Influences of the Transverse Static Magnetic Field on the Droplet Evolution Behaviors during the Low Frequency Electroslag Remelting Process

Abstract

To visualize the electroslag remelting (ESR) process, a transparent experimental model was adopted. The droplet evolution process at the consumable electrode tip was recorded by a high-speed camera. Different intensities of the transverse static magnetic field (TSMF) were imposed during the ESR process with a low frequency current of 5 Hz. The representative processes of formation and detachment of the droplets under different conditions were given. The results showed that when the intensities of the TSMF were equal or greater than 0.3 T, the droplet evolution processes would be influenced remarkably. When the TSMF reached 0.5 T, the liquid neck would be broken up into two arrays of smaller droplets. The mechanism of the breakup phenomenon appearing on the droplet neck was discussed. Statistical analysis of the videos captured by the high-speed camera under different conditions had been done. The results showed that the separation degree of the droplet necks and the remelting rate could be increased to a certain extent as the increase of the TSMF. The features of the collected droplets with the TSMF of 0.7 T showed a smaller average size and more tiny droplets comparing with what were obtained without the external magnetic field. The decrease of the droplet size and the generation of numerous tiny droplets enlarged the interfacial area between metal and slag tremendously.

1. Introduction

To settle the problem of nonmetallic inclusions damaging the mechanical properties of steels, electroslag remelting (ESR) technology^1,2,3,4,5,6) was developed for its excellent purification efficiency. Nowadays, the ESR process has become more and more important for the increasingly high requirements of the steels in areas such as electricity, aviation, metallurgy, machinery and so on. From a dynamic view, a larger interfacial area between metal and slag leads to a better removal efficiency of the inclusions. Li et al.⁷⁾ conducted the ESR experiments and took the samples from the consumable electrode, the molten film on the electrode tip, metal droplets in the slag and the final ingot. The quantitative determination of inclusions in the samples were made by metallographic, chemical and radioisotopic methods. The results showed that the cleaning happened the most at the consumable electrode tip. The droplet evolution behavior at the consumable electrode tip is one of the key points in influencing the interfacial area between metal and slag. Thus, it is necessary to study the evolution process of the droplets during the ESR process. To improve the interfacial area, the technique named Magnetically Controlled ESR (MC-ESR)^8,9,10) process was considered as shown in Fig. 1. In this work, a transverse static magnetic field (TSMF) was adopted during the MC-ESR process.

Fig. 1.

Schematic of Magnetically Controlled ESR process.

Up to now, it is still difficult to find out a perfect way to observe the droplet evolution behaviors during the actual ESR process due to the dangerous operating conditions with the strong electric current and the high temperature. Therefore, a lot of reports were focused on physical and numerical simulations. J. Campbell¹¹⁾ developed a transparent model to observe the phenomena occurring in the slag during the ESR process. Pb, Zn, Al and Cu bars were chosen as the consumable electrodes and molten LiCl-KCl was selected as the slag. The droplet evolution pictures were given but very ambiguous, and the external magnetic field was not considered in his studies. Wood alloy (the electrode) and NaCl solution (the slag) were used by Dong et al.^12,13) to study the mechanisms of the droplet formation, the effects of filling ratio and the current distribution ratio of passing electrode and conductive part of the mold. Wang et al.^14,15) developed a three-dimensional numerical model coupling electromagnetic, two-phase flow and temperature field. Their results showed that the momentum and heat could be carried out to the metal pool by the falling metal droplets. A. Kharicha et al.¹⁶⁾ established a 3D magneto-hydrodynamic model to simulate the dripping process of droplets. They found that superimposing a TSMF could make the droplets become smaller.

When a TSMF is imposed during the ESR process, a periodic reverse Lorentz force will be generated by the interaction of the alternating current and the external magnetic field, and this phenomenon is called as the electromagnetic vibration (EMV).^17,18,19) The effect of the EMV has the same frequency as the alternating current when the external magnetic field is static. The alternating current will shuttle through the metal melt preferentially to minimize the electrical resistance during the ESR process. Obviously, the EMV will influence the droplet evolution behaviors. Meanwhile, a low frequency remelting current can increase the power factor, meaning that a large electric energy can be saved, especially for the large scale ESR process.^20,21) Some researches^22,23,24) had been done about the influences of the low frequency current on the ESR process, but few reports were released in regard to the droplet evolution behaviors under the MC-ERS process.

