Six scaling methods are adopted to create six grading curves for numerical simulation based on one in-situ grading curve. The rate of number of particles in an interval to total number of particles is regarded as the number of measuring fractal

and thus a fractal model is developed. Fractal characteristic of particle size distribution is analysed. Using the particle flow cod

six sets of numerical samples corresponding to six grading curves are generated and used to carry out biaxial compression tests. In this test

the effect of scale method on the macroscopic and mecroscopic mechanical properties of numerical samples is observed

and the relationship between the fractal feature of particle size distribution and the mechanical properties of numerical sample is determined. The results show that the particle size distribution of numerical samples has a fractal feature

its fractal dimension D is from 1.463 to 1.783. With the increase of similar scale M based on the different scale methods

fine particle number in numerical samples increases and the filling rate between coarse and fine particles becomes better reasonable. As a result

the mechanical property of numerical sample is gradually ameliorated. Fractal dimensions D of particle size distribution linearly agrees with the mechanical properties of numerical samples.