Regular Article
Effect of Exit Wear Length on the Behavior of Coherent Jet
2024 Volume 64 Issue 12 Pages 1775-1782
Details
2024 Volume 64 Issue 12 Pages 1775-1782
The copper was the main manufacturing material to produce the coherent lance for enhancing the cooling effect. Due to the low hardness of copper and high-temperature environment, the exit of Laval nozzle would be worn off, resulting in suppressing the impaction ability of supersonic oxygen jet. In order to investigate the effect of wear length on the behavior of coherent jet, both high-temperature experiment and numerical simulation have been carried out, and the axial velocity, total temperature and oxygen fraction were measured in the experimental test to verify the accuracy of simulation model. Based on the result, the overexpand phenomenon was generated due to the Laval nozzle exit wear off, which improved the shock wave intensity at the tip of Laval nozzle, resulting in a lower axial velocity at the velocity potential core. With a longer wear length, the vorticity of the coherent jet periphery is increased, which causes more thermal energy of combustion flame being released prematurely near the coherent lance tip, leading to a shorter velocity potential core.
In order to reach the chemical contents standard of the end-point molten steel, the oxygen gas has been adopted to remove the excess elements in the liquid iron, such as the [C], [Si], and [P]. The oxygen gas is injected into the molten bath by the Laval nozzle, which increases its impaction ability, resulting in a greater dynamic condition for the oxidizing reactions.1,2,3) Additional, the coherent jet technology is further prolonged the velocity potential core length of the oxygen jet by using the shrouding flame. The combustion shrouding flame reduces the entrainment phenomenon between the ambient gas and the main oxygen jet. Additionally, the expansion effectiveness of the combustion shrouding flame increases the concentration of the main oxygen jet, thereby improving the impaction ability of the supersonic oxygen jet. Furthermore, the coherent lance can also function as a burner by controlling the flow rates of oxygen gas and shrouding fuel, which enhances the melting rate of steel scrap. Consequently, the coherent lance has been widely used in the Electrical Arc Furnace (EAF).4,5,6)
This means that the shrouding combustion flame has a great impaction on the technical index of main oxygen jet. Consequently, the primary focus of the research is on the effect of structure of coherent lance, shrouding fuel type, arrangements of shrouding and main nozzles on the main oxygen jet behavior.7,8,9,10,11) Odenthal et al.12) reported that the coherent lance designed by the SIS (Siemag Injection System) method could obviously prolong the combustion flame length, which increased the impaction ability of oxygen jet and reduced refractory erosion rate. Feng et al.13) indicated that the use of shrouding H2 increased the axial expansion rate of the combustion flame by the same heat value of shrouding gas principle, which improved the impaction ability of the main oxygen jet compared with the shrouding CH4. Wei et al.14) introduced the advantage of the coherent lime powder injection (C-LPI) technology in the EAF steelmaking process and proposed that the high velocity of the lime powder can delay the velocity attenuation of the main oxygen jet. Zhao et al.15) presented the supersonic shrouding gas injection method could effectively prevent the kinetic energy exchange between the ambient gas and main oxygen jet, which prolonged the velocity potential core of main oxygen jet to 33 times exit diameter of the Laval nozzle at the ambient temperature of 1800 K.
At present, the CH4 has been selected as the shrouding fuel gas in the coherent jet technology, due to the appropriate inflammability and heat value. When the CH4 is ignited by the pure oxygen gas, the theoretical temperature of the combustion flame is about 2871°C, which also releases the massive thermal radiation.16,17) Thus, the external surface of Laval nozzle exit is kept heating by the thermal radiation released by the shrouding flame. Although the heat conduction of the copper is quite great, its hardness is much lower than that of steel. Therefore, when the cooling capacity of the coherent lance is insufficient, the temperature of the external surface of Laval nozzle exit is quickly improved, which obviously reduces its hardness.18,19) Then, the external surface of Laval nozzle exit would be worn off by the high-pressure oxygen jet.
However, there is limited research on the behavior of a coherent jet formed by various exit wear lengths, and few studies have reported on the relationship between the impaction ability of an oxygen jet and wear length for traditional supersonic oxygen lances.20,21) In order to analyze the effect of the Laval nozzle exit wear length on the behavior of coherent jet, three wear lengths at the Laval nozzle exit are selected for testing through numerical simulation and combustion experiment at the ambient temperature of 1700 K.
