Abstract: In order to construct the digital pore network structure of coal to simulate the microscopic seepage process of gas in coal pores, the pore-reconstruction study of coal rock based on the morphological improvement of watershed algorithm was carried out. Firstly, a method based on improved watershed algorithm to identify coal matrix particle distribution parameters and pore network was proposed, and the noise influence was reduced by median filtering and mathematical morphology transformation to avoid the phenomenon of "over-segmentation". Then, taking 5-10 mesh, 10-40 mesh and 40-80 mesh size coal as examples, the particle distribution parameters were identified and the pore network structure of the coal was established. Finally, the seepage simulation test was carried out to verify pore network gas based on the LBM. The results show that the improved watershed algorithm reduces noise interference and the reconstructed coal pore structure has high complexity and connectivity, and that the seepage simulation of the pore structure by LBM algorithm can show local details such as gas seepage velocity and direction.

Key words: watershed segmentation(WS), pore network, particle size, seepage simulation, lattice Bltzmann method (LBM)