Numerical simulations of a centrifuge model of caving

Validation and calibration of numerical models is vitally important, particularly in the field of cave mining where our ability to monitor the caving rock mass is limited. As part of a project investigating caving mechanics, physical models of caving were tested in a geotechnical centrifuge (Cumming-Potvin et al. 2016b). This paper describes numerical simulations of one of the centrifuge tests. Two approaches were used, so the relative strengths and weaknesses could be compared. The Itasca caving algorithm approach was implemented in FLAC3D and a bonded block model (BBM) approach implemented in 3DEC. The results showed that there was a good match with the physical model in some respects. Both numerical approaches were able to capture the discontinuity in the damage profile seen in the physical model, and the shape of the failed zones also matched the physical model well. The match appeared to be better for the BBM approach. There is some uncertainty as to whether the mechanism of failure seen in the physical model was exhibited in the numerical models. The match between the numerical and physical models could have been improved via further calibration. However, this was outside the scope of this study. Overall, the results show that the numerical approaches used are suitable for practical use in the modelling of caving, particularly if they are suitably calibrated and/or validated using in situ monitoring data. Keywords: numerical modelling, cave mining, physical modelling, centrifuge, fracture banding