Abstract：Cleat network dominates the migration of coalbed methane in coal reservoirs, therefore, to identify its permeability is vital for quantitative estimate productivity of coal reservoirs. Compared with experimental test, the numerical method has such features as high repeatability and not being affected by the environment, test precision and human factors. It has become an effective method for description of the reservoir microstructure and excavation of fluid migration laws. In this paper, combined with self-affine property of surface geometries and spatial configuration of cleats, an algorithm representing cleat networks is developed based on the Weierstrass-Mandelbrot ( W-M) function The fluid flow in cleat network was simulated at pore scale by Lattice Boltzmann method ( LBM) and its permeability was calculated correspondingly. The results show that without butt cleats, the permeability of cleat networks follows the triple-effect fractal fracture-permeability model. After butt cleats being introduced, the Hurst exponent, the aperture of butt cleats and face cleats all affect the permeability of reservoir. The permeability increases with the increases of Hurst exponent and the aperture of butt cleats.However, the permeability of coal reservoir is dominated by spatial configuration and surface geometries of face cleats when the aperture of butt cleats is smaller than that of face cleats.

Keyword：coal reservoir;cleat network;LBM;permeability;self-affinity;triple-effect fracture-permeability model;