Powder Blasting in Hot Metal Desulfurization by Mechanical Stirring Process

Abstract

The effect of adding CaO-based desulfurization flux on desulfurization efficiency in hot metal desulfurization by mechanical stirring was investigated. It was found that desulfurization flux dispersion is enhanced by powder blasting with a carrier gas. In this research, the behaviors of the desulfurization powder under different blasting conditions were investigated, and the optimum blasting conditions for achieving higher desulfurization efficiency were examined. The velocities of the gas jet and particles were measured by pressure measurement by the pitot tube technique and LDV (Laser Doppler Velocimeter), respectively. The results showed that the powder velocity was accelerated by the gas jet, which agreed with the calculated velocity. The condition of powder penetration into the hot metal was examined based on the calculations. As a result, the higher carrier gas flow rate, 200 Nl/min, was categorized as a penetrating condition, and the lower carrier gas flow rate, 100 Nl/min, was not a penetrating condition. The difference in the desulfurization behaviors under those carrier gas flow rate conditions is caused by the difference in the penetrating condition. Based on the obtained blasting conditions, powder blasting tests were carried out in 300 t-scale hot metal desulfurization, and desulfurization flux consumption was decreased by 19% compared with conventional top addition.

1. Introduction

In recent years, increasing demand for high quality steel has driven the development of new high performance steel products. In line with these trends, reduction of the impurity content in steel is required. Heightened demand for mass production high performance steels is especially remarkable in the fields of steel plates and linepipe steel. In particular, in steel products used in the development and transportation of oil and natural gas, strict specifications are applied to properties such as ductility, low temperature toughness, weldability and hydrogen-induced cracking resistance. Sulfides which form as inclusions in steel affect the mechanical properties of the steel. For example, manganese sulfide (MnS) increases the anisotropy of steel when it is elongated during hot rolling, and can also become a point of origin of rust under corrosive environments. Thus, it is essential to reduce the sulfide content to as low as 0.001%. Numerous desulfurization methods have been developed to meet performance requirements such as weldability and hydrogen-induced cracking resistance. Among these, desulfurization by the mechanical stirring method was found to be a simple and efficient technique whereby the desulfurization flux is entrained into the molten hot metal by the vortex formed by stirring with an impeller. The mechanical stirring process¹⁾ was industrialized and is now widely used by steel makers.

Many studies focused on various aspects of this process with the aim of improving desulfurization efficiency,^2,3,4,5) for example, evaluation of dispersion by water model experiments, increasing dispersion by use of an eccentric impeller, increasing agitation power and optimizing the flux composition. On the other hand, from the viewpoint of reducing flux consumption and slag generation, further enhancement of the efficiency of the desulfurization reaction is necessary.

The authors analyzed the fundamental dispersion behavior in impeller stirring and reported that suppression of flux aggregation together with improvement of flux particle dispersion are significantly important.^6,7) An investigation of the effects of flux addition methods on hot metal desulfurization with mechanical stirring revealed that continuous addition suppresses flux aggregation. It was also found that blasting the flux powder makes it possible to deliver small particles directly into the molten hot metal, which results in high desulfurization efficiency by simultaneously promoting flux dispersion and suppressing flux aggregation.⁸⁾ However, cases in which those effects were obtained and cases in which only the effect of continuous addition was obtained were observed, depending on the blasting conditions.

In this study, the behavior of blasted powder flux was investigated and the optimum conditions for desulfurization were considered. Based on the results, the blasting conditions for practical hot metal desulfurization were designed, and the effect of flux blasting on desulfurization efficiency was confirmed in a plant trial.

2. Effect of Flux Addition Method on Hot Metal Desulfurization Efficiency

According to a previous report⁸⁾ in which the effect of the flux addition method in hot metal desulfurization with mechanical stirring was investigated in water model and hot model experiments, continuous addition of desulfurization flux is effective for preventing flux aggregation, and powder blasting makes it possible to increase desulfurization efficiency and deliver flux with a smaller diameter directly into molten hot metal. The mechanism of this effect is attributed to two reasons, promotion of flux dispersion during addition and suppression of flux aggregation by reducing the addition rate. Figure 1 shows schematic diagrams of the flux dispersion and aggregation behavior with the conventional batch addition method, continuous addition and power blasting, together with SEM images of the slag after desulfurization experiments.⁸⁾

Fig. 1.

