Model Predictive Control with horizons online adaptation: a steel industry case study

This paper proposes a two-layer Model Predictive Control strategy based on linear models, where the horizons are online adapted. The developed horizons online adaptation law is governed by combination of different conditions concerning the controlled variables included in the controller formulation and feedback information from the plant. The control strategy has been introduced within an Advanced Process Control framework composed by several functional blocks, aimed at controlling and optimizing a pusher type billets reheating furnace located in an Italian steel plant. The synthesized control system replaced the control action performed by local standalone temperature controllers manually driven by plant operators. Significant improvements on process control have been obtained and the conflicting objectives have been successfully met. Optimal operating points for energy efficiency obtainment, production targets meeting and product quality specifications fulfillment have been achieved.