Abstract: The discrete element method (DEM) is commonly used for simulating particle systems at the engineering scale, but computational efficiency remains a limiting factor for large-scale particle system simulations. Existing coarse-grained methods have limited applicability and lack a universal theoretical basis. To address this issue, this study utilizes dimensional analysis to describe the scaling laws of physical quantities in an exact scaled system. By using representative volume elements (RVE), approximate conservation relationships for mass, momentum, and energy are established between the coarse-grained system and the original system. Scaling relationships for corresponding physical quantities at two different scales (global and particle level) are obtained.To validate the correctness of the proposed two-scale coarse-grained DEM method, the method is applied to the falling and rotation tests of bulk coal. The particle size in the actual tests is 4mm, while in the simulations, five groups of coal particles with different scaling coefficients ranging from 4 mm to 12 mm are used. The total volume of bulk coal in both the tests and simulations is 0.001 m³. The falling test of bulk coal uses the average impact force and the reposeangle as comparative indicators. The results show that as the particle size of coal increases, the relative errors of both indicators generally increase (at a particle size of 12 mm, the error in impact force is 14.36%, and the error in the reposeangle is 19.05%).The rotational test of bulk coal uses the correlation coefficient between the applied force on the upper sample and the profile curve of the bulk coal heap as a comparative indicator. The results show that as the particle size increases, the relative error increases, and the correlation coefficient generally decreases (at a particle size of 12 mm, the force error is 39.29%, and the correlation coefficient is0.9574).The simulations require less time compared to the predicted duration, fully demonstrating the effectiveness of the dual-scale coarse-graining DEM method. The coarse-grained system can significantly improve computational efficiency within an acceptable range of errors.