Abstract: The simulation accuracy of the particle discrete element method (DEM) mainly depends on the accuracy of calibrated mesoscopic parameters. To improve the calibration accuracy and efficiency of mesoscopic parameters, firstly，the rolling-resistance linear contact model was used to simulate sand triaxial compression tests. Then, the variance analysis method in orthogonal experiments is used to select the controlling mesoscopic parameters (CMP), and response surface equations are established to relate the macro parameters to the CMP. By substituting the macro parameters determined from triaxial tests into the response surface equations, the values of the CMP are solved. Finally, the accuracy of the calibrated micro parameters is verified by comparing the simulation results with the triaxial test results. The results show that the macro parameters include elastic modulus, Poisson’s ratio, dilatancy angle, and peak friction angle. The CMP for the elastic modulus are effective modulus and friction coefficient, while the controlling micro parameters for Poisson’s ratio are stiffness ratio and friction coefficient. The CMP of the peak friction angle and dilatancy angle are the friction coefficient. The elastic modulus, Poisson’s ratio, peak friction angle and dilatancy angle from the triaxial tests, are 83.83 MPa, 0.45, 34.47° and 8.93°, respectively. The corresponding mesoscopic parameters are determined as follows: Effective modulus of 153.35 MPa, stiffness ratio of 2.16, and friction coefficient of 0.45. Comparing simulation results from the calibrated mesoscopic parameters with the triaxial tests, the difference in peak strength is 7.2% at a low level of confining pressure (100 kPa), and the errors are within 15.0% at higher level of confining pressures (300 and 500 kPa).

Key words: sand,  , discrete element method, mesoscopic parameter, orthogonal test, response surface method, marco parameter, elastic modulus, Poisson’s ratio