Hierarchical Model Predictive Control Applied to a Cement Raw Material Mixing Process

A practical model predictive control method is proposed which was applied to a cement raw mate- rial mixing process. The method has a hierarchical structure of 3 layers consisted of set point calculation with an optimizer, multi-input multi-output material mixing ratio control system with the Model Predictive Control scheme, and DCS local control system with PID controllers. It realized priority management of multi-objective control specifications and constraints, avoidance of infeasibility in optimization, handling of control degree of freedom, feed-forward control for transient manipulation, and adaptation to the time-varying components in each raw material with Kalman filtering estimator. A 72-hour test operation was executed for a real plant with regular raw materials and some additive industrial waste materials, which revealed good con- trol performance and reached an economical optimal operation point automatically.