Improved operation of a large-scale blast furnace using a hybrid dynamic model based optimizing control scheme

In the steel industry, the blast furnace (BF) is a core piece of equipment along the route from iron ore to steel, with the largest energy consumption and CO2 emissions. The stable, efficient, and economically viable thermal control of blast furnaces is still a challenging task, and fully automatic solutions are not applied industrially. The process exhibits multi-phase and multi-scale physio-chemical phenomena, in the presence of fast and very slow dynamics with latency periods of 6-8 h. Direct online measurements of key inner variables are missing, and unknown disturbances are strongly influencing the process; in particular, the quality of solid raw materials that are fed at the top of the furnace. Partial automation is the state-of-the-art strategy in the operation and control of industrial blast furnaces, and the experience and the dedication of the operators have a direct impact on the stability and efficiency of the operation. Model-based optimizing control schemes, which can be deployed either in closed -loop or as a guidance system to achieve a smooth, efficient, and reliable operation of blast furnaces, are promising. In this work, we propose a hybrid dynamic model based optimizing control scheme for the stable, energetically efficient, and economically optimal thermal control of blast furnaces. The underlying optimization problem is formulated as a combined tracking and performance optimizing control problem, where the goal is the tight control of the hot metal silicon content ([Si]) and the slag basicity (SB), while simultaneously maximizing the CO-efficiency of the furnace. These two variables are key product quality indices, and [Si] is an indicator of the internal thermal state of the BF process. Simulation results using real operational data of a large-scale industrial blast furnace are presented to demonstrate the potential of the approach for an improved operation of blast furnaces.& COPY; 2023 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).