(a) Schematic diagrams of flux dispersion and aggregation mechanisms and (b) Sulfur mappings of desulfurization slag after treatment.

As an example of the experimental results, Fig. 2 shows the changes in the desulfurization ratio with ① batch addition, ② continuous addition (addition during 0–10 min) and ③ powder blasting (addition during 0–10 min) with low and high gas flow rates. With the low gas flow rate (100 NL/min), the increase in the desulfurization rate is small and the powder behavior is similar to that in continuous addition. This means that even though the powder is blasted, the effect is limited to that in continuous addition. Although the difference is assumed to be due to differences in the behavior of the blasted powder, few studies have examined the powder behavior in the blasting process.

Fig. 2.

Changes of [S]/[S]_i in hot metal experiments (conditions A, B, C and D). (Online version in color.)

Therefore, in this study, the behavior of blasted flux powder was investigated and the optimum conditions for desulfurization were considered. Based on the results, blasting conditions for practical hot metal desulfurization were designed.

3. Behavior of Powder Blasted from Top Lance

3.1. Experimental Procedures

To investigate the behavior of powder blasted from a top lance, the jet velocity was measured by Pitot tubes. The experimental conditions are listed in Table 1.

Table 1. Experimental conditions for powder blasting.

Lance	Center hole	diameter 12 mm
	Surrounding holes	10 holes
		throat diameter 3.6 mm
		exit diameter 4.2 mm
		nozzle angle: 14°
Gas flow rate	Center hole	N2, 20 Nm3/h
	Surrounding holes	O2, 310 Nm3/h
Particle	Powder	CaO
	Average particle diameter	130 μm
	Particle density	3000 kg/m3
	Feeding rate	5.2 kg/min

The lance had a center hole and 10 surrounding holes. The surrounding holes had a nozzle angle of 14° and a de Laval tube geometry. The powder used in the experiment was lime powder with the average diameter of 130 μm. The feeding rate was 5.2 kg/min. In these measurements, the lance was set horizontally, and the jet velocity and powder velocity were measured by the Pitot tube technique and LDV (Laser Doppler Velocimeter), respectively.

The jet velocity was also calculated by using the general-purpose fluid analysis code FLUENT Ver.6. The standard k-ε module was used for the turbulent model. The simulated region was a 1/10 circumferential region to the distance of 600 mm from the top lance nozzle tip.

3.2. Turbulent Jet Flow and Powder Behavior

Figure 3 shows the result of the turbulent jet flow simulation. The jet trajectory curved inward just below the nozzle outlet, and all the jets coalesced by approximately 200 mm from the nozzle.

Fig. 3.

Numerical analysis of gas jet velocity (gas flow rate: 310 Nm³/h).

The presumed mechanism is as follows. Since a multiple-hole lance was used, mutual interference occurs between the neighboring jets, creating a strong negative pressure space around the central axis of the lance. As a result, the jets from each hole are sucked inward and merge. To evaluate the validity of the numerical calculation, the jet velocity distribution was measured, and the results were compared with the calculated results. The measured velocity distribution is shown in Fig. 4, and a comparison of the measured and calculated results is shown in Fig. 5.

Fig. 4.

Velocity distribution of gas jet in radial direction. (Online version in color.)

Fig. 5.

Comparison of calculated and measured gas jet velocities in axial direction. (Online version in color.)

Fig. 6.

Velocity of gas jet and particle distribution in radial direction. (Online version in color.)

Figure 4 shows the jet velocity distribution along the radial axis for each vertical distance from the nozzle outlet. The plots indicate values measured with Pitot tubes, and the lines indicate calculated results. Although the measured values are slightly larger than calculated ones, the values are in good agreement. According to the measured values, the jet flow has not yet coalesced at 100 mm from the nozzle, but the jets gradually coalesce as the distance from the nozzle increases. The calculated result was also in good agreement with the observed jet coalescence behavior.

Figure 5 shows the maximum velocity in the axial direction. Again, the measured and calculated values show good agreement. The results also show that the jet from the center hole and those from the surrounding holes coalesce at around 200 mm from the nozzle outlet.

