Calibration of a Dynamic Model for an Industrial Comminution Circuit

"Several grinding mill dynamic simulators have been developed in the last half-century. However, complete circuit simulations are mainly limited to academic investigations, and on-site applications for advanced control or real-time optimization are scarce. One particular issue that has certainly curbed a more widespread use is the time and efforts required for model calibration. This paper illustrates this challenge and presents a calibration method based on dynamic simulation using industrial data and laboratory testwork. The simulator models discussed originate from the work of Austin and Luckie (1972) for grinding mills, and Plitt (1976) for hydrocyclones. A two-stage approach based on a least squares minimization allows determining calibration parameters values. Firstly, laboratory testwork are required for characterizing breakage coefficients of the ore. Secondly, step change responses and their simulation— effect of ore and water feed rates on output flow rates and size distributions—allows identifying transport/mixing conditions. Simulation results are in agreement with actual process responses to variations of feed conditions obtained from a plant survey.INTRODUCTIONDynamic models for grinding circuits has been intensively studied since the 1970's notably by Austin and Luckie (1972b), Klimpel and Austin (1984), Rogovin and Hogg (1988), Napier-Munn et al. (1996), Liu and Spencer (2004), and King (2012) among others. The resulting models primarily serve in simulation applications for process equipment sizing, which incidentally also helps developing a better understanding of comminution operations. An obstacle curbing a more widespread and targeted use of grinding simulation for process control and optimization purposes is the time and efforts required for dynamic model calibration. It remains a cumbersome task that is typically not well detailed in technical papers. This could explain why dynamic process simulators are mainly confined to academic applications."