Linear multivariable dynamic model of a hot strip finishing mill

Abstract

A steel hot strip finishing mill presents significant interactions between variables that considerably reduce performance. Traditionally, the control techniques implemented are based on classical approaches assuming an ideal model. Modelling interactions would allow the opportunity to suppress them by the application of modern multivariable control techniques. In this paper the derivation of a linearized multivariable model of a real-life hot strip mill based on previous schemes is presented as a case study. Linearization of the non-linear static relations around an operating point is performed. The operating conditions are the mill set-up calculations obtained from rolling schedules and non-linear relations used in the plant. Some useful details for practitioners are given. Simulations and experimental model validation are also presented. Since the mill in the real hot rolling line is in closed loop, it is not possible to carry out an experimental validation of the plant itself (i.e. in open loop); therefore, the model was experimentally validated in closed loop. As an example of the use of the multivariable model an H_∞ multivariable optimal controller to attain nominal performance is designed for the looper system and results are presented.