In the present study, the numerical accuracy of a continuum species transfer (CST) model coupled with the volume-of-fluid (VOF) method for high Schmidt number fluids with or without chemical reactions was evaluated for the purpose of creating a numerical model of multiphase flow for use in a steel-making process. Even under the condition without a chemical reaction, the original CST model is incapable of accurately calculating mass transfer under conditions characterized by a high Péclet (Pe) number. However, the compressive CST (C-CST) model has been demonstrated to possess the capacity to accurately calculate mass transfer under conditions characterized by a high Pe number. The C-CST model, when coupled with a chemical reaction, was also validated for chemical absorption. The C-CST model with a chemical reaction can calculate the entire mass transfer involving a chemical reaction, particularly in conditions characterized by high Schmidt and Péclet numbers. However, numerical inaccuracies have been observed in the local mass transfer and chemical reaction due to calculation errors in the interfacial curvature necessary for the VOF method.

To clarify the sticking mechanism and discuss its suppression during hydrogen reduction in the fluidized bed, in-situ observation of the reduction behavior of wüstite using high-temperature microscope and the reduction experiment were carried out under various conditions such as gas composition, temperature, and total pressure.

First, hydrogen reduction leads to the formation of porous structure of wüstite phase. Then, reduction to metallic iron proceeds concentrically, and porous iron forms. A part of metallic iron nuclei grows vertically to the wüstite surface, and iron whisker forms under specific conditions. Under atmospheric pressure conditions, higher temperature and higher hydrogen partial pressure make the surface porous, and iron whisker forms through 100%H₂ reduction at 900°C and 950°C. Increasing total pressure enlarges the formation area of iron whiskers to lower temperature. Furthermore, three-step fluidized bed reduction process was suggested to control the sticking phenomenon of raw materials.

Microsegregation in alloys has a significant effect on the steel properties. Numerical simulations using cellular automaton (CA) models are often used to predict microsegregation behavior. However, the CA model has problems quantitatively predicting the microsegregation behavior owing to the inaccuracy of the calculation of the interface curvature. In this study, we developed a two-dimensional quantitative CA model that incorporates interface curvature calculations using the height function (HF) method, correction of mass-balance errors, and solid–liquid interface movements caused by solidification and melting. In a simple calculation of the curvature of circles using the HF method, we confirmed that the HF method could calculate the curvatures of circles of various radii with an error of less than 1%. In addition, we performed simulations of the unidirectional solidification of a single dendrite of an Fe–C alloy using two CA models that implemented curvature calculation models using the conventional and HF methods, and investigated the effects of both models on dendrite morphology. The development of secondary dendrite arms was observed only in the CA model that implemented the HF method, thereby confirming the effect of the curvature calculation on dendrite morphology. Finally, we performed simulations of the multidendrite growth of an Fe–C alloy under continuous cooling at different cooling rates using the CA model that implemented the HF method. Consequently, the solute concentration in the solid exhibited an appropriate distribution following the lever rule, and the microsegregation behavior based on the cooling rate was reasonably simulated.

The formation of manganese sulfides, (Mn,Fe)S, during solidification of steel and subsequent cooling is a critical factor influencing the quality of continuously cast products, generally impacting hot tearing sensitivity and surface cracking on the strand. Understanding and predicting the formation of sulfides in the process is, therefore, essential for steelmakers. In this study, phase equilibria in two isopleth sections in the systems Fe-0.02C-0.50Mn-S and Fe-0.02C-2Mn-S (in wt.-%, S = 0.002–0.30 wt.-%) were experimentally characterized between 700°C and 1550°C. Alloys were produced in a high-frequency remelting (HFR) furnace. The morphology and distribution of (Mn,Fe)S inclusions in selected HFR samples were then examined via Scanning Electron Microscopy (SEM) coupled with Energy Dispersive X-Ray Analysis (EDX). High-temperature phase equilibria were analyzed by Differential Scanning Calorimetry (DSC) over the temperature range of 700 to 1550°C. To further elucidate the dissolution behavior of (Mn,Fe)S during the melting, phase transformations in a selected sample were observed in situ using High-Temperature Laser Scanning Confocal Microscopy (HT-LSCM). A CALPHAD-type thermodynamic database for the Fe–C–Mn–S system was created based on literature assessments and used to perform thermodynamic calculations of (Mn,Fe)S stability, showing excellent agreement with the DSC data. The critically evaluated database was used to derive the analytical equilibrium constant for the stability of stoichiometric manganese sulfide, (MnS) → [Mn] + [S], in equilibrium with the liquid, δ-ferrite (BCC), and austenite (FCC) phases.

When using steel scraps containing Cr and Mn as raw materials in electric arc furnace (EAF) steelmaking, EAF steelmaking slag containing Cr₂O₃ and MnO is inevitably produced as a by-product. Since Cr and Mn are valuable elements, it is economically and environmentally important to recover by reduction of EAF slag and reuse those elements into steelmaking process as alloying materials. In this study, to improve the recovery efficiency, flux addition to EAF slag was investigated both thermodynamically and experimentally. At first, thermodynamic calculations were performed using Al₂O₃, CaF₂ and SiO₂ as candidate fluxes, and it was found that SiO₂ has the greatest potential to increase the slag liquid fraction and to improve the recovery efficiency. Then, reduction experiments using an actual EAF slag containing 8 mass% Cr₂O₃ and 6 mass% MnO were carried out at 1573 K with SiO₂ flux addition. As a result, slag liquid fraction and the recovery efficiency of Cr and Mn increased with increasing the amount of SiO₂ addition. Up to 81% of Cr and 61% of Mn in an EAF slag could be recovered. Furthermore, reduction experiments using different reducing agents were conducted. Despite the slag liquid fraction being the same, differences in recovery efficiency were observed. It was suggested that not only slag liquid fraction but also liquid fraction of reduced metal is necessary to improve the recovery efficiency.

A three-dimensional (3D) mathematical model was established, coupling the large eddy simulation (LES) turbulence model, heat transfer model, solidification model, discrete phase model (DPM), and dynamic mesh model. Based on the actual continuous casting (CC) end process, the numerical simulation was divided into 4 stages. A method for calculating the motion velocity of the dynamic wall and the solidified shell was proposed to achieve the coupled simulation of transient flow, heat transfer, and solidification. In stage 1, molten steel speed and jet depth decreased as casting speed reduced, while the shell thickness steadily increased. In stage 2, after the submerged entry nozzle (SEN) removal, the molten steel speed and temperature dropped rapidly, and the circulation flow dissipated. In stage 3, the remaining molten steel region gradually decreased, and complete solidification occurred at approximately 4647.5 s, with the final solidification position located about 0.28 m below the end of the last slab. Based on the actual CC end stages and the entrapped positions of inclusions, a method for calculating the actual positions of inclusions was introduced to predict the 3D spatial distribution of inclusions in a CC end slab. The normalized number (NN) of inclusions exhibited a fluctuating downward trend with the increasing distance below the end of the last slab. It was recommended to cut the last slab at 4 m to ensure cleanliness, while the 3D normalized number density (NND) of inclusions in regions beyond 7 m below the end of the last slab reached a lower stable value.