Predictive Control of Water Discharge in the Secondary Cooling Zone of a Continuous Caster

Abstract

This article examines the problem of determining water discharge in the secondary cooling zone of a continuous caster in the automated-control regime. It is proposed that the problem be solved by the Model Predictive Control (MPC) method. The method is described along with aspects of its use to solve the given problem. Results are presented from numerical experiments. One of the functions of the automated control system (ACS) of a continuous caster is maintaining a certain thermal regime for the ingot. This is done in the caster's secondary cooling zone (SCZ) by using a subsystem that controls the dis- charge of coolant water in each section of the SCZ. The design and operation of such a subsystem requires the use of algo- rithms that can perform real-time calculations of the control variable - the water discharge which is needed at a given moment to establish the best thermal regime for solidification of the ingot. Water discharge in the SCZ of a continuous caster is presently determined by two monitoring methods: open and dynamic (the latter can also be considered "open" if data from measurement of the current state of the controlled parameters are not used). Open monitoring means that the output data are not taken into account during the monitoring operation. In this case, the algorithm is constructed only on the basis of a previously prescribed rule for responding to the input signals and is not later corrected to reflect the actual values of the controlled parameters. Open control algorithms set the values for water discharge in every section of the SCZ in relation to the current casting speed. The most important step in creating algorithms of this type is experimentally choosing values of the control variables as a function of the perturbing parameters. Designers of continuous casters usually determine the values of the control variables for different ingot withdrawal speeds in the steady-state regime. Experiments we conducted led us to select an open control algorithm developed in accordance with the cooling- regime instructions for a continuous-caster ACS (Fig. 1). The instructions for a certain group of grades of steel include graphs of water discharge for each section of the SCZ in relation to ingot withdrawal speed. Since open control systems perform poorly under unsteady operating conditions, work is now actively proceeding on the development of dynamic control algorithms.