Mixed-Integer Predictive Control for a Three-Phase Electric Arc Furnace Producing Silico Manganese

In the metallurgical industry, Electrical Arc Furnace (EAF) are usually controlled through simple rules, without necessarily handling the coupling among their various components leading to inefficiency in the operation (e.g., instability in multi-phase control, unbalanced power distribution). Herein, we first develop a mathematical model of the EAF which is able to capture the behavior of the three-phase electrical evolution in time. Then, we formulate a mixed-integer optimal control problem in an MPC (Model Predictive Control) framework for the plant's linearized model. The goals are to concurrently control power and intensity across the three phases to track a priori given set points, handle integer inputs and limit transformer tap switching frequency. This contributes to enhancing the furnace stability and energy efficiency. The approach is showcased over a three-phase furnace simulator developed by Eramet Ideas.