In this study, a visualized physical simulation equipment was adopted. The influences of different intensities of the TSMF were studied during the ESR process with a low frequency current of 5 Hz. The representative processes of the droplet evolution behaviors under different conditions were given. The statistical analysis of the videos captured by the high-speed camera had been done to provide the quantitative results. The mechanism of the special phenomenon found in this study was discussed.

2. Experimental Procedure

The visualized physical simulation equipment was used to conduct the experiments, as shown in Fig. 2. Quartz glass tube with the internal diameter of 31 mm was used as the mold. Molten ZnCl₂ with the low melting point (548 K) was chosen as the molten slag for its fine transparency and low electrical conductivity. Zinc bar (melting point: 692 K) with the diameter of 15 mm was selected as the consumable electrode. The melting point of zinc bar is 144 K higher than the melting point of ZnCl₂, which is a suitable temperature difference. Meanwhile, the chemical reaction will not happen between the zinc bar and the molten ZnCl₂. The purity of the zinc and ZnCl₂ used in the experiments were all in AR grade. All the experiments were conducted with the argon atmosphere, because the ZnCl₂ was easy to deliquesce. To observe the droplet evolution process elaborately, a high-speed camera was utilized to record the whole process happening at the consumable electrode tip at the record frequency of 200 frames per second. The depth of the molten slag was about 70 mm. The consumable electrode was immersed into the molten slag for about 10 mm. The feed rate of the consumable electrode was manually controlled to maintain the intensity of the remelting current constant. The remelting current was 8 amperes with the low frequency of 5 Hz. The experiments were done under different intensities of the TSMF generated by an electromagnet. Water cooling system of the bottom electrode was kept constant in each experiment. The detailed physical properties, geometry and operating conditions of the experiment are listed in Table 1.

Fig. 2.

Diagram of the visualized physical simulation equipment.

Table 1. Physical properties, geometry and operating conditions of the experiment.

Parameters	Value
Physical properties of electrode (Zn)
Density (kg/m3)	7140
Melting point (K)	692
Electric conductivity (S/m)	2.7×106
Physical properties of slag (ZnCl2)
Density (kg/m3)	2370
Melting point (K)	548
Electric conductivity (S/m)	30
Geometry
Electrode diameter (m)	0.015
Slag diameter (m)	0.031
Slag height (m)	0.07
Immersion depth of electrode (m)	0.01
Operating conditions
Current (A)	8
Frequency (Hz)	5
External static magnetic field (T)	0/0.1/0.3/0.5/0.7
Atmosphere	Argon

A circular resistance furnace surrounding the quartz glass tube was used to melt the ZnCl₂ powder. The zinc bar was heated to 473 K, and then polished before it was fixed on the lifting device. After the zinc bar was inserted in the molten ZnCl₂, the resistance furnace was turned off and lowered down. Then the AC power supply was turned on so that the remelting current could shuttle through the zinc bar, molten ZnCl₂ and the bottom electrode. Large Joule heat would be generated by the remelting current due to the low electrical conductivity of the molten ZnCl₂. The zinc bar would be gradually remelted from its tip when the temperature of the molten ZnCl₂ reached above 692 K. The physical theory of the experiment was almost the same as the industrial ESR process. Each experiment was conducted for five minutes and was repeated three times. On account of the high cooling strength of the bottom electrode, the droplets could be collected after each experiment. The collected droplets were cleaned, dried and weighed after each experiment.

3. Results and Discussion

3.1. Visualization of the Droplet Evolution under Different Intensities of the TSMF

The different representative processes of formation and detachment of the droplets recorded by the high-speed camera under different conditions are shown from Figs. 3, 4, 5, 6, 7. To demonstrate the droplet evolution process better, the moment of the droplet separating from the electrode tip is set as T second and the time intervals between the continuous pictures are different. It can be discovered that the formation process of the droplet takes much more time comparing with the time spent by the detachment process through the time intervals.

Fig. 3.

Droplet evolution process without the external magnetic field.

