This study develops a novel framework that integrates numerical simulation, machine learning, and operating diagram to predict the production state of the hydrogen-based shaft furnace. The aim is to accurately describe and rapidly predict furnace performance and its deviation from the ideal point. A validated computational fluid dynamics (CFD) model of the shaft furnace that enable to calculate the operating condition was employed to generate a comprehensive database comprising 500 simulation cases under varying operating parameters (reducing gas flow rate, temperature, and composition). This database was then used to train and compare multiple machine learning (ML) models to predict the key internal reaction rates and equilibrium parameters necessary for constructing the operating diagram. An optimized ensemble of ML models achieved high prediction accuracy, with the R2 exceeding 0.986 for all targeted parameters. The framework successfully predicts the critical deviation (w value), which measures the distance between the actual operating line and the ideal point in the diagram. ML analysis reveals that reducing gas temperature and flow rate exert the most significant influence on this deviation. Unlike conventional numerical simulations requiring high-performance computers and 1 h of computation plus data processing per case, the presented machine learning-based approach generates outputs in under a minute from simple condition inputs, achieving orders-of-magnitude improvement in computational efficiency while maintaining high accuracy, demonstrating superior convenience, flexibility, and computational speed. The developed model provides a promising approach for achieving real-time prediction and optimization of shaft furnace performance.

Mineral analysis is a fundamental task in geological exploration and resource development that provides critical technical support for ore-genesis identification,resource evaluation,and process-parameter optimization.However,conventional mineral analysis methods rely significantly on expert knowledge and specialized instrumentation,thus resulting in high costs,low efficiency,and limited applicability in complex field environments.Moreover,their dependence on single-modal data restricts their performance in fine-grained mineral recognition.Recent advances in multimodal large-language models(MLLMs)have introduced new possibilities for mineral analysis by enabling a unified understanding of visual and textual information using advanced image encoders and cross-modal alignment mechanisms.Whereas MLLMs have shown promising results in domains such as education,healthcare,and geology,general-purpose models remain inadequate in mineral analysis,including insufficient domain-specific knowledge,weak generalization in fine-grained mineral image recognition,and limited capability to generate professional and structured geological reports.Hence,this study proposes MineralMLLM,which is a multimodal mineral-analysis system developed based on Qwen2.5-VL-a state-of-the-art vision-language model optimized for Chinese scenarios with native support for high-resolution dynamic image processing,precise spatial grounding,and robust multimodal document understanding.To enhance domain adaptability,the model was fine-tuned on a self-constructed mineral image-text dataset comprising approximately 10,000 samples across 20 mineral categories.Two parameter-efficient fine-tuning strategies,i.e.,Low-rank adaptation(LoRA)and Infused adapter by attention(IA3),were employed and systematically compared.The dataset was obtained from multiple sources,including mineralogy textbooks,academic literature,and online resources;subsequently,it was subjected to manual verification,data augmentation,and stratified random splitting in a training:validation:test ratio of 7∶1∶2 to ensure data quality and representativeness.Furthermore,retrieval-augmented generation(RAG)was integrated to incorporate domain-specific knowledge and establish a complete training-retrieval-inference pipeline.The RAG module adopts a hybrid retrieval strategy that combines dense vector retrieval(weight,0.6)and BM25-based sparse retrieval(weight,0.4),along with semantic chunking optimized at a threshold of 0.82 to balance between semantic coherence and retrieval efficiency.A lightweight web-based interactive system was implemented using Vue.js and Flask to support mineral-image upload,semantic recognition,and structured result visualization.Experimental results show that fine-tuning via LoRA and I A3 improved the BERT score by approximately 10％and 1％,respectively,compared with that of the base model.LoRA achieved superior performance owing to its larger trainable parameter capacity(approximately 190 million parameters,which constitute 2.24％of the total model parameters)and stronger feature adaptation capability.When combined with RAG,the LoRA-enhanced model further improved the BERT score by an additional 10％(reaching 0.806),with significant gains in terms of the bilingual evaluation understudy(BLEU)(0.4786 vs.0.4151)and ROUGE-fl(0.5755 vs.0.2258).These improvements significantly enhance both the mineral-identification accuracy and the generation of professional structured descriptions,including mineral classification,characteristic analysis,and reference citations.Ablation studies and robustness evaluations confirm the effectiveness and stability of MineralMLLM under challenging conditions,including blurred images,low-light environments,and partial occlusions.The model consistently outperformed baseline models under these conditions.Additionally,semantic chunking threshold analysis(τ ∈[0.70,0.90])indicates that τ=0.82 achieved optimal performance by balancing between chunk granularity and semantic integrity.In conclusion,MineralMLLM effectively bridges domain-specific geological knowledge with the general reasoning capabilities of MLLMs,thus providing a scalable and practical solution for intelligent mineral analysis.The proposed framework not only advances mineral phase identification but also offers a transferable technical paradigm for deploying large multimodal models in other professional and industrial domains.

