Model predictive control simulation, implementation and performance assessment of a coal comminution circuit

Abstract

This paper describes the simulation, implementation and performance assessment of model predictive control (MPC) for a coal communition circuit, and describes, as far as can be ascertained, the first implementation of MPC in the coal processing industry.

Dynamic models of an actual coal comminution circuit were derived in previous publications using the principle of mass conservation with unknown parameters that were identified with actual plant production data. The identified dynamic models are used in nonlinear MPC simulations to determine the process control objectives and to calculate a potential process improvement in throughput. This increase in potential throughput was estimated to be 5% and was used as motivation in a business case to obtain funding to implement MPC on the actual plant.

After performing a tender process, General Electric was contracted to implement the MPC. The MPC was implemented successfully on the comminution system and compared to current manual plant operational performance. After evaluating the MPC through a statistical performance assessment, it was found that throughput was improved by a significant 8.22%. The methodology followed is presented in this paper and the improved production throughput motivates for further MPC technologies to be implemented at other plant operations.