In this paper, Section 2 presents detailed information on the coherent lance structure, experimental system, and simulation model. Subsequently, the axial velocity, total temperature, and impaction cavity distributions achieved by the simulation model are depicted and comprehensively discussed in Section 3. Finally, Section 4 presents the main findings of this research.
The one-dimensional isoentropic flow theory was adopted to design the structure of the Laval nozzle, and the design Mach number and O2 flow rate were 2.0 and 2800 Nm3/h, respectively. As a result, the throat and exit diameters of the main Laval nozzle were 27.0 and 35.0 mm, respectively. Besides, three concentric rings have been selected to arrange the shrouding nozzles, and their diameters, with diameters from the inner to the outer concentric ring being 61.2, 80.6, and 161.8 mm, respectively, as shown in Fig. 1. The inner, intermediate and outer hydraulic diameters of shrouding nozzles were 13.3, 14.3 and 13.3 mm, respectively. The shrouding O2 was injected into molten bath by the inner and outer shrouding nozzles, and the shrouding CH4 was injected by the intermediate shrouding nozzle.
Figure 2 represents the wear condition of the Laval nozzle. In order to investigate the effect of wear length on the behavior of coherent jet, a wear angle (αw) was selected as 45°, and the wear lengths (Lw) were 0, 1 and 3 mm, respectively. A wear length of 0 mm for the coherent lance indicated that no wear has occurred. The same parameter of coherent lance was applied in numerical simulation and combustion experiment.
For verifying the accuracy of the numerical result, a series of combustion experiments have been carried out at the high ambient temperature. In the preparation stage, the burner was ignited to improve the ambient temperature of experimental furnace, and 12 thermocouples were fixed at the different locations of the furnace side-wall. When the ambient temperature reached at 1700 K, the burner was snuffed out, and the Pitot tube would be first positioned at specific locations in an axial direction to achieve the dynamic and static pressures of the coherent jet at centerline of the Laval nozzle. And then, the thermocouple was positioned at the same position to achieve the total temperature of the coherent jet. Moreover, for improving the accuracy of measurement data, both pressure and temperature of the coherent jet have been measured by 30 s to calculate the average values. The measurement method and experimental system details are given elsewhere,22,23,24) and only the core experimental apparatus is briefly shown here.
A quarter 3D model of coherent lance has been built to research the behavior of coherent jet generated by various wear length. The computational domain included the main Laval nozzle, shrouding nozzles and coherent jet flowing zone. This domain ranged from the tip of the main Laval nozzle to 2100 mm and 200 mm downstream in the positive and negative axial directions, respectively, and 400 mm in the radial direction. Figure 3 depicts the mesh profile of the numerical model, and the blue, red and gray plane represented the mass inlet, pressure outlet and wall boundary conditions, respectively. Moreover, the mesh density has been improved near the coherent lance exit, due to the greater velocity and pressure gradient. The mesh number of this model was 6574 056, and 3711 768 mesh nodes were deployed in the axial direction from −0.20 m to 0.75 m, and in the radial direction from 0.00 m to 0.25 m.
The SST k-ω turbulence model was adopted to effect of the viscous and vortex structure on the coherent jet. The species model and GRI-3.0 were used to simulate the combustion process between CH4 and O2. The steady solution was selected to achieve the flow field of coherent jet. Initially, the computational domain started with no CH4 or O2 flow passing through the Laval or shrouding nozzle, and the computational domain was filled with static air. Table 1 shows the detail information about the boundary conditions of the simulation model.
| Name of boundary | Type of boundary conditions | Values |
|---|---|---|
| Main Laval nozzle | O2 mass flow rate (kg/s) | 0.27778 |
| Mach number | 2.00 | |
| Inlet temperature (K) | 300 | |
| Shrouding nozzle | CH4 mass flow rate (kg/s) | 0.00744 |
| O2 mass flow rate (kg/s) | 0.01488 | |
| Inlet temperature (K) | 300 | |
| Outlet | Static pressure (Pa) | 101325 |
| Mass fractions (%) | N2=79/O2=21 | |
| Ambient temperature (K) | 1700 |
For the solution method of the simulation model, the coupled and least squares cell based schemes were selected to solve the pressure-velocity coupling process and improve the computational efficiency, respectively. Both pressure and gas species discretization rate were calculated by the second order upwind method scheme. Meanwhile, the other variables (density, momentum and turbulence kinetic energy, etc.) were solved by the QUICK (Quadratic Upstream Interpolation for Convective Kinematics) scheme. Two criterions were used to determine the solution convergence of simulation model, as follows: The energy residual was <10−7 and the other variables residuals were <10−4; The variations of average total temperature and velocity were kept within 1.0 K and 1.0 m/s, respectively, at the outlet boundary condition.
