Einstein-Roscoe Regression for the Slag Viscosity Prediction Problem in Steelmaking

In classical machine learning, regressors are trained without attempting to gain insight into the mechanism connecting inputs and outputs. Natural sciences, however, are interested in finding a robust interpretable function for the target phenomenon, that can return predictions even outside of the training domains. This paper focuses on viscosity prediction problem in steelmaking, and proposes Einstein-Roscoe regression (ERR), which learns the coefficients of the Einstein-Roscoe equation, and is able to extrapolate to unseen domains. Besides, it is often the case in the natural sciences that some measurements are much more expensive than the others due to physical constraints. To this end, we employ a transfer learning framework based on Gaussian process, which allows us to estimate the regression parameters using the auxiliary measurements available in a reasonable cost. In experiments using the viscosity measurements in high temperature slag system, ERR is compared favorably with various machine learning approaches in interpolation settings, while outperformed all of them in extrapolation settings. Furthermore, after estimating parameters using the auxiliary dataset obtained at room temperature, increase in accuracy is observed in the high temperature dataset, which corroborates the effectiveness of the proposed approach.

Effect of Fe/SiO2 Ratio and Fe2O3 on the Viscosity and Slag Structure of Copper-Smelting Slags

Secondary copper smelting is an effective means of treating waste resources. During the smelting process, the viscous behavior of the smelting slags is essential for smooth operation. Therefore, the effects of Fe/SiO2 ratio and Fe2O3 contents on the viscous behavior of the FeO−Fe2O3−SiO2−8 wt%CaO−3 wt%MgO−3 wt%Al2O3 slag system were investigated. The slag viscosity and activation energy for viscous flow decrease with increasing Fe/SiO2 from 0.8 to 1.2, and increase as the Fe2O3 content increases from 4 wt% to 16 wt% at Fe/SiO2 ratio of 1.2. However, under the conditions of Fe/SiO2 of 0.8 and 1.0, the viscosity and activation energy for viscous flow show a minimum value at Fe2O3 content of 12 wt%. Fe2O3 exhibits amphoteric properties. In addition, the increase in Fe2O3 content raises the breaking temperature of the slag, while the Fe/SiO2 ratio has the opposite effect. Fourier transform infrared spectroscopy (FTIR) and Raman spectroscopy show that increases in Fe/SiO2 ratio lead to simplification of the silicate network structure, while increases in Fe2O3 content improves the formability of the network. This study provides theoretical support for the related research and application of secondary copper smelting.

Thermodynamic modeling of iron recovery processes

One of the promising directions in metallurgy is the use of iron-containing waste, such as converter production sludge, iron-containing concentrates, rolling scale, iron ore processing waste and others. Development of new resource-saving technologies using such waste requires preliminary research and accumulation of information in the field of iron recovery. The paper considers the processes of iron recovery from oxides under various conditions. The authors used the method of thermodynamic modeling based on the search for the entropy maximum. The thermodynamic modeling tool was TERRA software package created at the Bauman Moscow State Technical University. TERRA complex is designed to calculate the thermodynamic properties and composition of the phases of equilibrium state of arbitrary systems with chemical and phase transformations. Using this software package, studies of the processes of iron recovery by various reducing agents (carbon, manganese, and silicon) in model thermodynamic systems were carried out, and optimal conditions for temperature and consumption of reducing agents were determined. The paper presents the results of a study of processes in the metal-slag system in equilibrium. The analysis of the metal-slag system equilibrium state was carried out for the temperature range of 1773 - 1973 K with different amounts of slag. Boundaries of the areas of redox processes were determined and the influence of metal components on conditions for iron oxides recovery from slag to metal was evaluated. The dependences of the system equilibrium composition on temperature at different ratios of metal and slag were obtained, as well as the optimal conditions for iron recovery.

Phase Equilibria in the System CaO-SiO2-La2O3-Nb2O5 at 1400 °C

CaO-SiO2-La2O3-Nb2O5 system is of great significance for the pyrometallurgical utilization of Bayan Obo tailing resources. In the present work, the phase equilibrium of this quaternary system at 1400 °C was determined by a thermodynamic equilibrium experiment. On the basis of the recently determined CaO-La2O3-Nb2O5 phase diagram, some boundary surfaces of primary phase fields of CaO-SiO2-La2O3-Nb2O5 phase diagram were modified; then, the 1400 °C isothermal surface in the primary phase fields of SiO2, CaNb2O6, Ca2Nb2O7, and LaNbO4 was constructed, respectively. On this basis, CaO-SiO2-Nb2O5 pseudo-ternary phase diagrams with w(La2O3) = 5%, 10%, 15%, and 20% were determined, respectively. Considering the importance of equilibrium crystallization reaction type, we proposed a new rule named Tangent Line Rule to judge the univariant reaction type in the quaternary phase diagram. By applying Tangent Line Rule and Tangent Plane Rule previously proposed, some univariant and bivariant crystallization reaction types in the CaO-SiO2-La2O3- Nb2O5 phase diagram were determined, respectively. The current work can provide original data for the establishment of a thermodynamic database of Nb-bearing and REE-bearing slag system; the proposed Tangent Line Rule will promote the application of a spatial quaternary phase diagram.

