High-Pressure Adsorption of Pure Gases on Coals and Activated Carbon: Measurements and Modeling

High-pressure gas adsorption of pure methane, nitrogen, and CO₂ were measured on five coals of varying rank and an activated carbon. All measurements were conducted at 328.2 K and pressures to 13.8 MPa using a volumetric method. The adsorption isotherm data are for dry Beulah Zap, Illinois no. 6, Pocahontas, Upper Freeport, and Wyodak coals and Filtrasorb (F-400) activated carbon. At about 7 MPa pressure, the ratio of excess adsorptions for N₂/CH₄/CO₂ varied from about 1/1.9/2.8 for Pocahontas coal to 1/2.1/5.9 for Beulah Zap coal. A distinct maximum in excess adsorption was observed for CO₂ on each coal between 6 and 8 MPa. For activated carbon at 7 MPa, the ratio of excess adsorptions for N₂/CH₄/CO₂ was 1/1.5/2.3, which is consistent with our earlier data on this activated carbon at a different temperature (318.2 K). The isotherm measurements have expected experimental uncertainties that vary from 1 to 7% for the adsorption of the three gases. The simplified local-density (SLD) model was fit to describe the pure-gas adsorption data. Overall, the weighted average absolute deviation (WAAD) was about 0.7 for both the coals and activated carbon. Generalized expressions were then developed to facilitate the prediction of nitrogen and CO₂ adsorption isotherms, solely on the basis of the methane adsorption measurements and coal characterization information. In general, the newly developed generalized expressions were capable of predicting the nitrogen and CO₂ adsorption within two times the experimental uncertainties for coals. The generalized expressions were further tested on an external data set comprised of three dry coals from the literature. For each coal, the CO₂ adsorption isotherms were predicted on the basis of the methane isotherms. The generalized predictions provided average absolute percentage deviation (% AAD) of about 10–12% for CO₂ adsorption on the three coals, which were about two times the deviations observed through direct regressions of the SLD model for these systems.