Figure 4 presents the axial velocity profiles of main oxygen jet formed under various operational conditions at the centerline of the Laval nozzle. The coherent lance with wear length of 1 and 3 mm are addressed as the C-1 and C-3, respectively. The coherent lance with wear length of 0 mm and with none shrouding flow rate are represented as the original coherent lance and traditional oxygen lance, respectively. Meanwhile, the measurement data of the main oxygen jet are depicted by various symbols (□, ○, △ and ▽), and solid and dotted lines are selected to show the simulation result. The result shows that the average axial velocity variation of the main oxygen jet between the experimental test and numerical simulation is 5.1%, which proves the experimental data is in good agreement with the simulation result.
The results show that, when the main oxygen jet passes through the shock wave near the Laval nozzle exit, the velocity potential core is formed, and the axial velocity of the main oxygen jet keeps unchanged. Then, the axial velocity of the main oxygen jet is reduced, after reaching the end of the velocity potential core. It indicates that the axial velocities of the main oxygen jets at the Laval nozzle exit are same of 491 m/s at the centerline of the Laval nozzle, for all the operational conditions. Thus, both the shrouding flow rate and the wear length have limited impaction on the axial velocity of the main oxygen jet at the center of the Laval nozzle exit.
During the oxygen jet flowing through the Laval nozzle, its high pressure potential is transformed into its kinetic energy by adiabatic expansion phenomenon. However, the wear length of Laval nozzle alters the normal expansion process of the oxygen jet, resulting in an overexpansion of the oxygen jet at the Laval nozzle exit. In consequence, the shock wave intensity of the oxygen jet is increased.
In this research, the axial velocity variation is used to represent the shock wave intensity of the main oxygen jet. Near the Laval nozzle exit, the simulation model achieves both the maximum and minimum axial velocities of the main oxygen jet. And the axial velocity variation of main oxygen jet is the difference between its maximum and minimum values. Figure 4(b) shows that the maximum axial velocity of the main oxygen jet generated by the traditional oxygen lance, original coherent lance and CL-1 and CL-3 is 501, 501, 511 and 527 m/s, respectively. Meanwhile, the minimum axial velocity of the main oxygen jet formed by the traditional oxygen lance, original coherent lance and CL-1 and CL-3 is 468, 468, 463 and 448 m/s, respectively. As a result, the axial velocity variation of main oxygen jet using the traditional oxygen lance, original coherent lance and CL-1 and CL-3 is 33, 33, 48 and 79 m/s, respectively.
Thus, with increasing the wear length of the Laval nozzle exit, the axial velocity variation of main oxygen jet is obviously improved, resulting in a greater shock wave intensity. In addition, both maximum and minimum axial velocities of the main oxygen jet generated by the traditional oxygen lance and the original coherent lance are same. This presents that the shrouding gas flow has no effect on the adiabatic expansion phenomenon of the main oxygen jet near the Laval nozzle exit.
Meanwhile, due to a greater shock wave intensity, the kinetic energy of main oxygen jet is suppressed, which decreases the axial velocity value of main oxygen jet. Hence, at the velocity potential core, the axial velocity value of main oxygen jet formed by the traditional oxygen lance, original coherent lance and CL-1 and CL-3 is 480, 480, 478 and 473 m/s, respectively.
Based on the result, the length of the velocity potential core generated by the traditional oxygen lance, original coherent lance and CL-1 and CL-3 is 0.587, 0.936, 0.929 and 0.919 m, respectively. Hence, the length of the velocity potential core is also decreased by the wear length. As mentioned, the shrouding combustion flame is the main factor on the length of the velocity potential core. It seems that the wear length has an influence on the shrouding combustion process, which would be discussed in the 4.2 Section.