Si3N4-SiC composite ceramics are attractive for high-temperature applications in hydrogen metallurgy due to their mechanical strength and oxidation resistance. This work investigates the microstructural evolution and kinetic mechanisms under pure CO and 50 vol% H2-50 vol% CO atmospheres between 600 and 1300 degrees C. Under pure CO atmosphere, oxidation followed gas-phase diffusion-controlled parabolic kinetics, where the rate was governed by CO diffusion through the porous SiO2-rich product layer, with an activation energy of 26388.211 J/ mol. Under H2-CO mixed atmospheres, the corrosion involved coupled oxidation-reduction reactions. Below 1000 degrees C, kinetics were dominated by gas-phase diffusion of CO through the product layer coupled with interfacial surface reactions, with an activation energy of 61017.652 J/mol. Above 1000 degrees C, hydrogen reduced SiO2 to volatile SiO, and the overall kinetics were controlled by coupled gas-phase diffusion and reduction-driven volatilization processes, leading to porous product layers and an activation energy of 99533.554 J/mol. The results clarified the competing roles of CO oxidation and H2 reduction in Si3N4-SiC composite ceramics degradation and provide kinetic models for predicting performance in reducing atmospheres. These findings provide theoretical guidance for the design and optimization of Si3N4-SiC ceramics for service in hydrogen-based steelmaking and other reducing environments.

To enhance the utilization efficiency of industrial byproducts, a synergistic carbothermal reduction process was developed for the synthesis of ferrosilicon alloy from fly ash and red mud. The reaction mechanisms were systematically analyzed through thermodynamic calculations. The effects of key parameters, including the iron-to-silicon molar ratio (n((Fe))/n((Si))), carbon-to-oxygen molar ratio (n((C))/n((O))), roasting temperature, and holding time on ferrosilicon formation were investigated using X-ray diffraction, scanning electron microscopy, and energy-dispersive spectroscopy. The results indicate that the ferrosilicon phase emerges when the n((Fe))/n((Si)) ratio decreases to 1.01 and remains detectable at 0.76. Within this range, an n((C))/n((O)) ratio between 0.33 and 0.48 is required to ensure complete reduction. At a fixed holding time of 2 h, ferrosilicon formation commences at 1723 K with increasing roasting temperature from 1673 K to 1823 K. Lower temperatures hinder the complete reaction between metallic Fe and mullite. Although higher temperatures (up to 1823 K) do not induce significant phase transformations, they markedly reduce residual carbon content. At a constant temperature of 1723 K, extending the duration from 1 h to 2 h or longer promotes ferrosilicon formation, with a duration of 4 h substantially reducing the residual carbon, indicating a more complete reaction. The optimal synthesis conditions were determined as follows: n((Fe))/n((Si)) = 0.76-1.01, n((c))/n((O)) = 0.33, roasting temperature of 1723-1823 K, and holding time 2-4 h. In particular, satisfactory results were achieved either at 1823 K for 2 h or at 1723 K for 4 h. Combined thermodynamic and experimental analyses revealed the primary reaction pathway as follows: Al2O3 + SiO2 -> mullite, and Fe2O3 -> Fe3O4 -> FeO -> Fe, followed by Fe+mullite -> Fe3Si; SiO2 + Fe -> Fe3Si; Nosean -> NaAlSiO4 + Na2SiO3, followed by NaAlSiO4 / Na2SiO3 + Fe -> Fe3Si.

