The main factor affecting the cleanliness of steel is inclusions, most of which are deoxidation products. The greater the oxygen content fluctuation at the converter's end point, the more unstable the inclusion control. A large amount of charging, operation, and sampling data is collected during the converter smelting, but it is not fully utilized, and the deoxidizing alloy addition still relies on manual experience. Herein, the representative convolutional neural network (CNN) of deep learning is adopted, and the number of convolutional layers, convolutional kernel size, and the number of convolutional kernels are optimized to establish the best prediction model of the converter end point. Its root mean square error, mean absolute error, and mean absolute percentage error are 35.29%, 25.59%, and 7.30%, respectively, which are superior to the back propagation neural network. This CNN prediction model uses 1300 sets of preprocessed data containing 23 indicator variables, of which 1200 sets of data are used for training and 100 sets for model validation. Within the ±50 ppm error scope, the model prediction hit rate can reach 87%, and when within the ±70 ppm error scope, the hit rate is 93%.

With the use of high phosphorus iron ore, there is a large amount of high phosphorus steel slag formed, which is difficult to handle. How to separate the elemental phosphorus has become a key issue in the secondary utilization of steel slag. Experiments found that there were distinct phosphorus-rich phases, iron-rich phases and matrix phases in the high-phosphorus steel slag cooled with the furnace. In this study, the effects of heat treatment conditions and slag basicity on the P2O5 content, as well as the size of the phosphorus-rich phase were investigated. Taking all factors into consideration, the optimal experimental conditions were determined as the holding temperature and time of 1350 °C and 60 min, respectively, and the slag basicity of 1.8. At this time, the P2O5 content in the phosphorus-rich phase reached 24.2%, and the average size of the phosphorus-rich phase was 63.51 μm. The phosphorus-rich phase is separated by crushing and magnetic separation for making phosphate fertilizer, and the residual steel slag is used again for steelmaking, which enables the realization of the resource utilization of high phosphorus steel slag.

In article http://doi.wiley.com/10.1002/srin.202200342, Bao, Wang, and Gu use the convolutional neural network model to predict the oxygen content at the end point of the converter. Experiments determine the number of convolutional layers, kernels, and convolutional kernel size. Its root mean square error, mean absolute error, and mean absolute percentage error are 35.29, 25.59, and 7.30%, respectively, which is superior to the back propagation neural network.

Given the accelerating depletion of iron ore resources, there is growing concern within the steel industry regarding the availability of high-phosphorus iron ore. However, it is important to note that the utilization of high-phosphorus iron ore may result in elevated phosphorus content and notable fluctuations in molten iron, thereby imposing additional challenges on the dephosphorization process in steelmaking. The most urgent issue in the process of converter steelmaking is how to achieve efficient dephosphorization. In this study, the influence of various factors on the logarithm of the phosphorus balance distribution ratio (lgLp), the logarithm of the P2O5 activity coefficient (lgγP2O5), and the logarithm of the phosphorus capacity (lgCp) were examined through thermodynamic calculations. The impact of each factor on dephosphorization was analyzed, and the optimal conditions for the dephosphorization stage of the converter were determined. Furthermore, the influence of basicity and FetO content on the form of phosphorus in the slag was analyzed using FactSage 7.2 software, and the precipitation rules of the slag phases were explored. The thermodynamic calculation results indicated that increasing the basicity of the dephosphorization slag was beneficial for dephosphorization, but it should be maintained below 3. The best dephosphorization effect was achieved when the FetO content was around 20%. The reaction temperature during the dephosphorization stage should be kept low, as the dephosphorization efficiency decreased sharply with the increasing temperature. In dephosphorization slag, Ca3(PO4)2 usually formed a solid solution with Ca2SiO4, so the form of phosphorus in the slag was mainly determined by the precipitation form and content of Ca2SiO4. The phases in the dephosphorization slag mainly consisted of a phosphorus-rich phase, an iron-rich phase, and a matrix phase. The results of scanning electron microscopy and X-ray diffraction analyses were consistent with the thermodynamic calculation results.

In the LF refining process, argon blowing at the bottom of ladle can play an important role in unifying the composition and temperature of molten steel and removing inclusions. However, unreasonable bottom argon blowing process can also cause many problems. Slag entrapment and slag surface exposure may occur, affecting the steel quality. Since the working conditions of different enterprises are very different, corresponding optimization is required for specific parameters. There were some problems in 70t ladle of a steel plant, such as unclear control of bottom argon blowing system in different refining periods, unobvious floating removal effect of inclusions in ladle, high total oxygen content and large fluctuation, etc. In this study, a 1:3 physical model was established according to the similarity principle. Then, on this basis, the experimental schemes with different blowing hole positions and argon flow rates were designed for simulation experiments. By means of mixing time measurement, flow field display and oil film measurement, the optimal argon blowing position was double holes 6, 12 (2/3R), and the included angle between them was 135°. The optimal argon flow rate for wire feeding and soft blowing should be 7.6 L/min (corresponding to the actual production of 180 L/min) and 0.6 L/min (corresponding to the actual production of 15 L/min), respectively. According to this scheme, the industrial experiments showed that the contents of total oxygen and nitrogen in the whole process were reduced, the surface density of inclusions in billet was reduced by 11.81% on average, and calcium sulfide and various inclusions containing aluminum were reduced to varying degrees.