AG Feed Particle Size Distribution Measurement of Iron Ore

"Feed particle size is an important variable for the primary grinding mill operation. Several commercial vision systems have proven to successfully measure online many types of ores and are used for automatic control. However, due to limitations associated with segmentation algorithms, these systems fail to provide good measurements when the feed is composed of a specific bimodal particle size distribution: very big rocks and a large amount of fine particles. This is typically the case for crushed iron ore. This paper presents a technique to overcome the limitations of existing vision systems by using 3D cameras and appropriate processing methods. It provides online particle size distribution indicators and bulk volume flow rate measurement. It is designed to be very cost-effective, robust, precise, easy to install and to maintain. The 3D system was validated in two iron ore processing plants and the results clearly show that it can be used to optimize primary grinding mill operation where 2D segmentation vision systems fail.INTRODUCTIONFragmentation measurement using digital image processing has successfully been applied in mining and mineral processing operations. Several maintenance and operating applications are already deployed in the industry: Measurement of rock fragmentation after blasting (Maerz, 1999; Palangio, 1999) (Pragyan, 2012) Truck inventory (Palangio, 2005) Monitoring of crusher mantle wear Automatic control of crushers and SAG mills (Bouajila, 2000) (Bartolacci, 2001)Commercial technologies like WipFrag (Maerz, 1999), VisioRock and others proved to be very efficient to precisely measure the particle size distribution of many types of materials. These products are mainly composed of a camera, a lighting system and an image processing system based on a segmentation algorithm. By controlling adequately the lighting conditions and with periodic maintenance on the segmentation algorithm calibration parameters, it is almost always possible to achieve reliable measurements.However, the segmentation algorithm is not adequate when the particle size distribution contains big rocks and a large amount of fine particles. In this case, the area containing the fine particles is accounted as a big rock while big rocks get segmented as multiple smaller sized particles, leading globally to a wrong measurement. This situation prevails for crushed iron ore and coke, among other materials. Figure 1 shows a picture of a material which is typically very difficult to process through segmentation techniques."