Predicting the Effect of Grinding Media Size Distribution on the Performance of a Ball Mill Using Discrete Element Method (DEM) and Population Balance Techniques

"Optimizing the ball charge in a grinding mill may be needed to maximise throughput or alternatively, to achieve proper product size in more challenging situations. Efforts to optimise the ball charge however are essentially empirical and suffer from the lack of appropriate, accessible and reliable tools. To address these limitations, a new model-based technology was developed that relies on 3D Discrete Element Method and population balance techniques: it takes on the form of a simulator that aims to significantly decrease the time required to optimise the size distribution and the grinding media load within a ball mill. This paper summarizes the research work related to the development of this simulator. The calibration of the different constituting sub-models will be highlighted. The successful use of the simulator to predict the effect of ball make-up on the grinding product size distribution in one plant is also reported.INTRODUCTION Adjusting the characteristics of the grinding media charge in grinding mills holds significant potential for optimisation of mill performance. For example, properly adjusting the ball size distribution in a ball mill can help achieving specified targets for final product size distribution and/or throughput capacity. However, several problems appear when reviewing the process and the available tools to optimise ball mill load: • The current empirical simulation tools are lacking with regard to the ability to predict the performance of grinding when changes to mill ball make-up are made; • Generally, these tools estimate the top size of the charge so it is difficult to predict the effect of mixed ball make-up; • As the quality of prediction of these tools is low, the operators try to optimise the ball charge by a ""trial and error"" method and proceeding with conservative changes. The influence of these changes on grinding performance is often less than the measurement precision and the improvements then remain imperceptible; • A change in ball make-up may require an average of about 6 months waiting period before reaching the stationary state of mill ball load. This exercise is due to be costly if applied change decreases the mill performance. In its effort towards the comminution process optimisation, COREM R&D portfolio had addressed the herein described opportunity during the last five years This paper describes the technology already developed for the optimisation of ball charge. It takes on the form of a simulator that aims to predict the evolution of the grinding media load, the grinding energy available in the ball mill and the final product size distribution. The simulator was first validated in a pilot scale ball mill and then applied in an industrial case."