Abstract
The permeability of rocks is a critical parameter in many subsurface geological applications, and pore properties measured on rock samples (including rock fragments) can be used to estimate rock permeability. A major use of MIP and NMR data is to assess the pore properties of a rock in order to estimate the permeability based on empirical equations. Although sandstones have been extensively studied, permeability in coals has received less attention. Consequently, in order to obtain reliable predictions for coal permeability, a comprehensive study of different permeability models was performed on coal samples having a range of permeabilities from 0.003 to 1.26 mD. The model results showed that the seepage pores in coals account for the bulk of the permeability, while the contribution of adsorption pores to permeability is negligible. The models that only consider a single pore size point on the mercury curve, such as the Pittman and Swanson model, or those that use the entire pore size distribution, like the Purcell and SDR model, are inadequate for predicting permeability in coals. This study modifies the Purcell model to determine permeability from the seepage pores of coal, resulting in the enhancement of the predictive capability, with an increased R(2) and reduction in the average absolute error by approximately 50% compared to the Purcell model. To apply the modified Purcell model to NMR data, a new model was developed that provides a high degree of predictive capability (∼0.1 mD). This new model can be used for cuttings, which could lead to a new method for field permeability estimation.