Comminution is an energy intensive process. In SAG-mills, it is achieved by rotating a drum in which large metal balls crush ore particles. In-situ monitoring of particle size would be of considerable interest to optimize their operation. However, there is no established solution to measure particle size in such a harsh mechanical environment. We show here that the acceleration of the grinding media, which can be monitored using embedded accelerometers, can be used to sense the particle size and size distribution during operation. In DEM simulations, we find that a machine learning classifier is able to detect the size and distribution of small particles solely based on the knowledge of the acceleration of larger grinding media particles. Results show that this kinematic sensing is effective over a wide range of particle size ratios, size distribution, mixture ratio and mill charge. Beyond their potential applications in mineral processing, these results point out that the kinematics of large particles is affected by the size of the smaller particles, an observation which can help advance rheological models for bi-disperse granular flows.Graphical
Read full abstract