Fig. 4.

Droplet evolution process with the TSMF of 0.1 T.

Fig. 5.

Droplet evolution process with the TSMF of 0.3 T.

Fig. 6.

Droplet evolution process with the TSMF of 0.5 T.

Fig. 7.

Droplet evolution process with the TSMF of 0.7 T.

Figure 3 shows the representative process of the droplet evolution without the external magnetic field. It can be found that the liquid film gathers at the consumable electrode tip gradually, and soon a liquid bulge appears. After that, the bulge connects with the electrode tip by a slender liquid neck, and then the neck becomes more and more slender until it is snapped under the main action of gravity and the pinch force.²⁵⁾ When the big droplet separates from the liquid neck, the liquid neck which still connects with the electrode tip may rise upwards quickly. However, sometimes there is one or two small droplets may form from the liquid neck under the action of the interfacial tension and gravity. In the meantime, the big detached droplet sinks down in the shape of a saucer. Here the big droplet is called main droplet and the small droplets formed from the droplet neck are called satellite droplets for distinction.

As shown in Fig. 4, when the external TSMF is 0.1 T, the droplet evolution behavior is not changed a lot comparing with the situation without the external magnetic field. However the liquid neck becomes a little longer before it is snapped under the action of the EMV, as shown in (T-0.005) second of Fig. 4.

It can be discovered that the droplet evolution has been changed by the action of the EMV when the intensity of the TSMF reaches 0.3 T, as shown in Fig. 5. It is known that the intensity of the EMV is proportional to the intensity of the external magnetic field, so the effect of the EMV is strong enough to change the droplet evolution behavior observably with the TSMF of 0.3 T during the low frequency ERS process. The growing bulge at the consumable electrode tip can be compelled back and forth by the effect of the EMV, which means that the liquid bulge is not always located on the focusing plane of the high-speed camera, so sometimes the contour of the liquid bulge becomes blurred. When the liquid neck appears, the remelting current will mainly flow through the slender neck to minimize the electric resistance for the high electrical resistivity of the slag, and this causes the liquid neck to suffer an enormous Lorentz force. In the meantime, it can be noticed that the liquid neck has become more clearly from a blurry figure. This indicates that the Lorentz force changes its direction and the liquid neck is compelled from back to front or vice versa, but the main droplet does not keep up with the liquid neck for its large inertia. So the flimsy neck can be avulsed from connecting with the electrode tip and the main droplet under the effect of the EMV. After that, the slender columnar liquid neck is long enough to be broken up into an array of small satellite droplets for the well-known phenomena of Plateau-Rayleigh instability^26,27,28) as shown in (T-0.005) second of Fig. 5, which indicates that the interfacial area between metal and slag can be improved.

The droplet evolution can be influenced remarkably during the MC-ESR (I = 8 A, f = 50 Hz) when the TSMF reaches 0.5 T.²⁹⁾ It indicates that a lower intensity of the TSMF can improve the droplet evolution behavior during the MC-ESR with a low frequency current. As a rule, the frequency of the EMV is equal to that of the remelting current. When reducing the current frequency under the MC-ESR process, the decrease of the EMV frequency leads to a larger displacement of the droplet neck, because there is more time for the Lorentz force to transform the liquid neck before the direction of the Lorentz force has been reversed in a cycle. There seems to be a higher efficiency for the EMV with a low frequency to change the droplet evolution process.

Figure 6 shows the representative droplet evolution process with the TSMF of 0.5 T. The effect of the EMV is so fierce that the whole molten slag is stirred severely which makes the molten slag become milky comparing with that in the condition of no external magnetic field. From the continuous pictures, it can be observed that a stronger liquid neck appears with the main droplet falls behind for its blurry figure. Shortly afterwards, the liquid neck is separated from the position near the middle line of the liquid neck. Then the whole liquid neck is broken up into a lot of smaller satellite droplets sinking after the main droplet as shown in (T+0.1) second of Fig. 6. As shown in Fig. 7, the representative droplet evolution process with the TSMF of 0.7 T is almost the same as the process with the TSMF of 0.5 T. But the molten slag is stirred more violently under the action of a stronger EMV, which can be observed through the video captured by the high speed camera.

