Machine Learning Approach for Accurate Slag Eye Predictions in Steelmaking Ladles

In the arena of steelmaking process optimization, managing slag-eye formation in gas-stirred ladles is critical. The slag eye area in steelmaking ladles is a critical parameter influencing the efficiency of inclusion removal and the control of slag-metal interactions. This study introduces a machine learning methodology for predicting the slag eye area, which has traditionally been a challenge due to the complex, dynamic nature of the ladle metallurgical process. Various machine learning models were evaluated using a comprehensive dataset from the available literature derived from physical and mathematical modeling. The model incorporates parameters such as slag thickness, melt height, ladle dimensions, number and position of gas inlets, physical properties of the constituent phases, etc., to estimate the slag eye area in both single and dual-purged steelmaking ladles. This research demonstrates superior slag eye area predictive capabilities, with implications for enhancing energy efficiency, alloy recovery, and overall process efficiency in the steelmaking industry.