Development of HPGR Circuit Based on the Operating Parameters and Material Characteristics

Design and selection of equipment for a grinding circuit strongly depends on the mill function, ore properties and requirements of the down-stream circuit. Primary mills usually work in open circuit and size distribution of product would be controlled in the final steps using a classifier like cyclone. Although HPGRs are mostly utilized as pre-grinding mills, it is preferred to utilize them in a closed circuit due to their specific mechanism of comminution. Unique mechanical design of HPGR and its completely different crushing process than tumbling mills needs precise consideration on its circuit design. In this paper, we investigate the impact of different parameters on design of pre-grinding HPGR circuit and selection of the proper classifier. Our study focuses on iron ore properties (e.g. moisture, particle size distribution of feed and product, compatibility and de-agglomerability). We also investigated various options of HPGR circuit according the classifier efficiency and influences on down-stream processes. The optimum operating parameters of HPGR (working pressure, speed and level of hopper) for the selected circuit were also determined based on pilot and industrial data. Our pilot and industrial tests and observations revealed that HPGR should be arranged in closed circuit because of the edge effect, as a negative factor of non-uniform distribution of pressure force along the roller width and presence of non-crushed particles in the mill product. Wet vibratory screen could be the optimum option for screening of its discharge. Fine part of the wet screen discharge could be delivered to the concentration stage without over-grinding in ball mill using a hydrocyclone working in a reverse closed circuit mode. Moreover, our pilot and industrial tests during design and operation of ten HPGRs in various iron ore processing plants proved that higher pressure and lower roller speed in a fixed level of material inside the hopper could give the desired product size distribution.