A dynamic flotation model for real-time control and optimisation

A state- and parameter observable flotation model was developed to be used in industrial flotation control and optimisation applications. What distinguishes this model from other models available in literature is that key parameters characterising flotation operation - that could previously only be calculated from manual sampling campaign results - can now be estimated in real-time using measurements that are commonly available on industrial flotation circuits.

The effect of air recovery on different flotation mechanisms is illustrated in Figures 2_1 – 2_8, highlighting the need to include non-linear effects such as air recovery as part of a model-based control strategy. A model-based control strategy using this approach needs to include both a controller and an estimator to compensate for variability, as shown in the structure in Figure 3_3. A linear non-model-based control solution was configured (Figure 3_4), to provide an indication of potential improvements of the model-based control solution over conventional linear control strategies. Simulation results are shown in Figures 4_1 – 4_9, with the non-linear model-based solution consistently outperforming the linear non-model-based solution. The non-linear model-based solution has the potential to improve circuit performance most when both pulp levels and air flow rates into flotation cells are used as manipulated variables, with significant potential to improve recovery to the concentrate without compromising grade. 

The ability to reject operational disturbances effectively and to compensate for variability in process characteristics are key components of a successful long-term control solution. Both these objectives (disturbance rejection in Figures 7_1 and 7_2, and parameter estimation in Figures 6_1 – 6_6) can be achieved when a non-linear model-based control strategy with a parameter- and state estimator capabilities are integrated in a structure as shown in Figure 3_3.