This study uses iron–carbon dust in place of iron ore powder to prepare ferrocoke. A gasification experiment is conducted and several kinetic models are used to simulate the ferrocoke gasification process, aiming to clarify how iron–carbon dust affects its kinetics. The microstructure of ferrocoke and the degree of carbon structure ordering are characterized using scanning electron microscopy, X‐ray diffraction (XRD), and Raman spectroscopy. Gasification data show that the addition of iron–carbon dust reduces the initial gasification temperature of ferrocoke. Kinetic fitting results reveal that iron–carbon dust alters the reaction‐controlling step from chemical reaction limitation to internal pore diffusion control while catalyzing the reaction. This confirms that the gasification process is influenced by pore/characteristics, aligning with the RPM model. XRD and Raman spectroscopy analyses further show that as the proportion of iron–carbon dust increases, the structural orderliness of the iron–carbon matrix decreases, while the amount of amorphous carbon increases. This structural transformation significantly enhances the gasification activity of the iron–carbon matrix.

Tap hole clay requires binders that provide sufficient plasticity, mechanical reliability across broad temperatures, and resistance to physical impact during blast furnace operation. Coal tar offers favorable thermoplasticity but raises environmental concerns, whereas phenolic resin provides high residual carbon yield but inadequate workability. This study examines the complementary functions of coal tar-phenolic resin composite binders and their influence on the performance and microstructure of tap hole clay. The composite system exhibits a more balanced property profile than single-binder formulations and reduces coal tar usage by approximately 20%-25% compared with traditional coal-tar-only binders. When resistance to physical impact is the primary requirement, a 3:1 coal tar-thermosetting resin ratio achieves the highest strength, with a Marshall value of 1.650 MPa, dried Cold crushing strength of 24.65 MPa, and sintered Cold crushing strength of 13.91 MPa. Microstructural analysis shows that improved pore refinement is closely associated with enhanced mechanical properties. Additionally, carbonized structures observed after heat treatment suggest the possible coexistence of lamellar carbon from coal tar and amorphous carbon domains from phenolic resin. These findings provide guidance for designing high-performance composite binders for tap hole clay.

As a key cooling component of the blast furnace, the dominant failure mechanism of copper cooling staves remains controversial, primarily focusing on two pathways: friction wear by burden and erosive wear by gas. This study employs a systematic approach combining the analysis of actual failure morphologies of copper cooling staves from blast furnaces with friction-wear tests and airflow erosive tests to clarify the primary controlling mechanism of copper cooling stave wear and failure. The results indicate that the friction-wear morphology aligns more closely with actual failure characteristics, suggesting that mechanical abrasion by the burden is the main cause of thickness reduction and structural failure in copper cooling staves. High-temperature friction-wear tests confirm that the wear mechanism of copper cooling staves under burden action is predominantly adhesive wear, accompanied by distinct abrasive wear characteristics, with the abrasive wear effect becoming more pronounced as temperature increases. With rising temperature, the wear volume of copper cooling staves increases from 207 to 444 mu m(3), while the wear rate increases from 8.6 to 18.5 mu m(3)/(N m). This study clarifies the controversy regarding the failure mechanism of copper cooling staves and holds significant guiding implications for the development of protective technologies for copper cooling staves.

Coal pyrolysis is a critical step in its clean and efficient utilization. This study employed a combination of thermogravimetric analysis (TGA), pyrolysis-gas chromatography/mass spectrometry, and ReaxFF molecular dynamics (MD) simulations to systematically reveal the pyrolysis kinetics and reaction mechanisms of bituminous coal. TG experiments provided data on mass change and kinetic parameters, while ReaxFF MD simulations elucidated dynamic processes at the atomic scale, such as chemical bond cleavage, product formation, and structural evolution. The results show that the weight-loss curves obtained from ReaxFF MD simulations are in excellent agreement with the experimental thermogravimetric data (R & sup2; > 0.76). Kinetic analysis indicates that the activation energies calculated by the Flynn-Wall-Ozawa and Kissinger-Akahira-Sunose methods exhibit consistent trends with the conversion rate. The activation energy obtained from ReaxFF calculations is higher than the experimental value, primarily due to the picosecond timescale used in the simulations. The study also found that both slower non-isothermal heating rates and higher isothermal reaction temperatures promote the generation of H-2 and CH4, and the results indicate that their generation predominantly occurs via radical recombination. Pyrolysis begins with the cleavage of C-H and C-C bonds, followed by aromatization and condensation reactions. Notably, lean coal demonstrated greater thermal stability due to its higher initial aromatic content. This macro-micro framework validates ReaxFF MD for simulating complex pyrolysis, supporting atomic-scale clean coal technology design.

