1 INTRODUCTION In mining and quarrying, blasting is the main meth- od for rock excavation. The explosives are placed in series of boreholes with a predetermined amount and detonation delay time, aiming at achieving a man- ageable muck pile and a desired fragment size dis- tribution (FSD). An under- or over-charged bore- holes are often a financial liability for the companies. That is, in addition to the time and re- sources costs, the resultant fragment sizes might be smaller than the acceptable size for the processing (recovery) units. Even in well-designed blasting rounds fine particles are generated. For example, in EU quarries around 2.5 billion tons of rocks are an- nually blasted of which 10 – 15 % is unsellable waste fine particles (Moser 2003). There are many theories on the source of fine par- ticles due to rock blasting, e.g. the traditional crushed zone model (CZM) that assumes fines are originating from an annular crushed zone around the blast hole, the two component model (TCM) devel- oped by Djordjevic (1999, 2002) and further im- provements of the CZM by Onederra et al. (2004) who assumed that the fine particles are originating from a star-shaped crushed zone. Another yet plau- sible one is that the fines are generated by a mecha- nism involving dynamic crack branching and merg- ing (Ouchterlony & Moser 2012). By the advancement of computation power and numerical tools, many researches have started to in- vestigate blast induced damage, dynamic crack branching and merging and thus fragmentation through numerical simulations. Cho & Kaneko (2004) e.g. studied the dynamic fracture process of a two dimensional disk with a borehole at its center subjected to different dynamic wave forms, different rise time and decay time of the pressure function, peak value of the applied pressure and stress loading rates. Zhu et al. (2007) studied blast induced damage and dynamic crack propagation of a circular rock model with a centrally located borehole using the fi- nite element method (FEM) code Autodyn 2D. Ma & An (2008) implemented a Johnson-Holmquist (JH) constitutive model into the commercial FEM code LS-Dyna and studied the borehole blast in- duced rock fracture and fracture pattern under dif- ferent circumstances i.e. different stress loading rates, effect of free surface and joint plane, pre- existing compressive stress, notches, etc.. Wang & Alonso-Marroquín (2009) used the 3D discrete ele- ment method (DEM) to model the fracture process and the size distribution in a sphere resulting from different impact rates. Banadaki & Mohanty (2010) investigated the crushed zone, radial and spalling cracks of cylindrical Barre and Laurentian granites subjected to blast load using the built-in JH2 materi- al model in Autodyn with explicit time integration scheme, Nordendale (2013) studied the damage in- duced in ultra-high strength concrete and ashcrete panels of size 305 mm × 305 mm × 27 mm by ballis- tic impact. Modelling blast fragmentation of cylinders of mortar and rock A. Iravani, I. Kukolj, F. Ouchterlony Dept. Mineral Resources Engineering, Montanuniversitaet Leoben, 8700 Leoben, Styria, Austria T. Antretter Institute of Mechanics, Montanuniveristaet Leoben, 8700 Leoben, Styria, Austria J. Åström CSC-IT Center for Science, P.O. Box 405, 02101, Espoo, Finland ABSTRACT: This paper investigates the blast fragmentation of a mortar cylinder by numerical simulations. The aim of the project is to understand the underlying mechanisms causing blast induced fines. Two numeri- cal methods: Finite and Discrete Element Methods (FEM, DEM) with explicit time integration were used and the results were compared with the results of blasting tests. In FEM thin cylindrical disk (Ø140 mm) with 1 layer of 3D continuum elements and in DEM a 3D cylinder with Ø140×200 mm were modelled. They were loaded by a pressure evolution acting on borehole wall. Both models reproduce realistic crack patterns con- sisting of through-going radial cracks, with branching and interconnecting cracks, around a crushed zone at the borehole. The FEM models, however, for slight changes contain unrealistic areas of deleted elements, whereas the DEM models were more robust and delivered realistic fragment size distribution of the expected Swebrec function type.