Model predictive control of froth flotation processes aided by a dynamic simulator

Abstract

A model and simulation based methodology is used to implement a multi-layer model predictive control (MPC) strategy for a rougher row of mechanical flotation cells. Pilot-scale tests are done to calibrate and validate both the process simulation models and the predictive simulation models. The hierarchical control strategy considers three layers: orchestrator, advanced control and basic control; is deployed, in a commercial control system and, tested in a pilot row. The orchestrator is divided in two: the row supervisor and the row optimizer. The row supervisor monitors and manages all the other components of the control structure. The optimizer is a MPC-based controller which optimal criterion is separation efficiency (SE) and; according to recent developments, that happens with a balanced mass-pull profile along the row. The advanced control layer includes individual cell MPC in coordination with a symbolic MPC for all pulp levels along the row. The basic control layer consists of single loop proportional and integral (PI) controllers and their corresponding valves and instruments. After simulation, the control layers are successively downloaded in an industrial controller, starting from the basic control layer and ending with orchestrator s algorithms. Then, the control structure provides good disturbance rejection against feed variabilities. Regarding the orchestrator, it supports smooth and logical transitions between control modes as well as good abnormal situation management. This work shows promising results of the power of integrated process control design and model based methodologies; allowing earlier and better selection and validation of: flotation machine technology, cutting edge instrumentation and, advanced control structure and strategies. Given pre-defined economic assumptions, estimated results are obtained for the simulated industrial scenario: almost 40 percent reduction of capital expenditure (Capex), with almost the same operational expenditure (Opex). From the total Capex reduction, almost 80% is due to integrated process and control design (IPCD), being the other 20% a consequence of advanced process control and optimization structure and strategies. MPC-based control algorithms show their potential to have a main role in mineral processing processes feasibility and optimality.