https://journal.issiran.com/ International Journal of Iron & Steel Society of Iran en daily 1 Sat, 22 Nov 2025 00:00:00 +0330 Sat, 22 Nov 2025 00:00:00 +0330 https://journal.issiran.com/article_736136.html Iron ore pellets have emerged as the dominant burden material in modern ironmaking due to their uniform quality, superior mechanical strength, and enhanced transportability. The induration process, consisting of drying, preheating, firing, and cooling stages, is critical to achieving these desirable properties, with the firing stage being particularly important for pellet recrystallization and sintering. Among the key reactions, magnetite oxidation not only strengthens internal bonding by producing hematite but also contributes to the furnace's thermal balance. Ensuring complete conversion is essential to minimize residual FeO, which directly affects pellet reducibility and cold crushing strength (CCS). Pilot-scale packed-bed tests are commonly employed to simulate industrial firing conditions. However, these experiments can show high variability due to radial non-uniformity in voidage and gas flow. In the present study, experimental data revealed that, at corresponding heights, wall-adjacent pellets reached temperatures on average 60 °C higher than pellets in the center during the firing stage, with differences persisting for up to 23 minutes. This thermal disparity led to a higher conversion fraction near the walls by approximately 3–6 %, as determined by residual FeO measurements. Moreover, CCS tests on 35 carefully selected pellets indicated that lateral pellets exhibited strengths between 14 and 56 kg/pellet, higher than those of their central counterparts, depending on bed height. Such discrepancies are often overlooked in one-dimensional models, which assume uniform voidage and flow. In contrast, two-dimensional axisymmetric CFD modeling offers a more accurate representation by incorporating radial variations in voidage and accounting for magnetite oxidation reaction kinetics. In this study, the developed CFD model accurately reproduced the measured temperature profiles and conversion fractions, with a maximum deviation of less than 5 % from the pilot-scale experimental data. These findings underscore the importance of wall effects in small-to-medium induration setups, where the affected zone can account for a substantial portion of the bed volume. The distinct thermal histories of pellets in lateral versus central positions can mislead average property assessments if inappropriate sampling is employed. Therefore, integrating quantitative understanding of radial gradients into both modeling and operational practice is essential for accurate process optimization, scale-up, and quality control in iron ore pellet production. https://journal.issiran.com/article_736344.html This study investigates the application of machine learning, specifically Support Vector Machines (SVM), to predict the metallurgical length in continuous casting. The metallurgical length, defined as the distance from the molten metal surface to the point of complete solidification, significantly impacts product quality. Traditional methods for predicting metallurgical length, such as the K-factor model and numerical simulations, face limitations in accuracy, computational cost, and adaptability to real-time industrial applications. To address these limitations, this study proposes a novel approach using Support Vector Machines (SVM), a machine learning algorithm, to predict metallurgical length with high precision. Numerical simulations were conducted to model fluid flow, heat transfer, and solidification processes, validated against experimental data. The SVM model was trained on metallurgical length data derived from simulations at various casting speeds. Results demonstrated that the SVM model achieved a mean square error (MSE) of 0.0789 compared to numerical data, significantly outperforming empirical methods (MSE = 0.5353). The study highlights the potential of machine learning to enhance real-time decision-making in continuous casting, offering a computationally efficient and accurate alternative to traditional methods. This approach can be extended to analyze other process parameters, such as cooling water flow rate and initial superheat temperature, further optimizing steel production.