Model predictive control via system identification for a hot blast stove

In this paper, a control system is designed for a hot blast stove. Different from thermal models which take the complex form of partial differential equations and algebraic equations, system identification was used to model the dynamics for both the on-gas cycle and on-blast cycle based on the empirical data from No. 17 hot blast stove at Mittal Steel in East Chicago, Indiana, U.S.A. Model predictive control (MPC) was employed to minimize the natural gas usage while satisfying all the constraints. Simulation results verified the applicability of the approach. It is estimated that this proposed controller could save Mittal Steel $100,000 annually.