Simulation environment for advanced control development of a multiple hearth furnace

Abstract

The aim of this thesis is to setup a simulation environment that prepares the ground for Multiple Hearth Furnace (MHF) advanced process control development based on Economic Model Predictive Control (EMPC) and/or MPC techniques.

Additionally, the interest is to design an Economic MPC for the Multiple Hearth Furnace, aiming to minimize energy consumption of the furnace while maintaining the specified product quality. The implementation of the EMPC requires a dynamic model which is simplified from a previously developed mechanistic model of the MHF. The simplified model is developed in the form of a nonlinear Hammerstein-Wiener model, which is linearized at every sampling time to carry out the state estimation and MPC optimization tasks. As the accuracy of the simplified process model is crucial for performance of the EMPC, the thesis aims to compare the simulation results of the mechanistic model and the simplified one. The comparison of the models show that the simplified model follows accurately the mechanistic model in all cases.

A description of the process of interest is given, with an emphasis in outlining the overall control strategy currently implemented. Next the components of the EMPC design are illustrated, including the overall strategy, the cost function and the necessary models of the process. Afterwards an implementation algorithm is provided comprising all the elements of the design, in order to obtain the optimal control of the MHF. Finally, practical problems regarding industrial implementations of temperature control in the hearth 4 are discussed and further research items outlined.