Using hydrogen as a reducing agent can significantly reduce the carbon emissions in the steel industry. Research on H-2 reduction and foaming of FeO slag is limited. Unlike prior numerical studies that only simulated inert gas injection (e.g., Ar) without chemical reaction, this study develops a 3D model that couples H-2-FeO reduction kinetics with population balance modeling (PBM) for foaming. The model is validated against experimental data (discrepancy <4.4%). The results show that after injecting hydrogen into the slag for 10 min, vortices form in the liquid, the unreacted FeO is transported to the reaction interface. The presence of vortices is associated with enhanced mixing and lower local H2O concentrations near the reaction interface, which may facilitate the forward reaction. When the hydrogen injection time was extended from 10 to 30 min, the average height of the foam phase increased from 0.531 to 0.645 cm, the volume fraction of the foam phase rose from 0.54 to 0.83, the foaming index rose from 0.512 to 0.622, the average diameter of the bubbles rose from 0.86 mm to 2.57 mm. A practical strength of this validated model is its ability to predict foaming index and foam height for different slag compositions and gas flow rates without costly high-temperature experiments, accelerating the design of hydrogen-based smelting processes.

Low-silicon smelting is a critical approach to achieving low-carbon and high-efficiency blast furnace ironmaking. However, its implementation still faces considerable technical challenges. In this study, the behavior of silicon across different zones of the blast furnace is systematically analyzed, and corresponding process optimization strategies are proposed and validated through industrial trials on a commercial blast furnace in China. The results show that increasing slag basicity reduces the activity of SiO2, promoting the transfer of silicon from molten iron into the slag phase. A higher carbon content increases the activity coefficient of silicon, enhancing the reduction of SiO2. In the tuyere zone, SiO2 is initially reduced to gaseous SiO, which is subsequently reduced to silicon upon contact with molten iron. Lowering the theoretical combustion temperature at the tuyere effectively suppresses SiO2 reduction. In addition, reducing the height of the dripping zone limits the reaction time for silicon generation. Industrial application of the proposed strategies led to a 0.08 wt% reduction in silicon content in hot metal, an 11.12 kg/t decrease in fuel rate, and an estimated reduction of 96,900 tons of CO2 emissions annually. These findings provide an effective pathway for improving the sustainability and operational efficiency of blast furnace ironmaking.

In high-temperature smelting, the thermal stability of slag-defined as its integrated ability to maintain stable physicochemical properties under temperature fluctuations—is essential for process continuity, energy efficiency, refractory protection, and subsequent utilization. However, the multicomponent and multiphase nature of slag leads to complex thermochemical responses, and no unified evaluation framework currently exists. This review summarizes the major indicators used to assess slag thermal stability and clarifies their applicability and limitations. Key factors affecting stability, including chemical composition, phase evolution, microstructural features, and operating conditions, are critically examined. Strategies for improving slag thermal stability are further discussed from the viewpoints of component design, process optimization, advanced characterization, predictive modeling, and functional utilization. The aim is to provide a clearer conceptual basis and practical guidance for enhancing slag performance in high-temperature metallurgical systems.

Decarbonizing the iron and steel industry has become critically urgent, and one promising strategy to address this challenge is to adjust the proportion of pellets in the burden structure, replacing sinter with pellets. This study investigates the impact of varying pellet proportions on blast furnace burden distribution using Discrete Element Method (DEM) simulations. Key findings reveal that increasing the proportion of pellet results in a decrease in the stockpile's angle of repose, while the porosity of the stockpile rises. Specifically, as the pellet proportion increases from 30% to 90%, the angle of repose decreases from 38.66 degrees to 26.57 degrees, and the porosity increases from 32.66% to 35.93%. Additionally, the segregation of pellet in the radial direction initially increases, then decreases, while segregation of sinter steadily declines as its proportion increases. Increasing the proportion of pellet will effectively improve the permeability of the burden, thus improve the airflow distribution and increasing the furnace efficiency.