Consistency between the measured and calculated coalescence behavior of the jets from the center hole and surrounding holes was also confirmed under other conditions (different number of surrounding holes, throat diameter and flow rate), and the measured values also showed good reproducibility. Therefore, in the following discussion, the calculated values will be used in the discussion and explanation of jet behavior.

4. Discussion

4.1. Model of Blasted Powder Particle Velocity

The behavior of the blasted powder was modeled. When a particle flows through a fluid with a relative velocity difference, the resistance force F_D is generally expressed as shown in Eq. (1).   

$$F_D=C_d\pi r^2\left\{\frac{{\rho }_g{\left(u_g-u_p\right)}^2}{2}\right\}$$(1)

where, C_d is the resistance coefficient, r is the radius of the particle (m), u is velocity (m/s), ρ is density (kg/m³) and the subscripts g and p indicate gas and particle, respectively. The resistance coefficient C_d is expressed by Eqs. (2) and (3).⁹⁾   

$$C_d=24\left(1+0.125{Re}^{0.72}\right)/Re\hspace{1em}:Re\le 1\mathrm{\hspace{0.17em} }000$$(2)

  

$$C_d=0.44\hspace{1em}:Re>1\mathrm{\hspace{0.17em} }000$$(3)

The Reynolds number Re is expressed by Eq. (4).   

$$Re=2r\left|u_g-u_p\right|\frac{{\rho }_g}{{\eta }_g}$$(4)

The equation of motion is expressed by Eqs. (5) and (6), and the weight of a particle is expressed by Eq. (7).   

$$f=ma$$(5)

  

$$f=C_d\pi r^2\left\{\frac{{\rho }_g{\left(u_g-u_p\right)}^2}{2}\right\}+mg$$(6)

  

$$m=\frac{4}{3}\pi \cdot r^3\cdot {\rho }_p$$(7)

where, η is viscosity (Pa·s), a is the acceleration of the particle (m/s²), g is gravitational acceleration (m/s²) and m is the mass of the particle (kg).

The particle velocity was calculated by using the model described above. The calculated values are shown together with the measured values in Fig. 7. Since the jet velocity at the center hole outlet was about 30 m/s, the initial velocity of the particle was set at the same value on the assumption that the flow has developed to a steady state in the tube. The calculated jet velocity along the central axis was used for the gas velocity, and the particle velocity was successfully calculated from the jet velocity by using the model.

Fig. 7.

Comparison of calculated and measured particle velocities on central axis. (Online version in color.)

4.2. Behavior of Particle Blasted from Single Hole Lance

Using the above-mentioned model, the difference between the blasting conditions when the blasting effect is obtained and when that effect is not obtained were discussed in terms of powder velocity. Assuming the powder feeding rate in practical metal desulfurization by mechanical stirring, a single hole lance was analyzed. Laboratory experiments previously confirmed that there are three effects of blasting, namely, feeding of smaller-diameter particles, improvement of particle dispersion in hot metal, and suppression of particle aggregation. However, the effects of blasting were not obtained under all blasting conditions. Therefore, the particle velocity was analyzed for the case in which the effect of blasting was observed and for the case in which that effect was not observed.⁸⁾

The calculation conditions are listed in Table 2. A single hole lance with the throat diameter of 4 mm was used. The lance height LH (distance from lance tip to molten hot metal surface) was 50 mm. The carrier gas for blasting was nitrogen, and the flow rate was 100 NL/min or 200 NL/min. The particle diameter was 0.05, 0.1 or 1.0 mm. The initial velocity was set to 30 m/s for the previously mentioned reason.

Table 2. Experimental conditions for 70 kg-scale hot metal experiments.

Lance	diameter 4 mm (straight)
Lance height	50 mm
Gas flow rate for powder blasting	200 Nl/min 100 Nl/min
Powder density	3000 kg/m3
Initial velocity of powder	30 m/s

First, the powder velocity was calculated with the nitrogen flow rate of 200 NL/min as a condition under which the effect of blasting was observed in a laboratory experiment. The jet behavior was analyzed using the single jet behavior model by Ito and Muchi.¹⁰⁾ The parameters are determined by fitting with the experimental results.¹¹⁾

The jet velocity and particle velocity are shown in Fig. 8. The particle is accelerated by the high velocity jet immediately after leaving the nozzle outlet. Acceleration then weakens as the jet velocity decreases, and finally the particle travels at a basically constant velocity by inertia. The maximum velocity that a particle achieves depends on the particle size. Although the maximum velocity increases as the particle size decreases, the velocity of smaller particles decreases more rapidly. The influence of the initial jet velocity on these behaviors was estimated by calculating the case of the initial particle velocity of 60 m/s. Although the initial velocity was doubled, the maximum particle velocity was almost the same. Therefore, the influence of the initial particle velocity is considered to be small.

