Stereological bias, which defines the overestimated degree of phase liberation in 2D measurements, is an important phenomenon in liberation evaluations of breakage products and yet remains insufficiently clarified. A proper liberation-evaluation method would greatly advance the field of mining. Here, we develop a novel numerical simulation using the Discrete Element Method (DEM), which models particles with random texture, including the mixed-phase and phase-liberated particles, and quantifies the stereological bias by comparing the cross-sectional measurement of the granular sample with the 3D information. The stereological bias was found to be affected by fractal dimension of the image intensity of particle sectional texture, which is easily obtained via 2D measurements and following image analysis of the cutting plane of the sample. Therefore, this factor is a potentially useful indicator in stereological bias modeling. (C) 2016 Elsevier B.V. All rights reserved.