High-pellet-ratio smelting is regarded as an important pathway for green blast furnace ironmaking, offering significant potential for reducing raw material consumption and carbon emissions. However, variations in slag properties under high pellet ratios exert a pronounced influence on process stability and hot metal quality. In this study, typical pellet SiO2 contents (3.2-4.2 wt%) and pellet proportions (30-50 wt%) were investigated to quantitatively evaluate the effects of slag amount and composition on the comprehensive metallurgical performance of blast furnace slag. The results show that increasing the pellet ratio from 30 to 50 wt% reduces slag generation by approximate to 15%. Pellets with a SiO2 content of 3.6 wt% provide a balanced control of slag quantity and viscosity. As slag basicity increases from 1.10 to 1.30, the sulfur partition ratio rises from 12.8 to 24.7, significantly enhancing desulfurization performance under high-pellet conditions. Increasing the MgO content to 10 wt% reduces slag temperature fluctuations to 64 degrees C, while further increasing the slag amount to 330 kg/t suppresses temperature variation to within 10 degrees C. From an engineering perspective, maintaining slag basicity at 1.10-1.15 with an MgO content of 8-9 wt% effectively stabilizes slag fluidity and enhances thermal buffering capacity under high-pellet-ratio operation. When the pellet proportion is increased to 50 wt%, the theoretical blast furnace fuel rate decreases by 25 kg/t, corresponding to an annual carbon emission reduction of 146,000 t. These findings provide both theoretical insights and industrial guidance for low-carbon, high-efficiency blast furnace smelting with high pellet ratios.

This study investigated the mechanical properties of briquettes (CDB) through uniaxial compression tests, and CT was used to quantitatively analyze the porosity, particle structure, and carbon matrix. The relationship between porosity, particle characteristics, and CDB compressive strength was revealed, along with the structural evolution of three-dimensional pores in ferro-coke during gasification. The effects of porosity and particles on CDB compressive strength and the consumption of carbon-based components in ferro-coke during gasification were also analyzed. The results show that with the increase of carbonization temperature, the effect of pores on the strength of ferro-coke is mainly at 500 degrees C and 800 degrees C. With the increase of dust ratio, the porosity increases from 6.52 % to 25.01 %, the pore sphericity and large particle dust also increase accordingly, the critical fracture stress decreases, and the compressive strength decreases from 4826.778 N to 4060.227 N. In addition, with the gasification reaction of ferro-coke, the open porosity of ferro-coke gradually increases from 0.99 % to 10.20 %, and CO2 can quickly diffuse into ferro-coke, and the carbon matrix of ferro-coke was rapidly consumed. At the same time, the iron, calcium and other metal elements in the dust particles will play a catalytic role in promoting the reaction of the surrounding carbon matrix with CO2, resulting in faster consumption of the carbon matrix around the dust particles and increased porosity.

Ductile iron cooling staves are critical cooling equipment essential for ensuring the safety of the blast furnace. However, Ductile iron cooling staves are prone to fracture under thermal shock, compromising operational safety. Samples were taken from a blast furnace's spherical graphite cast iron cooling stave to investigate damage mechanisms during service. The results indicate that after high-temperature service, a network of intersecting cracks developed on the surface of the ductile iron cooling stave. The maximum thinning of the stave wall reached 140 mm, accounting for 58% of the original thickness. Results from metallographic microscopy and micro-CT analysis revealed significant differences in the graphite morphology across different regions of the cooling stave. The tensile strength of the cast iron stave decreased by 14.8%-31.4% after service. Through analysis of the temperature field of the cast iron cooling stave, it is believed that the location of cracks is related to the thermal stress experienced by the cast iron cooling stave. Crack initiation is primarily induced by regions of high stress concentration around nonspheroidal graphite, as well as debonding between spheroidal graphite and the metal matrix. Crack propagation occurs primarily through the graphite phases and extends intergranularly within the metal matrix.

Straight-grate induration of iron-ore pellets is a strongly coupled moving-bed thermal process involving alternating gas-flow directions, cross-zone heat recovery, pressure-driven flow, and temperature-dependent heat transfer. Fast reconstruction of the bed thermal state is essential for operation-window analysis, whereas repeated numerical solution of conventional mechanistic models is costly over wide operating ranges. This work proposes a zone-wise parameterized physics-informed neural-network surrogate for one-dimensional straight-grate bed modeling. Bed thickness, grate speed, pellet diameter, and bed void fraction are used as parametric inputs and are embedded into the heat-transfer correlation, Ergun pressure-drop constraint, and cross-zone continuity treatment. The surrogate reconstructs gas and solid temperature fields as well as the volumetric gas-solid heat-transfer coefficient. Under sparse supervision, it achieves higher sample efficiency and better local fidelity near process-zone transitions than purely data-driven baselines, with the mean solid-temperature error remaining within approximately 2 K when 500 or more samples per zone are used. The trained surrogate is further coupled with a four-variable multi-objective optimizer to balance productivity, bed thermal non-uniformity, and top-layer overheating risk. The median times for temperature-history prediction, single candidate evaluation, and pseudo-two-dimensional field reconstruction are 3.91 s, 3.94 s, and 5.10 s, corresponding to speed-ups of 33.08 & times;, 32.81 & times;, and 25.36 & times; over the refined-grid finite-difference solver. Considering the multi-hour cost of a complete NSGA-II workflow, the proposed framework is positioned for rapid thermal-state screening and offline operation-window decision support.