Fig. 8.

Calculated gas and particle velocity on central axis (Powder blasting gas flow rate: 200 Nl/min). (Online version in color.)

Secondly, the particle velocity was calculated with the nitrogen flow rate of 100 NL/min as a condition under which the effect of blasting was not observed in the laboratory experiment. Since the jet from the lance is a free jet flow under this condition, the jet behavior was calculated by Eq. (8).¹²⁾   

$$u_m=12r_o{u}_o/x$$(8)

where, u_m is the maximum velocity of the jet at the central axis (m/s), r_o is the hole radius (m), u_o is the jet velocity at the nozzle outlet (m/s) and x is the distance from the nozzle outlet (m).

The calculated jet velocity and particle velocity are shown in Fig. 9. As acceleration is smaller, and the particle velocity is calculated to be up to 30–60 m/s.

Fig. 9.

Velocity of particle distribution on central axis (Powder blasting gas flow rate: 100 Nl/min). (Online version in color.)

4.3. Criteria for Particle Penetration into Hot Metal

For the flux powder to penetrate into molten hot metal and disperse, it is necessary to accelerate the blasted powder to the critical penetration velocity. The penetration velocity is discussed in the following. The fundamental study of the penetration behavior of a solid particle into a liquid metal was reported by Ozawa et al.¹³⁾ When the velocity of a solid particle at the surface of a liquid metal is ν_C, the critical Weber number is represented as Eq. (9), and the criteria reported by Ozawa et al. are expressed by Eqs. (10) and (11).   

$$W{e}_c=\frac{r_p{v}_c{}^2{\rho }_l}{\sigma }$$(9)

  

$$\begin{array}{l}W{e}_c=\\ \frac{1}{0.044}\left\{\left(1-exp\left(\frac{0.66}{{\rho }^{*}+1/4}\right)\right)\left(\frac{{\rho }^{*}+1/4}{0.33}-1+cos{\theta }_c\right)+2\right\}\end{array}$$(10)

  

$${\rho }^{*}=\frac{{\rho }_p}{{\rho }_l}$$(11)

where, We_c is the critical Weber number (–), d_p is the diameter of the particle (m), v_C is the critical velocity (m/s), θ_c is the contact angle between the particle and the liquid (°), ρ_p is the particle density (kg/m³), ρ_l is the liquid density (kg/m³), ρ^* is the density ratio (–) and σ is the surface tension of the liquid (N/m).

On the other hand, in the case of low wettability, like that in a molten CaO-molten hot metal system, Ogawa and Matsumoto¹⁴⁾ reported that the critical velocity becomes larger because of a decrease in apparent density. They reported the critical Weber number using the parameter α (ratio of associated film thickness to particle diameter) as shown in Eqs. (12), (13), (14), (15).   

$$W{e}_c=\frac{1}{0.11}\left\{\left(1-exp\left(\frac{0.66}{{\rho }^{*2}}\right)\right)\left(\frac{{\rho }^{*2}}{0.33}-1+cos{\theta }_c\right)+2\right\}$$(12)

  

$${\rho }^{*2}=\frac{{\rho }_p\prime }{{\rho }_l}-\frac{1}{4}$$(13)

  

$${\rho }_p\prime =\frac{W_p}{\left(V_p+V\right)}$$(14)

  

$$\alpha =\frac{\mathrm{\Delta }r}{d_p}$$(15)

where, V is the volume of the associated film (m³), V_p is the volume of the particle (m³), W_p is the weight of the particle (kg) and Δr is the thickness of the associated film (m), ρ_ρ is the apparent density of the particle (kg/m³) and ρ^*2 is the density ratio (–).

Assuming the main component of desulfurization flux, the critical velocity in a CaO-molten hot metal system was calculated based on the criteria reported by Ozawa et al. and Ogawa and Matsumoto. The result is shown in Fig. 10. The calculation parameters were as follows: θ_c=132 (°), ρ_l=7000 (kg/m³₎, ρ_p=3000 (kg/m³) and σ=1.5 (N/m). The calculation results for the maximum velocity of the blasted particle, which were previously discussed in section 4.2, are also shown.

