Optimization of the Gas Content Evaluation Method for Deep Coal Seams Based on Diffusion Models

Gas content is a critical parameter in coalbed methane (CBM) exploration and development, with the estimation of lost gas being the primary source of error when using the on-site desorption method. The complex occurrence states of deep CBM and the extended diffusion time before sealing in desorption canisters exacerbate these errors when applying traditional gas content evaluation methods to deep reservoirs. This study compares the performance of various gas diffusion and lost gas prediction models in deep CBM reservoirs. Following this, a modified method for calculating gas loss, capable of accurately describing the entire methane diffusion process, is proposed, and its superiority and applicability are evaluated compared with pressure-preserved coring. It also identifies the initiation time and key factors influencing gas loss. Key findings include: The bidisperse diffusion model provides the best fit, exhibiting the highest correlation coefficient (<i>R</i> ²), and the lowest mean absolute error and mean square error. In deep coal seam rope core gas content testing, using the start of core extraction as the zero point for gas loss is more geologically accurate than using the midpoint. The United States Bureau of Mines method underestimates gas loss, the adaptive curve fitting (ACF) method is unstable, and the modified curve fitting (MCF) method overestimates it. The MCF20 method, proposed in this study, reduces errors and aligns more closely with results from pressure-preserved core extraction, making it the optimal method for deep CBM gas content evaluation. A strong correlation is observed between gas loss and escape time, <i>R</i> _o, fixed carbon content, and vitrinite content during rope core extraction.