Figure 5 shows the axial velocity profiles of the main oxygen under various operational conditions at the Laval nozzle exit. For the traditional oxygen lance and original coherent lance, the axial velocity ranging from 0.000 m to 0.016 m are constants of 491 m/s at the Laval nozzle exit. Then, with increasing the Y coordinate, the axial velocity gradually reduces to the 0 m/s, due to the velocity boundary layer at the Laval nozzle surface.
The diameter of Laval nozzle diameter increases with a bigger wear length. Subsequently, overexpansion phenomenon of high pressure oxygen jet is generated, when the oxygen jet reaches the worn face of the Laval nozzle. As a result, for the CL-1 and CL-3, there is a phenomenon of increasing axial velocity near the edge of the Laval nozzle exit. The Y coordinate value at the Laval nozzle exit, where the axial velocity of the oxygen jet begins to increase, is defined as the Pin. The maximum axial velocity of the main oxygen jet at the Laval nozzle exit generated by the CL-1 and CL-3 are 494 and 505 m/s, respectively. Moreover, the Pin formed by the CL-1 and CL-3 are 0.0158 and 0.0147 m, respectively. Hence, at the Laval nozzle exit, the maximum axial velocity of the main oxygen jet is increased, and the Pin is reduced, when the wear length is increased.
Due to the overexpanded phenomenon, the mixing rate of the main oxygen jet with the shrouding gas flow may be improved. Hence, the vorticity profile of coherent jet is proposed to analyze this mixing rate. The vorticity is used to investigate the rotation of a gas phase, which is also a measure of mixing rate among the fluids. A higher vorticity magnitude leads to a greater mixing rate among the gas phases. In the Cartesian coordinates, the vorticity (ζv) can be calculated as follows:
| (1) |
where,
The maximum vorticity of the coherent jet using the traditional oxygen lance, original coherent lance, CL-1 and CL-3 is 2.42×105, 2.01×105, 2.13×105 and 2.34×105 1/s, respectively, at the Z coordinate of 0.15 m. Without the shrouding flow gas, the main oxygen jet using the traditional oxygen lance is mixed with ambient gas in advance, resulting in a biggest vorticity at the Z coordinate of 0.15 m. Besides, the average of the maximum vorticity for the coherent jet generated the traditional oxygen lance, original coherent lance, CL-1 and CL-3 is 0.96×105, 0.99×105, 1.05×105 and 1.08×105 1/s, respectively, when the Z coordinates are 0.28 and 0.42 m. Although the shrouding flame flow can suppresses the mixing rate of main oxygen jet with the ambient gas, the main oxygen jet still will be entrained with the shrouding flame flow. Whereas, the density of the shrouding flame flow is lower than that of the ambient gas, due to a higher temperature. Thus, the velocity of the main oxygen jet formed by the coherent lance is still higher than that by the traditional oxygen lance, with a bigger vorticity.
Based on the result, the deascend order of the vorticity for the coherent jet is: CL-3, CL-1 and original coherent lance. Hence, with a longer wear length, both the vorticity maximum of the coherent jet and the main oxygen jet with the shrouding gas is obviously increased. Thus, the wear length of the Laval nozzle exit influences the behavior of the shrouding gas flow, which may alter the shrouding combustion process.
4.2. Total Temperature DistributionFigure 7 depicts the total temperature distribution of the main oxygen jet at the centerline at room and high temperatures with various shrouding fuel shrouding supplement methods. Meanwhile, the average total temperature variation of the main oxygen jet between the experimental test (□, ○, △ and ▽) and numerical simulation (solid and dotted lines) is 7.1%.
Because of none shrouding combustion flame, the main oxygen jet formed by traditional oxygen lance is rapidly entrained with the ambient gas, which leads to its temperature starting to rise first, as presented in Fig. 7. The result shows that the ascend order of the total temperature for the main oxygen jet is: CL-3, CL-1 and original coherent lance. This further demonstrates that a longer wear length suppresses the protection effectiveness of the shrouding combustion flame, which causes the premature mixing of the main oxygen jet with the high temperature ambient temperature.
For the Z coordinate, in the range from 1.00 m to 1.25 m, from 1.25 m to 1.50 m, from 1.50 m to 1.75 m, from 1.75 m to 2.00 m, the average total temperatures of the main oxygen jet formed by original coherent jet are 15.3, 5.9, 2.6 and 2.2°C lower than that by CL-1, respectively. Additional, the average total temperatures of main oxygen jet, along the Z coordinate, formed by original coherent jet are 16.5, 11.1, 9.9 and 9.4°C lower than that by CL-3 within the range of 1.00 m to 1.25 m, 1.25 m to 1.50 m, 1.50 m to 1.75 m, 1.75 m to 2.00 m, respectively. Hence, the average total temperature variation of the main oxygen lance is increased with a longer wear length, and is reduced with a larger Z coordinate.
