The accurate determination of methane adsorption isotherms in coals is crucial for both the evaluation of underground coalbed methane (CBM) reserves and design of development strategies for enhancing CBM recovery. However, the experimental measurement of high-pressure methane adsorption isotherms is extremely tedious and time-consuming. This paper proposed the use of an ensemble machine learning (ML) method, namely the gradient boosting decision tree (GBDT), in order to accurately estimate methane adsorption isotherms based on coal properties in the Qinshui basin, China. The GBDT method was trained to correlate the adsorption amount with coal properties (ash, fixed carbon, moisture, vitrinite, and vitrinite reflectance) and experimental conditions (pressure, equilibrium moisture, and temperature). The results show that the estimated adsorption amounts agree well with the experimental ones, which prove the accuracy and robustness of the GBDT method. A comparison of the GBDT with two commonly used ML methods, namely the artificial neural network (ANN) and support vector machine (SVM), confirms the superiority of GBDT in terms of generalization capability and robustness. Furthermore, relative importance scanning and univariate analysis based on the constructed GBDT model were conducted, which showed that the fixed carbon and ash contents are primary factors that significantly affect the adsorption isotherms for the coal samples in this study.

The production potential of highly deviated wells cannot be fully realized by conventional acid fracturing, as it can only generate a single fracture. To fully enhance the productivity of highly deviated wells, it is necessary to initiate multiple fractures along a prolonged well section to ensure the optimal number of fractures, thereby maximizing the economic returns post-stimulation. Thus, the number of fractures is a crucial parameter in the acid fracturing design of highly deviated wells. Considering factors such as the random distribution of natural fractures within the reservoir and interference between fractures during production, and, based on the oil–water two-phase flow equation, a three-dimensional reservoir–fracture production coupling model and its seepage difference model are established to simulate the production performance of highly deviated wells under varying conditions, including the number of fractures, fracture spacing, and conductivity parameters. A numerical model for the number of acid fracturing fractures in highly deviated wells is also established, in conjunction with an economic evaluation model. The simulation results indicate that the daily oil production of highly deviated wells increases with the increase in fracture number, fracture conductivity, fracture length, and reservoir permeability. However, over time, the daily oil production gradually decreases. Similarly, the cumulative production also increases with these parameters, but shows a downward trend over time. By conducting numerical simulations to evaluate the productivity and economy of highly deviated wells post-acid fracturing, it is determined that the optimal number of fractures to achieve maximum efficiency is six. The reliability of this result is confirmed by the pressure distribution cloud map of the formation after acid fracturing in highly deviated wells.

The characteristics and influencing factors of gas production in CBM wells are analyzed based on the field geological data and the productivity data of coalbed methane (CBM) wells in the Zhengbei block, and then the favorable areas are divided. The results show that the average gas production of No. 3 coal seam CBM wells in the study area is in the range of 0~1793 m³/d, with an average of 250.97 m³/d; 80% of the wells are less than 500 m³/d, and there are fewer wells above 1000 m³/d. The average gas production is positively correlated with gas content, critical desorption pressure, permeability, Young’s modulus, and Schlumberger ratio, and negatively correlated with fracture index, fault fractal dimension, Poisson’s ratio, and horizontal stress difference coefficient. The relationship between coal seam thickness and the minimum horizontal principal stress is not strong. The low-yield wells have the characteristics of multiple pump-stopping disturbances, unstable casing pressure control, overly rapid pressure reduction in the single-phase flow stage, sand and pulverized coal production, and high-yield water in the later stage during the drainage process. It may be caused by the small difference in compressive strength between the roof and floor and the coal seam, and the small difference in the Young’s modulus of the floor. The difference between the two high-yield wells is large, and the fracturing cracks are easily controlled in the coal seam and extend along the level. The production control factors from strong to weak are as follows: critical desorption pressure, permeability, Schlumberger ratio, fault fractal dimension, Young’s modulus, horizontal stress difference coefficient, minimum horizontal principal stress, gas content, Poisson’s ratio, fracture index, coal seam thickness. The type I development unit (development of favorable areas) of the Zhengbei block is interspersed with the north and south of the block on the plane, and the III development unit is mainly located in the east of the block and near the Z-56 well. The comprehensive index has a significant positive correlation with the gas production, and the prediction results are accurate.