In the meantime, it can be observed that there are more inclusions in Fig. 7 than in Fig. 3. The inclusions appearing from Figs. 3, 4, 5, 6, 7, are the oxide coating of the zinc bar. Although the experiment was conducted with the argon atmosphere and the zinc bar was polished before it was inserted in the molten ZnCl₂, the absolutely oxygen-free condition could not be kept. Then, the surface of the zinc bar would be oxidized more or less and the inclusions (ZnO crust) would appear when the zinc bar was dipped into the molten ZnCl₂ with the high temperature of about 883 K during the experimental process. The density of the inclusions is 5606 kg/m³, which is about twice the mass of the molten ZnCl₂ (2370 kg/m³). So the inclusions would sink down to the bottom area during the experiment without the external magnetic field. When the TSMF was superimposed, the forceful EMV would be generated. With the increase of the TSMF, the stronger EMV would stir the molten slag fiercely. Hence, the inclusions could not stay still at the bottom area. And that’s why there are more inclusions in Fig. 7 than in Fig. 3. The main purpose of this study is to find out the influences of the TSMF on the droplet evolution behaviors during the ESR process with a low frequency current of 5 Hz. From the videos captured by the high-speed camera, it can be found that the movement of liquid metal is much quicker than the movement of inclusions. The inclusions have little influence on the droplet evolution processes. So the inclusions appearing in the molten slag can be ignored.

3.2. Mechanism of the Breakup Phenomenon on the Droplet Neck

To explain the phenomenon found under the condition when the TSMF is greater than or equal to 0.5 T, simple three-dimension diagrams have been drawn as shown in Fig. 8. The main droplet detaches from the electrode tip when the interfacial tension of the droplet neck cannot withstand the detaching force (gravity and pinch force) during the ESR process without the external magnetic field as shown in Fig. 8(a). While the TSMF reaches 0.5 T during the MC-ESR process with a low frequency current, the droplet evolution process becomes more complicated as shown in Fig. 8(b).

Fig. 8.

Simple three-dimension diagrams of detachment of the droplets. (a) Without external magnetic field and (b) with the TSMF greater than or equal to 0.5 T.

Here, it is supposed that the direction of the TSMF (B) is toward the positive direction of Y axis, and the direction of the alternating current (J) goes along the direction of Z axis, so the direction of the EMV (F=J×B) is in X direction as shown in bottom-left corner of Fig. 8(b). The EMV is the Lorentz force, whose amplitude and orientation is periodic changed corresponding with the time behavior of the alternating current (5 Hz). Meanwhile, the intensity of the EMV is proportional to that of the current density or the external TSMF. The mechanism of the breakup phenomenon can be explained as following: firstly, due to the liquid droplet neck suffering a tremendous effect of the EMV, the shape of the liquid droplet neck can be changed from a column into a flat strip. The cross-sections of the liquid necks without and with the TSMF are shown in top-left corner of Figs. 8(a) and 8(b). That’s why the droplet neck seems stronger as shown in (T-0.02) second of Figs. 6 and 7. It is easy to understand that the remelting current concentrates on flowing through the middle line part of the liquid flat strip to minimize the electrical resistance between the two electrodes. The middle line part of the strip becomes thinner and the border begins to expand outward due to the more forceful EMV acting on middle line part of the strip. The periodic reverse force makes the middle line area of the strip become so thin that a gap appears and quickly grows bigger. The flat strip can be separated into two slender cylinders as shown in Fig. 8(b), which corresponds to the phenomenon found in T second of Fig. 7 obviously. Then, the alternative current focuses on flowing through the two liquid cylinders, which means that the instability of the two cylinders will be increased a lot by the EMV. Eventually, the two liquid cylinders are broken up into two arrays of small satellite droplets via the Plateau-Rayleigh instability.