Figure 8 presents the total temperature profiles under various operational conditions. The total temperature isolines of 2000 K and 2800 K generated by three types of wear lengths are measured in the simulation model, and are referred to as the 2000 K isoline and the 2800 K isoline, respectively. A higher Z coordinate of the 2800 K total temperature isoline indicates a more complete burning of CH4 closer to the coherent lance tip. Meanwhile, a higher Z coordinate of the 2000 K total temperature isoline represents the shrouding combustion flame achieving better protection effective for the main oxygen jet. Additionally, the grey and purple lines represent the positions, at the Z coordinate, of the 2800 K and 2000 K isolines, respectively, as shown in Fig. 8(b).
Based on the result, the Z coordinates of 2000 K isoline formed by the original coherent lance, CL-1 and CL-3 are 0.890, 0.859 and 0.842 m, respectively. The Z coordinates of 2800 K isoline formed by the original coherent lance, CL-1 and CL-3 are 0.268, 0.293 and 0.302 m, respectively.
Therefore, more shrouding CH4 is ignited by the O2 gas with a longer wear length near the coherent lance tip, which causes more thermal energy of combustion flame being released prematurely. In consequence, both the length of shrouding combustion flame and the Z coordinates of 2000 K isoline are reduced, which suppresses the protection effective of the hrouding combustion flame and decreases the velocity potential core of the main oxygen jet.
In this research, three cross-sections with radius of 0.12 m are selected at the Z coordinate of 0.105, 0.280 and 0.420 m, respectively. Then, the CO2 mass fraction is measured in those cross-sections, to represent the reaction rate of the shrouding CH4, as depicted in Fig. 9.
At the Z coordinate of 0.105 m, the CO2 mass fraction is enhanced with a longer wear length. This further proves that more shrouding CH4 is ignited by the O2 gas with a longer wear length near the coherent lance tip.
As the combustion reaction proceeds gradually, more CO2 is produced, resulting an increase in CO2 mass fraction at the Z coordinate of 0.280 m. The decrease rates of CO2 mass fraction at the Z coordinate of 0.420 m, generated by the original coherent lance, CL-1 and CL-3 are 0.2, 3.2 and 7.6%, respectively, comparing with the Z coordinate of 0.420 m. Since less shrouding CH4 consumed by O2 gas, the CO2 is still produced by the residual shrouding CH4, for the original coherent lance. As a result, the decrease rates of CO2 mass fraction generated by the original coherent lance is the minimum, compared to the CL-1 and CL-3.
Consequently, when the Laval nozzle exit is worn, the combustion rate of the shrouding CH4 near the coherent lance tip is improved, leading to reduced protection effectiveness of the shrouding combustion flame and a shorter velocity potential core.
4.3. Impaction Depth and Area DistributionsWhen the main oxygen jet reaches the surface of the molten bath, the liquid slag would be penetrated due to the total pressure variation, resulting in an impaction cavity. This means that the dynamic pressure distribution of the main oxygen jet is the key factor in the shape of the impaction cavity. In this paper, the two-layer liquid phases (liquid slag and molten steel) have been taken into consideration, and the impaction depth calculated by the dynamic pressure is as follows:25,26)
| (2) |
| (3) |
where, Hp, Hslag and Hsteel are the theoretical impaction depth, liquid slag thickness and steel molten penetrate depth (m), respectively. Pd is the dynamic pressure of main oxygen jet (Pa). g, ρslag and ρsteel are the gravitational acceleration of 9.81 (m2/s), liquid slag and molten steel density (kg/m3), respectively.
The impaction area is defined as the region, where main oxygen jet just penetrates the slag layer and directly contacts the surface of molten steel. Additional, Eq. (3) is selected to calculate the dynamic of main oxygen jet, when main oxygen jet can just penetrate the slag layer.
| (4) |
In this paper, the thickness of liquid slag, the densities of liquid slag, and molten steel are assumed as 0.15 m, 3000 kg/m3 and 7200 kg/m3, respectively. As a result, the dynamic pressure of the main oxygen jet is 4415 Pa, when the main oxygen jet can just penetrate the liquid slag.
