In the blast furnace hearth, excessive accumulation of molten slag and iron may lead to an increase in gas pressure drop, which can cause issues such as raw burden hanging. Based on simulations using cold model experiments and a numerical model, it was found that the gas flow region in the blast furnace hearth contracts due to the accumulation of melt, resulting in an increase in gas pressure drop, and, when the average liquid level rises to within 1 m from the tuyere, gas pressure drop increases rapidly. To prevent this issue, a prompt reduction of the blast volume and increase of the tapping rate are recommended.
Five data sets, each containing a year’s worth of electric arc furnace (EAF) steel production, were analyzed for electricity consumption trends using a literature multilinear regression (MLR) model, a modified MLR model with engineered variables, and a machine learning (ML) model utilizing XGBoost and Shapley analysis. While the literature model provided a stronger fit to data, the engineered model resulted in more physically probable results and is recommended due to a wider applicability. The results of the ML analysis provided an auxiliary analysis that confirmed several accuracies in the engineered model; however, the results exposed some weaknesses of the model, namely its sensitivity to outlier data points. It is determined that the analysis of EAF electricity usage is most effectively done by utilizing the MLR with engineered variables in conjunction with an ML analysis.
