm is used to divide the reservoir into three categories to reduce the impact of heterogeneity. Then, a LightGBM model is constructed for each category, and physical constraints based on Huang’s model and stress–gravity equilibrium are incorporated into the loss function to ensure that the prediction results conform to mechanical laws. Taking the fracturing clusters in Category I as an example, the proposed model achieves an MAPE of 2.78% and an R² of 0.89 on the test set. Comparative experiments show that the proposed model outperforms BP neural networks, random forests, and Transformers in prediction accuracy. Ablation experiments verify the independent contributions and synergistic effects of the clustering module and the physical information constraints. Transfer experiments demonstrate that the model has good applicability to blocks with similar geological conditions. This study provides an effective method for predicting in situ stress in deep coalbed methane reservoirs, balancing accuracy and physical interpretability.

Tight oil horizontal wells in the Mahu block of the Junggar Basin commonly show rapid production decline and limited target-domain samples. These characteristics make accurate production prediction difficult. This work aims to address the small-sample overfitting problem of tight oil horizontal well production prediction in Mahu Sag with rapid production decline and limited measured well data. The concept of transfer learning is introduced to address the issue of insufficient target domain samples, and the Pearson correlation coefficient is utilized to select the main controlling factors for production from the production data. Next, based on the features extracted by the temporal convolutional network at different data scales, a multi-head attention mechanism is introduced to capture the dependencies across different time steps. Subsequently, an improved sparrow search algorithm is employed to optimize the hyperparameters of the bidirectional long short-term memory network. Finally, the bidirectional long short-term memory network is integrated to further extract the nonlinear features learned by the temporal convolutional network to conduct production prediction. Tailored to the exploitation conditions of tight oil horizontal wells in this block, a tight oil production prediction model based on transfer learning and the multi-head attention mechanism is proposed. Experimental results demonstrate that, compared with the standard bidirectional long short-term memory network, the proposed model’s evaluation metrics show a 60.93% decrease in root mean square error, a 78.53% decrease in mean absolute percentage error, and a 43.68% increase in coefficient of determination. This verifies the effectiveness of transfer learning in solving small-sample modeling challenges, providing precise technical support for the optimization of tight oil fracturing parameters and stimulation treatments in the Mahu block. The novelty of this work lies in the integration of multi-head attention temporal convolution network, quantum sparrow optimized bidirectional long short-term memory network and cross-block transfer learning for small-sample tight oil forecasting.