Besides, there is a tilt angle for the application of the coherent lance in the EAF steelmaking process, and this tilt angle is selected as 45°, as depicted in Fig. 10. This implies that the shape of the theoretical impaction area is an ellipse, and the impaction diameter, discussed in Fig. 11, is calculated based on the hydraulic diameter of this area. Besides, the vertical distance from the coherent lance tip to the surface of molten bath is assumed to be 1.0 m.
Figure 11 shows that the average impaction depth and diameter formed by CL-1 and CL-3 are increased by 32.4 and 17.3%, respectively, compared to the traditional oxygen lance. This means that although the Laval nozzle exit is worn, the impaction ability of the coherent jet is still better than the traditional oxygen jet. Compared to the original coherent lance, the average impaction depth and diameter formed by CL-1 and CL-3 are decreased by 2.6 and 2.7%, respectively. Hence, both impaction depth and diameter reduces with a longer wear length.
At the impaction cavity, the oxygen gas is injected into the molten bath. With a higher oxygen flow rate, the oxidation reaction in the molten bath is improved, which reduces the melting time and increases the oxygen utilization rate. The oxygen flow rate on the impaction area is calculated by the simulation model.
Based on the result, the oxygen flow rate on the impaction area formed by the traditional oxygen lance, original coherent lance, CL-1 and CL-3 are 0.038, 0.050, 0.0492 and 0.0480 kg/s, respectively. As mentioned, the flow rate of the main oxygen jet is 0.2778 kg/s in each simulation model. Therefore, the oxygen flow rate on the impaction area formed by the traditional oxygen lance, original coherent lance, CL-1 and CL-3 are 13.8, 17.9, 17.7 and 17.3% of the main oxygen jet.
Hence, only a small fraction of oxygen gas is directly injected into the molten steel through the impaction area. It seems that a big fraction of oxygen gas would be supplied to the liquid slag, due to the lower impaction ability at the periphery of the jet. Subsequently, the (FeO) is formed in the liquid slag and plays a crucial role in transferring the oxygen element to the molten steel, facilitating the removal of excess elements such as the [C], [Si], and [P].
The results indicate that a longer wear length of the Laval nozzle exit would increase the oxygen flow rate allocated to the liquid slag. Meanwhile, the transmit rate of the oxygen gas to the liquid slag is further improved by a traditional oxygen lance, as there is no protective effect from the shrouding combustion flame.
In this paper, both numerical simulation and experimental test have been carried out to investigate the behavior of coherent jet formed by various exit wear length by discussing the axial velocity, total temperature and impaction cavity parameters. The main conclusions of this research have been summarized as following:
(1) The axial velocity value of main oxygen jet, at the velocity potential core, formed by the original coherent lance and CL-1 and CL-3 is 480, 478 and 473 m/s, respectively. This indicates that, with a longer wear length, the overexpansion phenomenon of high pressure oxygen will become more evident, resulting in bigger velocity variation and shock wave intensity. Subsequently, at the velocity potential core, the axial velocity of the main oxygen jet is gradually reduced.
(2) A longer wear length at the Laval nozzle exit enhances the vorticity of the coherent jet periphery, consequently altering the flow field of the shrouding gas. As a result, the Z coordinates of 2800 K isoline formed by the original coherent lance, CL-1 and CL-3 are 0.268, 0.293 and 0.302 m, respectively. Hence, more shrouding CH4 is ignited by the O2 gas with a longer wear length near the coherent lance tip, which causes more thermal energy of combustion flame being released prematurely, leading to a lower protection effectiveness of shrouding flame and a shorter velocity potential core.
(3) Because of the lower impaction ability generated by a longer wear length, both impaction depth and area are reduced. Meanwhile, the results show that the oxygen flow rate on the impaction area formed by the original coherent lance, CL-1 and CL-3 are 17.9, 17.7 and 17.3% of the main oxygen jet. Therefore, a longer wear length of the Laval nozzle exit would decrease the fraction of oxygen gas directly injected into the molten steel through the impaction area.
On behalf of all authors, the corresponding author states that there is no conflict of interest.
The authors would like to express their thanks for the support by National Natural Science Foundation of China (NSFC 52322407, NSFC 52074024 and NSFC 52274313).
