Nonlinear model predictive control for improved water recovery and throughput stability for tailings reprocessing

Abstract

The reprocessing of tailings aims to recover residual wealth, reclaim or rehabilitate valuable land, or mitigate safety and environmental risks. These aims all support environmental, social, and governance measures that are increasingly placed at the centre of corporate strategy. Tailings reprocessing operations are water intensive, and typically include surge tanks with both level and density averaging objectives to improve the efficiency of downstream water and mineral recovery. In this study, a rigorous dynamic model is derived to describe the rate of change of both the volume and density in these surge tanks. By simulation with industrial data it is demonstrated that the significant input disturbances typical to tailings reprocessing circuits drive a gain inversion in the density model of the surge tank. Since conventional linear averaging control approaches are not ideally suited to deal with gain inversion and multivariable control objectives a nonlinear model predictive controller (NMPC) was derived and implemented on an industrial tailings reprocessing surge tank. Results show a 5 % improvement in water recovery from the plant tailings product, and a 27 % reduction in the standard deviation of the tailings product mass flow.