Sand filtration is a widely used solution to purify water and provide safe water in addressing contemporary water crisis issues, which takes advantage of the porous nature of sand filter composed of sand particles to absorb and trap pollutants to get clean water. To optimize its performance, much research has been conducted to investigate fluid flow in the packing of sand particles. As sand particles are often randomly packed and result in complex pore network of intricated small channels, fluid flow within a sand filter is complicated. Many studies have attempted to establish the pore network flow in sand filters, but the local flow field around particles as a function of local pore and particle structures has not yet been described in detail. This research employs numerical simulations based on discrete element method (DEM) and computational fluid dynamics (CFD) to study fluid flow in sand filters. DEM is used to obtain the packed bed of sand particles with realistic packing structures, and CFD is then applied to solve the complex flow field with fully resolved sand particle boundaries considered. The model is verified by comparing the total pressure drop across the whole packed bed under varying inlet velocities with both the Ergun equation and experimental data. By using this model, our research delves into the analysis of velocity distribution, drag force, and suspended solid residence time for the flow in the sand filter under different inlet flow directions. Through the analysis, the complex flow characteristics inside the packed beds are related to the local pore and particle structures, and further related to the pore network structure. Our study not only contributes to new understandings of porous media flow at microscopic scale but also provides guidance for optimizing sand filtration efficiency, which helps advance sustainable water management practices.