期刊名称：IOP Conference Series: Earth and Environmental Science
印刷版ISSN：1755-1307
电子版ISSN：1755-1315
出版年度：2020
卷号：570
期号：2
DOI：10.1088/1755-1315/570/2/022022
语种：English
出版社：IOP Publishing
摘要：The mineral bonds between grains of rock are usually simulated by parallel bonds in discrete element method (DEM) simulations. The existence of multiple input parameters makes the calibration process tedious and time-consuming. In this paper, a series of parametric studies are firstly conducted by varying the value of each input parameter independently, based on which their distinct influences on the strength and deformation characteristics of simulated specimens are obtained. One-way and multiple-way analysis of variance are conducted to evaluate the sensitivity of macro strength and deformation parameters to the change of input parallel bond model parameters. The order of influence level of these input parameters on the macro behaviour of simulated specimens is obtained and the input parameters with the dominating influences on the macro mechanical properties are identified. Then, empirical correlations between the macro mechanical properties and their dominating input parameters are established separately. These empirical correlations are taken as the basic components in the regression analysis of multiple variables for each macro mechanical property. The regression equations serve as the first estimation of input parameters for the calibration. The adjustment of input parameters is conducted with reference to the partial differentials of the regression equations considering the difference between the simulation results and the experimental data as the increments/decrements. The effectiveness of the proposed calibration approach is validated by simulating a series of triaxial compression tests on mudstone.