3.3. Influences of the Intensities of the TSMF on the Droplet Evolution

Statistical work has been done frame-by-frame of the videos recorded by the high-speed camera to evaluate the separation degree of the droplet necks under different intensities of the external TSMF. The number of main droplets (N_m) generated in each experiment and the number of satellite droplets (N_s) generated in each dripping process were recorded. So the average number of satellite droplets accompanying with each main droplet (A_s) can be simply calculated by the equation as following:   

$$A_s=\frac{\mathrm{\Sigma }{N}_s}{N_m}$$(1)

The quantitative data of the average number of satellite droplets sinking after each main droplet during every dripping process are shown in Fig. 9 and error bars donate S.D. The average number of satellite droplets accompanying with each main droplet can manifest the separation effect of the liquid neck caused by the EMV. It can be easily found that the average number increases rapidly when the TSMF is above 0.1 T. And when the TSMF reaches 0.5 T, the average number achieves its maximum value of 8.5 which means that the liquid neck can be broken up into about 8 satellite droplets during each detachment process. While the TSMF is increased to 0.7 T, the average number decreases a little but it is still approximately equal to the value with the TSMF of 0.5 T.

Fig. 9.

Average number of the satellite droplets with each main droplet under different intensities of the TSMF.

The average remelting rate (A_r) can be calculated by the equation as following:   

$$A_r=\frac{W_d}{t_{tot}}$$(2)

where W_d is the weight of all the droplets which have been collected after each experiment, and t_tot is the total time of each experiment. The changes of A_r with different intensities of the TSMF are shown in Fig. 10 and error bars donate S.D. The results demonstrate almost the same tendency comparing with the changes of the average number shown in Fig. 9. As the increase of the TSMF, the effect of the EMV becomes larger. So the liquid metal film and the liquid bulge attaching to the electrode tip are compelled to drift away earlier, and the main droplet may be avulsed from the electrode tip when the growing bulge is still small if the effect of the EMV is strong enough. The EMV also improves the heat transfer from the molten slag to the consumable electrode. However the improvement of the remelting rate is not unlimited. When the TSMF is equal or greater than 0.5 T, the restrictive condition of the remelting rate becomes the intensity of the remelting current, so the remelting rates with the TSMF of 0.5 and 0.7 T are almost the same.

Fig. 10.

Variation of remelting rate under different intensities of the TSMF.

3.4. Features of the Droplets

The features of the collected droplets under the conditions without the external magnetic field and with the TSMF of 0.7 T are shown in Fig. 11. The number and the equivalent diameter of metal droplets have been counted and measured. Then, a quantitative graph has been drawn, as shown in Fig. 12. There are 19 main droplets with the equivalent diameter of 9.2 mm under the condition without the external magnetic field. In the meantime, there are 48 satellite droplets with the equivalent diameter less than 2 mm. While the TSMF of 0.7 T is superimposed, there are 47 main droplets with the equivalent diameter of 6.1 mm, and there are more than 500 satellite droplets formed from the droplet necks. Among the hundreds of satellite droplets, there are at least 450 satellite droplets with the equivalent diameter less than 2 mm. The following three major differences can be concluded: the first one is that there are more droplets being generated with the TSMF of 0.7 T for a higher remelting rate. The second one is that the size of main droplets with the TSMF of 0.7 T is smaller than that with no external magnetic field obviously. The reason is that the main droplets can be avulsed from the consumable electrode tip when the growing liquid bulges are still small due to the forceful EMV. The last one is that there are hundreds of satellite droplets being generated from the liquid necks due to the breakup phenomenon discussed earlier, while there are only dozens of satellite droplets being generated with no external magnetic field.

Fig. 11.

Droplets collected after the experiments. (a) Without the external magnetic field and (b) with the TSMF of 0.7 T.

Fig. 12.

Equivalent diameter distributions of the droplets obtained without the TSMF and with the TSMF of 0.7 T.

It is known that the shape of metal pool is one of the key factors to produce high quality ingots by the ESR process. It is well to achieve a typical U-shaped metal pool profile. Wang et al.³⁰⁾ carried out the numerical simulations about the ESR process with the vibrating and traditional electrodes. Their results showed that the horizontal vibrating ESR process could generate small droplets and the remelting rate could be increased, which were similar to our results, although the method was different. Their result indicated that the metal pool profile obtained with the horizontal vibrating electrode was shallower. In their study, the shallower metal pool profile was ascribed to the small droplets (equivalent diameter: 10.06 mm) which provided less energy into the metal pool. In our study, the main droplets with the equivalent diameter was 6.1 mm accompanied by hundreds of satellite droplets, when the TSMF of 0.7 T was imposed. It is believed that a shallower metal pool profile can be generated due to the smaller droplets obtained in our study, which is beneficial to directional solidification and restrain the macro-segregation significantly. Meanwhile, the droplets could be compelled back and forth by the forceful EMV, and this phenomenon was captured by the high speed camera. Thus, the droplets would not be concentrated in the center area of the metal pool, which was also beneficial to get a shallower metal pool.

