AbstractTo ensure the quality of grain storage, researchers have developed a variety of models for the morphological structure of bulk grain piles. However, traditional characterization methods suffer from issues such as inaccurate models, limited size ranges, and low precision. In this study, x‐ray computed tomography was employed for the first time to capture real three‐dimensional (3D) images, revealing the surface porosity distribution ranging from 30% to 38% along the slice direction and a fractal dimension primarily distributed between 1.45 and 1.47. Moreover, the box counting method was used to determine the representative elementary volume (REV) comprising 600 × 600 × 600 pixels, effectively characterizing pore structure using porosity as an index. Connectivity analysis of the REV was conducted by integrating the refined central axis method and the 3D watershed algorithm. Equivalent diameters of connected pores were mainly distributed between 0.5 and 3.5 mm, with an average pore diameter of 2.22 ± 0.02 mm, an average coordination number of 7.04 ± 0.07, and an average tortuosity of 1.58 ± 0.01. Based on the characteristic parameters of connected pores, an equivalent pore network model (EPNM) was reconstructed for numerical simulation of single‐phase percolation of bulk grain pile. In addition, the constructed experimental platform demonstrates that the constructed EPNM closely corresponds to the real pore structure of the seed body, accurately reflecting pore–throat size, connectivity, and morphological characteristics within the grain pile. Furthermore, this research model can be applied to the study of gas flow, heat transfer, and mass transfer within porous media.