A Decision Support System for off-gas and steam optimal management in integrated steelworks

Optimizing the energy distribution in energy intensive industries is a difficult task. It requires a global vision for synchronizing the energy demands of all the involved processes. In this work, a decision support system is presented, which aims at supporting process operators and plant managers in monitoring and control of energy flows, e.g., process off-gases, steam and electricity in integrated steelworks. The software includes a digital twin of the integrated steelworks, for modelling or predicting internal energy production and consumption, exploited by a hierarchical control system that calculates in real-time the scheduling of the power plant, steam boilers and all the involved manipulable equipment. The digital twin implements a wide set of continuous learning models based on machine learning methodologies and standard system identification techniques. The control system integrates a high-level controller for calculating optimal references for a time horizon up to one day. These references are exploited by a set of distributed economic model predictive controllers that focus on optimizing the behavior of each specific energy network. This system allows minimizing the waste of energy, and minimizing the costs’ management, by improving the synchronization of the main energy actors.