Model-based control of mineral processing operations

Abstract

The use of phenomenological models for on-line decision making represents the next logical step in mineral processing control system design. The key elements required for model-based control are good on-line models, accurate estimation procedures based on precise measurements and inexpensive distributed control hardware to handle stabilizing and optimizing levels of calculation. These elements are now available with the necessary characteristics. This paper discusses the development and implementation of the model-based control concept for mineral processing systems. Four examples of model-based control strategies for mineral processing operations are presented. One is a supervisory strategy for a semi-autogenous grinding mill circuit in which Kalman filter estimates of ball and rock holdup in the mill as well as ore hardness are used. The second is an optimal feedback control strategy to maintain at setpoint the slurry volume in the sump and the product particle size of a rod/ball mill grinding circuit. The third is a model-assisted control strategy which dynamically optimizes feed-rate and particle size for a single-stage grinding circuit. The fourth example is a feed-forward strategy where Kalman filter estimates of ore flotability in the first two cells of a flotation circuit can be used to predict the steady state recovery and concentrate grade. By minimizing an economic objective function involving recovery and grade, the optimal setpoint values of collector, frother and pulp level can be found. In all instances in which the authors and their coworkers have evaluated this concept in plants, it has been found to be very successful.