2025 Daniel Kavic Enhanced Temperature Prediction in Ladle Furnace Steel Refining Hybrid Process Modeling Based on Computational Thermodynamics and Statistical L

The ladle furnace (LF) is pivotal in secondary steelmaking, facilitating the addition of alloying elements, homogenizing the melt, and eliminating non-metallic inclusions. Precise temperature control is crucial for reducing energy consumption and carbon dioxide emissions during casting. This paper presents a hybrid model for temperature prediction in the LF, starting with an extensive literature review analyzing existing mechanistic, data-driven, and hybrid temperature models. The first part evaluates the strengths and weaknesses of these models, focusing on their predictive accuracies. The second part introduces a novel hybrid LF process model utilizing the effective equilibrium reaction zone method, combining computational thermodynamics with reaction kinetics. Statistical learning techniques were employed to analyze extensive LF production data from voestalpine Stahl GmbH, which included cleaning and descriptive statistical analysis. Using multiple linear regression, key process parameters impacting temperature were quantified, such as electrode heating power and cooling effects. The model demonstrated high precision, with a deviation of less than 5°C for 94% of predicted temperatures compared to actual measurements. The findings indicate that the hybrid temperature model is both accurate and reliable, offering significant potential for real-world applications in ladle furnace operations.