Accurate depiction of the adsorption capacity of supercritical CO₂ (ScCO₂) by existing adsorption models is an important focus for deep coal seams in CO₂-enhanced coalbed methane (CO₂-ECBM) recovery. To investigate the applicability of different adsorption models for the adsorption isotherms of ScCO₂, the validities of 10 different adsorption models were analyzed, based on analyses of the adsorption characteristics of ScCO₂ from deep coal seams of the Southern Qinshui Basin, China. These models include the Langmuir (L) model, two-parameter Langmuir (TL) model, Toth (T) model, Langmuir–Freundlich (LF) model, extended Langmuir (EL) model, double parameter Brunauer–Emmett Teller model, three-parameter BET (TBET) model, Dubinin–Radushkevich (D–R) model, Dubinin–Astakhov (D–A) model, and Ono–Kondo lattice (OK) model. These models were tested for both the excess and absolute adsorption capacities of ScCO₂ under various temperatures and pressures. The simulation accuracy of the different adsorption models was analyzed. The optimal models for the adsorption of ScCO₂ in deep coal seams were selected based on a comprehensive analysis of the simulation parameters, standard error, and residual sum of squares. There were obvious differences in the validity of the different adsorption models in terms of the excess adsorption capacity and absolute adsorption capacity of ScCO₂. The D–A and D–R models are the optimal adsorption models for the adsorption isotherms of the excess adsorption of ScCO₂ for the whole tested pressure range. The T, TL, and D–R models are the optimal adsorption models in simulation of the excess adsorption capacity of ScCO₂ for the selected adsorption models when the equilibrium pressure is divided into two sections at the point of 8.13 MPa. In simulation of the absolute adsorption capacity of ScCO₂, the TBET and LF models are the optimal adsorption models among the selected models when the equilibrium pressure is less than or equal to 8.13 MPa. The linear, exponential, logarithmic, power function, and polynomial adsorption simulation all have good precision in the simulation of the absolute adsorption capacity of ScCO₂ when the pressure is beyond 8.13 MPa.

Investigating the coal adsorption behavior on supercritical CO₂ (ScCO₂) is crucial for long-term CO₂ geological storage. In this paper, low-permeability coal samples from the Huainan-Huaibei coalfields in China were selected. The high-pressure isothermal adsorption of CO₂ was carried out at 36, 42, and 48 °C. The results of adsorption experiments were analyzed by fitting 9 types of modified adsorption models, including three different adsorption theories. Considering that different adsorption mechanisms may exist for CO₂ in coal, 14 mixed adsorption models were established. The accuracy of the coefficient of determination (R²) and root-mean-square error (RMSE) for ScCO₂ excess adsorption capacity was analyzed, mainly focusing on the accuracy of the key model parameters such as the adsorption phase density and the theoretical adsorption capacity. These parameters were discussed, combined with the predicted adsorption phase density of CO₂ based on the intercept method. The results indicate that among the 9 types of modified adsorption considered, based on the adsorption phase density screening, the deviation of the predicted adsorption capacity from the experimental value was then considered. The Dubinin-Radushkevich (DR) model can effectively fit the adsorption behavior of CO₂ at low pressure (<7.5 MPa). The Langmuir (L), Langmuir-Freundlich (LF), Extended-Langmuir (EL), and TOTH models can effectively fit the adsorption behavior of CO₂ at high pressure (7.5-20 MPa), while the multimolecular layer models were unsuitable for fitting ScCO₂ adsorption. The model fitting results showed that only the monomolecular layer and micropore-filled adsorption models were suitable for fitting the ScCO₂ adsorption capacity. The DR-LF model best fits the adsorption data based on its key parameters of adsorption phase density and theoretical adsorption capacity. The established mixed model DR-LF fitting results showed that the CO₂ in coal was dominated by microporous filling adsorption. The higher the temperature, the greater the contribution of microporous filling adsorption to the total adsorption. There still exists deviation in the adsorption phase density and theoretical adsorption capacity. The contribution percentage of different adsorption mechanisms of CO₂ in coal needs to be further investigated.

The connectivity and permeability of the coal seam pore structures control the occurrence and migration of coalbed methane. Coal samples were used from Huainan–Huaibei to reconstruct three-dimensional models of the pores and an equivalent pore network model, Statistical pore structure characteristic parameters. The pore structure of the coal reservoir was analyzed from the direction of multidimensional and multiangle. It shows that based on quantitative analysis, the representative Elementary volume of 500 × 500 × 500 was the most suitable experimental volume. The Y-axis direction of the Renlou sample had poor pore connectivity compared to that of other samples. Large volume connected pores dominated their pore systems. In terms of coal sample pore connectivity, the coal samples from the Liuzhuang and Qidong regions had pore connectivity better than those from the other regions. The pore connectivity of the Liuzhuang coal samples was the best. In terms of coal permeability, the Liuzhuang sample had better permeability than the other three samples, and the permeability was the best in the Y-axis direction. For all the combinations of the different types of throats, the shorter the throat, the greater the equivalent radius and the better the permeability. Conversely, the worse the permeability. During gas injection production, the closer the gas injection area was to the gas injection well, the poorer the connectivity and the lower the permeability over time. Near the production area, where the CO₂ did not reach the production area, the fracture porosity and effective connected porosity of the coal reservoir increased over time. When CO₂ reached the production area, the change in its connected pore structure was consistent with the change in the connected pores in the gas injection area. With this study, the coal seam pore structure on a microscale was characterized. A comprehensive analysis of the coal reservoir pore connectivity and permeability was completed. The study results are significant for the exploration and development of coalbed methane in the Huainan–Huaibei coalfield.

In the coal exploration of China, the commonly used direct method within 120 min has potential errors in lost gas calculation of deep coal seam for its complex geological conditions. The exploration of deep coal resources by drilling holes in Huainan of Eastern China offered an opportunity to starting research into developing a new method. A developed method with error analysis was constructed to estimate the lost gas using the total desorption process obtained from exploratory coal cores. The accuracy of the direct method was also evaluated comparatively. The result shows that the desorption curve of tested coal samples matches the fitted curve equation. Desorption temperature and the tectonic coal with associated pore characteristics significantly affect the variation of the adsorption characteristics and the estimation of lost gas. The direct method obviously underestimates the lost gas, and methodology using a new lost gas estimation procedure with additional residual gas allows for achieving relatively accurate results of the determination of gas content in coal seams. The calculated result of the new method is about 1.00–1.41 times that of the direct method. The error analysis of desorption results allowed us to determine the dependence between the time (retrieval time and desorption time) and determination method. The time used for desorption in the tank is allowed to extend to less than 400 min or more than 1000 min, which is very potentially important to accurately get the coalbed gas content for coring samples, especially deep exploratory cores for field application.