Tensile Strength Calibration in DEM Modeling

Tensile strengths from the direct tension test and the Brazilian test are compared for the purpose of material property calibration for DEM modeling. We show that size effect in direct tension is statistically insignificant. Sample size is not critically important in calibrating the tensile strength. On the other hand, failure mechanisms in the Brazilian test depend on not only the material properties but also sample size. In addition to the conventional diametrical splitting failure resulted from initiation and propagation of a center crack, alternative failure scenarios involving formation of damaged zones near the loading points are also possible. The Brazilian tensile strength could either underestimate or overestimate the intrinsic tensile strength. Direct tension instead of the Brazilian test should therefore be conducted for material property calibration. We also show that a displacement-softening contact law can resolve the issue of low compressive over tensile strength ratio in DEM modeling. By adjusting the softening coefficient, a realistic strength ratio can be obtained and the conventional diametrical failure pattern can be reproduced in the Brazilian test.