Evaluating the Effect of MgO/Al2O3 Ratio on Thermal Behaviors and Structures of Blast Furnace Slag with Low Carbon Consumption

In order to clarify the effect of the MgO/Al2O3 ratio on the fluidity of a low-alumina blast furnace slag system, the influence law of slag fluidity with different MgO/Al2O3 ratios was studied based on the composition of blast furnace slag through a viscosity experiment and themodynamic software. By studying the effect of the MgO/Al2O3 ratio on the activation energy of viscous flow of slag combined with FT-IR, the effect of the MgO/Al2O3 ratio on the thermal-stability of low-aluminum slag was interpreted from the microstructure level. Results indicated that the viscosity and the melting temperature of slag both showed a gradual downward trend due to the increase of the MgO/Al2O3 ratio. Besides, the temperature stability of the low aluminum slag became more stable due to the depolymerization of the complex structure of slag. Considering the actual operating conditions of blast furnace, the MgO/Al2O3 ratio of slag was suggested to be controlled to 0.60 and the basicity to be no higher than 1.20 under the conditions of this investigation. Industrial test results showed that the coke rate could be saved as 3.49 kg/t when the MgO/Al2O3 ratio decreased from 0.70 to 0.58.

A SULPHIDE CAPACITY PREDICTION MODEL OF BLAST FURNACE SLAG

A sulphide capacity prediction model of CaO-SiO2-MgO-Al2O3 slags has been developed based on the ion and molecule coexistence theory(IMCT). Sulphide capacity(Cs) of slag for blast furnace (BF) with high Al2O3 in the CaO-SiO2-MgO-Al2O3 system at 1773K were measured by applying slag-metal equilibrium method. The feasibility of the developed IMCT model is verified by the sulphide capacity measured in the experiment. Effects of R(w(CaO)/w(SiO2)), w(MgO)/w(Al2O3) and w(Al2O3) on sulphide capacity were discussed. There is a good linear relationship between the experimental value and the predicted value. Therefore, the theoretical model of ion and molecule coexistence can be used to calculate the sulphide capacity of CaO-SiO2-MgO-Al2O3 quaternary slag system. When w(Al2O3)=20% and w(MgO)/w(Al2O3)=0.5, the sulphide capacity of slag increased with the increase of R. When w(Al2O3)=20% and R=1.30, the sulphide capacity of slag increased with the increase of w(MgO)/w(Al2O3). When R=1.30 and w(MgO)/w(Al2O3)=0.4, the sulphide capacity of slag decreased with the increase of w(Al2O3).

Optimizing Slag Content to Control Ds-Type Inclusions in 10B21 Cold Heading Steel

Ds-type inclusions during production are an important factor affecting the performance and quality of manufactured 10B21 steel. To minimize Ds-type inclusions in steel and improve the production qualification rate of steelmaking plants, a refining slag system optimization scheme was proposed based on the analysis of current inclusion evolution during the steelmaking process, and industrial tests were conducted to verify improvements resulting from application of the proposed scheme. The results showed that the composition of Ds inclusions in 10B21 steel are mainly CaO–Al2O3–MgO–CaS–TiN, which exists in the form of calcium–magnesium aluminate coated with titanium nitride and calcium sulfide. The main reason for the formation of Ds inclusions is the poor fluidity of the refining slag and its low capability to absorb inclusions. The poor coverage of the refining slag on the molten steel during refining can easily cause secondary oxidation of the molten steel. Thus, the formation and growth of Ds-type inclusions are aggravated after the calcium feeding line and soft blowing operation. Here, we propose to minimize Ds inclusions using our optimized refined slag system. The mass percentage of the optimized slag system is CaO: 55–60%, Al2O3: 20–35%, SiO2: 3–7%, MgO: 4–8%, (MnO + FeO) < 1%, and the basicity is controlled within the range of 7–11. We observed that our optimized refining slag system has a significantly improved ability to remove inclusions, particularly Ds inclusions, which improves the qualification rate of 10B21 steel.