Apparently, the liquid droplet necks being broken up into many smaller satellite droplets and smaller main droplets indicate that the interfacial area between metal and slag can be increased a lot. So the removal efficiency of inclusions and impurities can be expected to be improved in practice production. Zhong et al.³¹⁾ studied the influence of the TSMF (0.05 T) on microstructure and properties of GCr15 bearing steel by the electroslag continuous casting method. Different current densities were adopted in their experiments, which meant that different intensities of the EMV would be generated during the remelting process. Their results showed that the impurity elements and the secondary dendrite arm spacing could be decreased with a more forceful EMV. The hardness, the impact toughness and the tensile strength of ingots could be improved by the action of the EMV, which indicated the quality improvement of the ingots produced by the MC-ESR process.

4. Conclusions

This work has provided a new way to increase the interfacial area between metal and slag by minishing the size of the main droplets and breaking up the droplet necks. Physical simulations of the ESR process were done with the aid of the transparent model apparatus. The results showed that superimposing a TSMF with a low frequency current of 5 Hz could influence the droplet evolution processes remarkably. The visualized results of the droplet evolution processes with different intensities of the TSMF were given. The mechanism of the breakup phenomenon appearing on the droplet neck was discussed. Main results obtained in this study are following:

(1) Applying a proper TSMF such as 0.5 T or 0.7 T during the ESR process with a low frequency current of 5 Hz, the shape of the liquid droplet neck will be changed from a column into a flat strip by the effect of the EMV, and then the flat strip can be separated into two slender cylinders from its middle line area. After that, the two slender cylinders will be broken up into two arrays of small satellite droplets via the Plateau-Rayleigh instability.

(2) The remelting rate can be increased by the effect of the EMV when the TSMF is below 0.5 T, and it will not be increased because the main restrictive condition becomes the intensity of the remelting current.

(3) The average size of the main droplets can be reduced and more small satellite droplets can be generated with a proper TSMF, which indicates that the technique of the MC-ESR with a low frequency current can increase the interfacial area between metal and slag enormously.

Acknowledgement

The authors gratefully acknowledged the financial support of the National Key Research and Development Program of China (2016YFB0300401), Science and Technology Commission of Shanghai Municipality (Key Project No. 13JC1402500, 15520711000), and Independent Research and Development Project of State Key of Advanced Special Steel, Shanghai University (SKLASS2015-Z021, SELF-2014-02).