The increasing utilization of low‐grade iron ores poses significant challenges to the stable and efficient operation of blast furnaces. The gangue components in the ore affect slag fluidity and the permeability of the burden to gas and liquid. In this study, the coupled influence of slag properties, burden dripping behavior, and blast furnace performance was systematically investigated based on the principles of low‐grade burden smelting and long‐term industrial verification. The results show that with a burden ratio of 78% sinter + 14% pellet + 8% lump, the optimal coupling between the molten layer thickness and gas flow resistance is achieved, resulting in the minimum characteristic value S and improved gas–liquid permeability in the softening and melting zone. Increasing Al 2 O 3 content significantly increases slag viscosity and reduces superheating, thus narrowing the thermal operating window. When the Al 2 O 3 content is above 16 wt%, adding an appropriate amount of TiO 2 up to 1.5 wt% can effectively compensate for the adverse effects of Al 2 O 3 . Applying the proposed slag and burden design scheme in a commercial blast furnace showed that when the ore grade decreased from 57.5% to 56.5%, the blast furnace maintained stable operation with good permeability, lower fuel consumption, and higher production efficiency. This provides guidance for the efficient and stable smelting of Al 2 O 3 ‐TiO 2 ‐containing low‐grade iron ore in modern blast furnace ironmaking.

This study investigates the effects of flow distributor placement and loosener configuration on particle-flow behavior in a hydrogen-based direct reduction shaft furnace using the discrete element method (DEM). A three-dimensional industrial-scale furnace model based on a MIDREX-type geometry was established, and four representative structural configurations were examined by varying the flow distributor position and loosener setting. The results show that flow distributor placement is the dominant factor controlling particle descending behavior and particle-flow uniformity. When the flow distributor was located in the cooling zone, the flow uniformity index reached 0.875, which was 40.9% and 20.9% higher than those for the transition-cooling interface and transition-zone configurations, respectively. Particle trajectory analysis indicates that the effect of flow distributor position is mainly confined to the region above the device, with limited influence on the lower burden trajectory. Although the loosener has little effect on particle-flow uniformity, it significantly suppresses particle degradation. Under the transition-zone flow distributor configuration, the predicted powder formation ratio decreased from 3.89% to 2.97% after introducing the loosener, corresponding to a relative reduction of 23.7%. Overall, among the four representative configurations investigated in this study, positioning the flow distributor in the transition zone while retaining the loosener provides a more balanced compromise between burden-flow regulation and powder suppression for shaft furnace design and industrial operation.

In order to study the influence of tuyere structure parameters on the gas flow distribution of oxygen blast furnace (OBF), a three-dimensional full-scale CFD mathematical model is established based on the internal model parameters of 430 m³ OBF of Bayi Steel. Without changing the structural parameters of the hearth tuyere and the blast parameters, the influence laws of tuyere area, tuyere number and tuyere angle on the gas flow distribution are analyzed, and five evaluation indicators, namely D, H, V Horizontal, V Vertical and V Depth , are put forward for the initial gas flow of the shaft tuyere to evaluate the gas distribution state. The results show that increasing tuyere number or decreasing tuyere area can significantly enhance the gas velocity at the upper part of OBF and effectively improve the reducing atmosphere, but the shaft tuyere angle has little effect on the gas velocity. The structural parameters of tuyere mainly affect the initial gas flow pattern. Reducing tuyere diameter from 100 mm to 80 mm can form a more penetrating concentrated gas flow, and its velocity attenuation depth D is reduced from 0.23 m to 0.17 m. However, if the tuyere angle is increased from 3 to 7, the gas flow will penetrate downwards, which will expand the low-speed area at the lower part of tuyere. The five evaluation indicators put forward in this study can effectively quantify the initial gas flow characteristics, and provide an important theoretical basis for the optimization of shaft tuyere structure of OBF.