Continuous cast width control using a data mining approach

Abstract

Twelve per cent chrome ferritic (non-stabilised) stainless steel cast at the continuous caster at Columbus Stainless exhibited notable differences in the width change between consecutive heats. The reason for these differences is related to the fact that the steel is in a dual phase region between austenite and ferrite during the solidification stages of the continuous casting process. A model was developed and is currently used as a production tool to predict the width change of a 12% chrome ferritic heat before it is cast based on heat composition. The strand width is altered based on the model predictions by changing the secondary cooling pattern. It was uncertain if the current model is the best suited for this application and a study was carried out using different but more advanced data mining techniques in an attempt to improve the existing model. It was found that advanced data mining techniques could not improve the original rule based model.