References

• 1)   H.  Halfa: Steel Res. Int., 84 (2013), 495.
• 2)   J.  Reitz,  B.  Wietbrock,  S.  Richter,  S.  Hoffmann,  G.  Hirt and  B.  Friedrich: Adv. Eng. Mater., 13 (2011), 395.
• 3)   B.  Naderi and  J. A.  Mohandesi: Metall. Mater. Trans. A, 42A (2011), 2250.
• 4)   E. R.  Gutman,  V. A.  Durynin,  G. Y.  Kalinin,  O. A.  Khar’kov and  V. V.  Tsukanov: Russ. Metall., 2009 (2009), 655.
• 5)   G.  Balachandran,  M. L.  Bhatia,  N. B.  Ballal and  P. K.  Rao: ISIJ Int., 40 (2000), 478.
• 6)   A.  Choudh: ISIJ Int., 32 (1992), 563.
• 7)   Z. B.  Li,  W. H.  Zhou and  Y. D.  Li: Iron Steel, 15 (1980), 20.
• 8)   Y.  Kompan,  I.  Protokovilov,  Y.  Fautrelle,  Y.  Gelfgat and  A.  Bojarevics: Int. Scientific Colloquium Modelling for Material Proc., University of Latvia Press, Riga, (2010), 85.
• 9)   M.  Murgaš,  A. S.  Chaus,  A.  Pokusa and  M.  Pokusová: ISIJ Int., 40 (2000), 980.
• 10)   G.  Chen,  Y. B.  Zhong,  M. L.  Feng,  Z. S.  Lei,  W. L.  Ren and  Z. M.  Ren: Shanghai Met., 34 (2012), 44.
• 11)   J.  Campbell: J. Met., 22 (1970), 23.
• 12)   Y. W.  Dong,  Z. H.  Jiang,  H. B.  Cao,  Z. W.  Hou and  K. A.  Yao: Metall. Mater. Trans. B, 47 (2016), 3575.
• 13)   Y. L.  Cao,  Y. W.  Dong,  Z. H.  Jiang,  H. B.  Cao,  D.  Hou and  Q. L.  Feng: Int. J. Miner. Metall. Mater., 23 (2016), 399.
• 14)   Q.  Wang,  Z.  He,  B. K.  Li and  F.  Tsukihashi: Metall. Mater. Trans. B, 45 (2014), 2425.
• 15)   Q.  Wang,  F.  Wang,  B. K.  Li and  F.  Tsukihashi: ISIJ Int., 55 (2015), 1010.
• 16)   A.  Kharicha,  M.  Wu and  A.  Ludwig: Metall. Mater. Trans. B, 47 (2016), 1427.
• 17)   K.  Iwai and  T.  Kohama: ISIJ Int., 50 (2010), 1357.
• 18)   A.  Radjai,  K.  Miwa and  T.  Nishio: Metall. Mater. Trans. A, 29 (1998), 1477.
• 19)   C.  Vivès: Metall. Trans. B, 20 (1989), 623.
• 20)   X.  Geng,  Z. H.  Jiang,  F. B.  Liu and  H.  Peng: Adv. Mater. Res., 79–82 (2009), 1747.
• 21)   A. V.  Dub,  V. S.  Dub,  Y. N.  Kriger,  L. Y.  Levkov,  D. A.  Shurygin,  M. A.  Kissel’man,  C. M.  Nekhamin,  A. I.  Chernyak,  A. V.  Bessonov,  S. V.  Kamantsev and  S. O.  Sokolov: Russ. Metall., 12 (2012), 1031.
• 22)   A.  Kharicha,  M.  Wu,  A.  Ludwig,  M.  Ramprecht and  H.  Holzgruber: Cfd Modeling and Simulation in Materials Processing, John Wiley & Sons, Hoboken, NJ, (2012), 139.
• 23)   L. Z.  Chang,  X. F.  Shi,  H. S.  Yang and  Z. B.  Li: J. Iron Steel Res. Int., 16 (2009), 7.
• 24)   E. K.  Sibaki,  A.  Kharicha,  M.  Wu,  A.  Ludwig and  H.  Holzgruber: Proc. 2013 Int. Symp. on Liquid Metal Processing & Casting, Springer, Berlin, (2013), 13.
• 25)   C.  Vivès: Metall. Mater. Trans. B, 27 (1996), 445.
• 26)   A.  Passian and  T.  Thundat: Nature, 487 (2012), 440.
• 27)   R.  Mead-Hunter,  A. J.  King and  B. J.  Mullins: Langmuir, 28 (2012), 6731.
• 28)   S.  Shabahang,  J. J.  Kaufman,  D. S.  Deng and  A. F.  Abouraddy: Appl. Phys. Lett., 99 (2011), 161909.
• 29)   H.  Wang,  Y. B.  Zhong,  Q.  Li,  Y. P.  Fang,  W. L.  Ren,  Z. S.  Lei and  Z. M.  Ren: ISIJ Int., 56 (2016), 255.
• 30)   F.  Wang,  Q.  Wang and  B. K.  Li: ISIJ Int., 57 (2017), 91.
• 31)   Y. B.  Zhong,  Q.  Li,  Y. P.  Fang,  H.  Wang,  M. H.  Peng,  L. C.  Dong,  T. X.  Zheng,  Z. S.  Lei,  W. L.  Ren and  Z. M.  Ren: Mater. Sci. Eng. A, 660 (2016), 118.