Prepared by
Mircea-Mihai Mihet
02/11/2016ACKNOWLEDGEMENTS I firstly would like to acknowledge Christopher Leonardi (my supervisor), as without his help I would not have achieved even a fraction of the work that I have. His knowledge in the field and general wisdom has been an invaluable resource and I thank him greatly. I would also like to acknowledge Mehdi Serati who conducted the annular Brazilian experiments for which the models in this paper are based on. Furthermore, acknowledgement should definitely be given to my mother and father who give me the financially means and encouragement to live in Brisbane to attend UQ.
i
ABSTRACTThe aim of this research paper is to investigate the efficacy of combined FEM/DEM modelling for accurately representing complex stress states and fracture processes. The necessity for this is significant when taking into consideration the range of applications that can benefit from accurately modelling complex stress states and fracture mechanisms. There are direct benefits to the mining and geotechnical industry by being able to confidently simulate large-scale real-world scenarios for which complex stress states and fracture processes must be modelled. To achieve this, a suite of five, 2D FEM/DEM models was suggested based on experimental annular Brazilian tests with quantitative and qualitative outputs for comparison to the experimental data (peak loads and fracture paths). The annular Brazilian test was specifically chosen due to the lack of research in the field of computational modelling of the method, the complex distribution of stresses during loading, the unique fracture mechanisms that follow, and the availability of experimental data and highspeed fracture footage.Three preliminary models were initially conducted on two sample geometries (models 1 and 2 having an inner to outer hole diameter ratio of 0.13 and model 3 having a ratio of 0.5) to validate the code's ability to represent the fractures based on expected outcomes and to investigate the effects of scaling density for increasing computational speed. From the preliminary investigation it was determined that the code could deliver the expected outcomes, though density scaling should be avoided due to its effects on fracture propagation. These models, however, were not compared to the experimental data as they were conducted for an arbitrary coal material for faster fracturing, not the materials used during experimentation.Although there was an initial aim to complete five final models based on the experimental data, only three managed to be (partially) completed. From these models it was evident that the high loading rate caused violent fracturing rather than tensile splitting. However, the preferential fracture planes were consistent with what was observed during experimentation. The peak loads however were not, and this was likely due to the un-even distribution of stresses observed due to the high loading rate. From the models developed in this study the code's efficacy is inconclusive as not enough models were in...