Characterization of shale gas adsorption behavior based on the gravimetric method

Based on gravimetric isothermal adsorption experiments, methane adsorption isotherms were obtained under different temperature and pressure conditions. High-pressure adsorption models were used to investigate methane adsorption behavior and its controlling factors in shale, and machine-learning methods were further introduced to predict adsorption capacity and quantify the influence of different parameters. The dual-site Langmuir model, supercritical Dubinin-Radushkevich model, and Ono-Kondo lattice model were employed to fit and compare high-pressure adsorption behavior over the full pressure range. For machine-learning prediction, temperature, pressure, total organic carbon, and vitrinite reflectance were selected as input variables, and excess adsorption capacity was used as the output variable. The results show that the absolute methane adsorption capacity of shale generally increases rapidly with pressure at the initial stage and then gradually approaches a plateau, while it decreases significantly with increasing temperature. The Ono-Kondo lattice model exhibits the best fitting stability and physical interpretability across different pressure ranges, effectively characterizing adsorption behavior under supercritical conditions. Factor analysis indicates that total organic carbon is positively correlated with methane adsorption capacity, whereas temperature shows a negative correlation. Thermal maturity, as represented by vitrinite reflectance, also affects methane adsorption capacity; however, its apparent influence is nonlinear and coupled with total organic carbon content. Among the machine-learning models, CatBoost achieves the highest prediction accuracy, with a coefficient of determination of 0.9528, indicating strong nonlinear fitting capability. SHAP-based interpretability analysis shows that the importance of different parameters controlling methane adsorption capacity follows the order: pressure > total organic carbon > temperature > vitrinite reflectance. This study provides experimental evidence and methodological support for understanding high-pressure shale gas adsorption mechanisms, accurately characterizing adsorption capacity, and evaluating geological gas-in-place in the Xujiahe Formation of the Sichuan Basin.