What is required from DEM simulations to model breakage in mills? M.S. Powell * , A.T. McBride Mineral Processing Research Unit, University of Cape Town, PB Rondebosch 7701, South Africa Received 23 December 2005; accepted 6 March 2006 Available online 5 June 2006 Abstract It is quite common to encounter discrete element method (DEM) simulations of mills that present images of the motion of grinding media, summaries of tangential and normal forces, and mill power. The usefulness of this data is questioned, with respect to modelling breakage. This work presents hypotheses of how the DEM simulations can be used as input to comminution modelling, and this guides the data logging and analysis requirements. Techniques are proposed for collecting and using this data in a manner useful for predicting breakage in a comminution device. Individual particle impact histories of contact angle, force, and impulse are required to realistically model breakage. It is argued that the majority of breakage results from cumulative damage, thus it is essential to track individual particle histories to realistically predict the breakage product from a mill. Ó 2006 Elsevier Ltd. All rights reserved. Keywords: Comminution; Grinding; Discreet element modelling; Simulation 1. Introduction The discrete element method (DEM) has been adopted by many researchers as a tool for simulating and under- standing comminution devices, and in particular tumbling mills. Simulations of mills can provide information on indi- vidual particle position and contact force history. The DEM outputs are driven by: the particle size, density, and material properties; and the mill size, liner configura- tion, filling, and rotation speed. These are the same as the inputs that change the milling throughput and product. The outputs can thus allow some predictive capability of how the energy spectra will vary with changes in milling conditions. The DEM produces large amounts of potentially useful information. For example, at every time step the particle’s position, and translational and rotational velocities may be recorded. If the particle is in contact with another particle or the mill shell, the magnitude of the normal and shear contact forces and energy absorbed in the interaction can be recorded. At the termination of each contact event the duration of the contact, the total energy dissipated or absorbed, and what particle classes (size, material, etc.,) were in contact can be logged. Accepting the underlying limitations of the DEM, i.e., simplified particle geometry, and the inability to simulate ore breakage, the challenge is to interpret and utilise the particle data to improve the predictive capability of com- minution models. It is proposed that a key aspect of this challenge is utilising the full extent of the data, and not col- lapsing the richly detailed outputs back to a few sets of sim- ple energy distributions, or worse, gross average energies. This needs to be conducted within the constraints of the computational expense incurred by calculating and logging extra information that is added to the vast data sets that can be generated by the DEM. 2. Hypotheses It is hypothesised that the following factors apply to the modelling of mechanical comminution devices: 1. Purpose of simulations (a) The purpose of conducting simulations of mills and other comminution equipment is to provide data 0892-6875/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.mineng.2006.03.009 * Corresponding author. Tel.: +27 21 650 3861; fax: +27 21 650 5501. E-mail address: mpowell@chemeng.uct.ac.za (M.S. Powell). This article is also available online at: www.elsevier.com/locate/mineng Minerals Engineering 19 (2006) 1013–1021