Control of Continuous Casting Process Based on Two-Dimensional Flow Field Measurements

Shereen Abouelazayem, M.Sc.

Doctoral thesis

Abstract:

Two-dimensional flow field measurement allows us to obtain detailed information about the processes inside the continuous casting mould. This is very important because the flow phenomena in the mould are complex, and they significantly affect the steel quality. For this reason, control based on two-dimensional flow monitoring has a great potential to achieve substantial improvement over the conventional continuous casting control. Two-dimensional flow field measurement provides large amounts of measurement data distributed within the whole cross-section of the mould. An experimental setup of the continuous casting process called Mini-LIMMCAST located in Helmholtz-Zentrum Dresden-Rossendorf (HZDR), Dresden, Germany, is used for this thesis. This thesis examines two alternatives of flow measurement sensors: Ultrasound Doppler Velocimetry (UDV) and Contactless Inductive Flow Tomography (CIFT). Both sensor variants can obtain information on the velocity profile in the mould. Two approaches were considered to create the process model needed for model-based control: a spatially discretized version of a model based on partial differential equations and computational fluid dynamics and a model obtained using system identification methods. In the end, system identification proved to be more fruitful for the aim of creating the model-based controller. Specific features of the flow were parametrized to obtain the needed controlled variables and outputs of identified models. These features are mainly related to the exiting jet angle and the meniscus velocity. The manipulated variables considered are electromagnetic brake current and stopper rod position. Model predictive control in several versions was used as the main control approach, and the results of simulation experiments demonstrate that the model predictive controller can control the flow and achieve the optimum flow structures in the mould using UDV. CIFT measurements can provide similar velocity profiles. However, further technical developments in the CIFT sensor signal processing, such as compensating for the effects of the strong and time-varying magnetic field of the electromagnetic brake on CIFT measurements, are necessary if this sensor is to be used for closed-loop control.

Abstract:

Two-dimensional flow field measurement allows us to obtain detailed information about the processes inside the continuous casting mould. This is very important because the flow phenomena in the mould are complex, and they significantly affect the steel quality. For this reason, control based on two-dimensional flow monitoring has a great potential to achieve substantial improvement over the conventional continuous casting control. Two-dimensional flow field measurement provides large amounts of measurement data distributed within the whole cross-section of the mould. An experimental setup of the continuous casting process called Mini-LIMMCAST located in Helmholtz-Zentrum Dresden-Rossendorf (HZDR), Dresden, Germany, is used for this thesis. This thesis examines two alternatives of flow measurement sensors: Ultrasound Doppler Velocimetry (UDV) and Contactless Inductive Flow Tomography (CIFT). Both sensor variants can obtain information on the velocity profile in the mould. Two approaches were considered to create the process model needed for model-based control: a spatially discretized version of a model based on partial differential equations and computational fluid dynamics and a model obtained using system identification methods. In the end, system identification proved to be more fruitful for the aim of creating the model-based controller. Specific features of the flow were parametrized to obtain the needed controlled variables and outputs of identified models. These features are mainly related to the exiting jet angle and the meniscus velocity. The manipulated variables considered are electromagnetic brake current and stopper rod position. Model predictive control in several versions was used as the main control approach, and the results of simulation experiments demonstrate that the model predictive controller can control the flow and achieve the optimum flow structures in the mould using UDV. CIFT measurements can provide similar velocity profiles. However, further technical developments in the CIFT sensor signal processing, such as compensating for the effects of the strong and time-varying magnetic field of the electromagnetic brake on CIFT measurements, are necessary if this sensor is to be used for closed-loop control.

Keywords

Continuous Casting, Contactless Inductive Flow Tomography, Industrial control, Model Predictive Control, Ultrasound Doppler Velocimetry

Language used: Czech

Date on which the thesis was submitted / produced: 16. 6. 2022

Thesis defence

• Supervisor: doc. Dr. Ing. Jaroslav Hlava

Technical University of Liberec

Faculty of Mechatronics, Informatics and Interdisciplinary Studies

Doctoral programme / field:
Electrical Engineering and Informatics / Technical cybernetics