A numerical model for analyzing microscopic fluid flow in granular media is developed using a coupled discrete element method (DEM) and computational fluid dynamic (CFD) approach to investigate fluid-particle interaction. The CFD-DEM coupling algorithm was implemented using open source code LIGGGHTS and OpenFOAM, within the framework of open source code CFDEM. A rigorous rolling resistance model was incorporated into the DEM code to approximately represent the effect of particle shape. Two benchmark examples, i.e., sand pile formation and upward seepage flow in sand, were presented to demonstrate the effectiveness of rolling resistance model and validate the CFD-DEM coupling model, respectively. The results show that the prediction of the relationship between seepage flow velocity and hydraulic gradient is consistent with classical Ergun theory. The incorporated rolling resistance model shows capability of representing the particle shape effect on the sand pile repose angle and packing porosity of sand columns.