This study established the three-dimensionally random structures of fibrous filters, with fiber diameters in the range of 10 to 20 mu m and solid volume fractions (SVFs) in the range of 12 to 19%, in order to explore the filtration process inside the fibrous filters. The computational fluid dynamics (CFD), coupled with the discrete element method (DEM), was applied to simulate the airflow, the motion of particles, and the interactions between particles, fibers and fluid. The filtration efficiency and pressure drop of the filters were calculated and analyzed in CFD-DEM and the results were validated by the semi-analytical models. In addition, the correlations between, filtration performance and fiber structures, particle parameters and fluid properties were investigated. It was found that the changes in key parameters (e.g., particle diameter, fiber diameter and fluid velocity) had significant effects on filtration performance. (C) 2016 Published by Elsevier Ltd.