The Particle Flow Code (PFC) is a useful tool for simulating the failure behavior of hard rocks. The introduction of clump logic into the PFC modeling has made it possible to better reproduce the brittleness characteristic of hard rocks, namely, the large difference between compressive strength and tensile strength. However, there is a problem in the PFC simulation with the clumped particle model, namely, when we simulate Brazilian tensile tests on hard rocks, the results do not fully express the rapid decrease in tensile stress after the peak stress. One reason for this problem is the existence of clumps that block the rapid propagation and the connection of the micro-cracks in the model by their interlocking effect. Focusing on the effect of clump configuration on the Brazilian tensile failure behavior of hard rocks, this paper carried out simulations by varying clump and ball size. The experimental mechanical properties of Lac du Bonnet granite were targeted in the simulations. From the simulation results, we confirmed that brittleness can be controlled by changing the ratio of clump size to minimum ball size and that under a constant ratio of clump size to minimum ball size, the local peak stress does not change with the minimum ball size.
The Particle Flow Code (PFC), a numerical simulation code based on the distinct element method (DEM), is an effective numerical tool for simulating the failure process of hard rocks. By introducing the contact- and parallel-bond and clumped particle model (Cho et al., 2007 ), PFC has become better able to express the brittle failure of hard rocks (e.g., Funatsu et al., 2008 ). On the other hand, a problem has been pointed out, namely, that a PFC simulation with the clumped particle model cannot reproduce the rapid strain-softening behavior in Brazilian tensile tests (Inoue et al., 2009 ; Nakashima et al., 2011, 2013 [4, 5]). In addition, the cracks generated in the simulation are too widely distributed in the specimen compared to those in experiments. This problem should be overcome before handling actual excavation problems, firstly because the stress field generated around real underground cavities may be more complicated than that of the Brazilian tensile condition, and secondly because Brazilian tests and uniaxial compression tests are the most commonly used element tests to calibrate the input parameters of a PFC simulation. If Brazilian tests are not simulated with adequate accuracy, reliable input parameters cannot be obtained to model the targeted rock.