A constraints softening decoupling strategy oriented to time delays handling with Model Predictive Control

In this work a two-layer Model Predictive Control strategy is presented. The upper layer searches for optimal steady state targets, while the lower layer manages the transient state of the considered process, calculating the future control moves. In order to efficiently handle time delays on the input-output channels, a suitable lower layer formulation has been developed. In particular, a decoupling strategy acting on the constraints relaxation guarantees a protection against critical situations tied to the presence of time delays. In the proposed Advanced Process Control system architecture, the Model Predictive Control module interacts with other key modules that assure major resilience in the plant management. Simulation and practical results deriving from the implementation of the designed controller on a clinker production phase have proven the consistency and the reliability of the proposed approach. With respect to the previous control system, based on local PID controllers, a remarkable variance reduction of the controlled variables has been observed, leading to the possibility to run the plant much closer to the operating boundary limits thus achieving, besides all, energy saving.