A novel permeability prediction model for coal based on dynamic transformation of pores in multiple scales

Abstract

Schlumberger-Doll research (SDR) model is popular in predicting rock permeability, however the intrinsic defects constrain its applicability in coal. Pore structure in coal is complicated while SDR model assumes it is homogeneous. To solve this problem, we conducted water flushing experiments to simulate coalbed methane (CBM) drainage process using nuclear magnetic resonance (NMR) on durain (DHB-9-3), clarain (HP-5-3) and semi clarain (MD-6-6). Through comparative analyses on stress sensitivity, adsorption pores in DHB-9-3, non-adsorbed pores in HP-5-3 and seepage pores in MD-6-6 show a greater sensitivity. The non-adsorbed pores, which are responsible for reservoir permeability, become more complicated under increasing effective stress. The fractal dimension of non-adsorbed pores in DHB-9-3 and HP-5-3 increases from 2.93 to 2.97, and from 2.92 to 2.93 in MD-6-6 when the inlet water pressure reduces from 8 MPa to 3 MPa, indicating a more complicated pore geometry and a longer water flow pathway. As a result, the permeability shows a downward trend. A novel coal permeability prediction model is constructed with a factor of fractal dimension embedded in. Compared with SDR model, the new model is verified to have 0.79 times, 0.16 times and 0.08 times less error rates in DHB-9-3, HP-5-3 and MD-6-6, respectively.