Abstract: Slag foaming is a key phenomenon in electric arc furnace (EAF) steelmaking, which improves thermal efficiency, suppressing metal splashing, and stabilizing the refining process. Accurate prediction and control of slag foaming are essential for achieving green and efficient steelmaking. This review aims to provide a systematic overview of the research progress on slag foaming prediction, clarifying the applicability, advantages, and limitations of different predictive methods to support intelligent control of foamy slags. Following the framework of "influencing factors-prediction methods-development trends", this review summarizes the coupling effects of multiple variables such as basicity, viscosity, surface tension, suspended particles, gas parameters, and temperature on foam formation and stability. Furthermore, it compares five major prediction approaches, including empirical formulas, dimensionless modeling, thermodynamic calculations, computational fluid dynamics (CFD) simulations, and machine learning models, and analyzes their core concepts, merits, and constraints. The results indicate that single models often struggle to balance real-time capability and accuracy, particularly under multi-variable coupling and complex operating conditions. Therefore, a hybrid prediction framework combining mechanism-based and data-driven models is proposed, emphasizing the importance of physical constraints, multi-scale coupling, and multi-source data fusion. This integrated approach is expected to advance slag foaming prediction from "computable" to "controllable and adjustable", offering methodological insights for the development of green and intelligent EAF steelmaking.

Key words: slag foaming, electric arc furnace, similarity modeling, thermodynamic simulation, CFD numerical simulation, machine learning, green metallurgy

摘要： 炉渣泡沫化是电弧炉炼钢中提升能效、抑制喷溅与稳定冶炼过程的关键环节，其行为预测与调控对实现绿色高效炼钢具有重要意义。本工作旨在系统综述炉渣泡沫化预测领域的研究进展，明确不同方法的适用范围与发展趋势，为智能化泡沫渣控制提供理论支撑。围绕“影响因素-预测方法-发展趋势”的主线，综述了碱度、黏度、表面张力、悬浮颗粒、气体参数及温度等多变量对泡沫形成与稳定的耦合效应，并系统比较了经验公式、无量纲建模、热力学计算、计算流体力学(Computational Fluid Dynamics, CFD)模拟和机器学习等主要预测方法，分析了其核心思想、优势及局限性。研究表明，单一模型难以兼顾实时性与精度，尤其在多变量强耦合和复杂工况下预测能力不足。为此，本工作提出结合机理模型与数据驱动模型的混合预测框架，强调物理约束、跨尺度耦合与多源数据融合的重要性，旨在推动炉渣泡沫化预测由“可计算”向“可控、可调”转变，为电弧炉绿色智能制造提供方法学参考。

关键词: 炉渣泡沫化, 电弧炉, 相似建模, 热力学模拟, CFD数值模拟, 机器学习, 绿色冶金