Industrial Experience on Process Transition Monitoring for Continuous Steel Casting Operation

Process transitions are common in the iron and steel industry. Our investigation shows that more than 50% of catastrophic process failures in continuous steel casting operation are related to abnormal operations during the process transition period. As a multivariate statistical (MVS) analysis tool, Multi-way PCA (MPCA) is applied in this paper to monitor one important process transition in the continuous casting process: submerged entry nozzle (SEN) change. A novel scheme is proposed for synchronizing process trajectories over the SEN change and the missing data existing in the synchronized trajectories are handled subsequently in both modeling and monitoring parts. The monitoring results are demonstrated by an industrial example. It is shown to provide good detectability of various process abnormalities. The proposed scheme can be further extended to monitor other process transitions in continuous casting process such as flying tundish change and product grade change. Copyright © 2003 IFAC