Fig. 10.

Relationship between particle diameter and critical CaO velocity for penetration into molten hot metal. (Online version in color.)

According to the criteria of Ozawa et al., when the diameter of a CaO particle is from 10 μm to 500 μm, the particle penetrates into molten hot metal when the particle velocity is approximately 5–40 m/s. On the other hand, according to the equation proposed by Ogawa and Matsumoto, the critical velocity is about 10–100 m/s, which means a larger velocity is necessary for penetration. The penetration behavior of particles with low wettability, for example, in a CaO-molten hot metal system, has not been fully discussed so far, and further discussion is necessary. However, for a CaO particle to penetrate into molten hot metal, it is assumed that the velocity must be larger than the value calculated by the criteria of Ozawa et al., in other words, larger than 100 m/s.

With the nitrogen flow rate of 200 Nl/min, blasted powder particles with diameters of 1 mm to 0.01 mm are accelerated and reach a velocity of approximately 40–210 m/s. This is much larger than the critical velocity calculated by the equation proposed by Ogawa and Matsumoto, which means that the velocity of the blasted powder is sufficiently large in comparison with the critical velocity, and there is a high possibility of penetration. On the other hand, when the nitrogen flow rate is 100 Nl/min, the maximum velocity of the blasted powder particles is only about 30–95 m/s, and the maximum velocity of particles smaller than 0.03 mm does not exceed the critical velocity for penetration according to Ogawa and Matsumoto. From these calculation results, the particle penetrates the surface and small particles disperse in the molten hot metal, when flux is blasted with a nitrogen carrier gas with the flow rate of 200 m/s. Enhancement of desulfurization by blasting was observed in hot model experiments under this condition. However, in the case of the nitrogen flow rate of 100 Nl/min, the particle velocity was not high enough for a large portion of the flux to penetrate the surface, and as a result, enhancement of desulfurization is attributed only to continuous addition.

These results suggest that, for flux powder to be added directly and disperse into molten hot metal, the particle velocity at the surface of the molten hot metal must be larger than the critical velocity for penetration.

5. Plant Experiment in Practical Desulfurization Process

The plant experiment was carried out at a 300 ton-scale desulfurization facility (steelmaking shop at East Japan Works Keihin District, JFE Steel Corporation). A schematic diagram of the facility is shown in Fig. 11, and specifications are shown in Table 3. A single hole lance for flux blasting was installed on the impeller carriage, and CaO was used as the desulfurization flux. The particle velocity at the molten hot metal surface and the critical velocity for penetration were calculated by Eqs. (1), (2), (3), (4), (5), (6), (7) for lime powder of the average particle size of the lime used, and the blasting conditions were determined so that the surface velocity was larger than the calculated critical velocity for penetration according to Ogawa and Matsumoto, assuming that wetting of the particle does not occur (α=1). Blasting was started after stirring by the impeller had generated a vortex so that the flux could be entrained into the molten hot metal by the vortex.

Fig. 11.

Schematic diagram of powder blasting equipment in mechanical stirring process at East Japan Works (Keihin District).

Table 3. Specifications of powder blasting at East Japan Works (Keihin District).

Heat size	300 t /heat
Metal	[S]i = 0.025–0.030 mass%
Initial temperature	1523–1663 K
Gas and gas flow rate for powder blasting	N2 6–20 Nm3/min
Lance	1 hole (straight)
Lance height	0.3–1.4 m
Desulfurization flux	Powder	CaO
	Feeding rate	100–400 kg/min

Figure 12 shows CaO efficiency for desulfurization for the cases of flux top addition (Batch addition) and flux powder blasting. CaO efficiency is higher at high temperatures and higher with blasted charges. The difference between the CaO blasted charges and the CaO top-added charges was larger at lower temperatures. Figure 13 shows a comparison of the unit consumption of desulfurization flux with top addition and powder blasting. Unit consumption of the desulfurization flux decreased by 19% with powder blasting.

Fig. 12.

Relationship between hot metal temperature and CaO efficiency for desulfurization. (Online version in color.)

Fig. 13.

Effect of powder blasting on desulfurization flux consumption.

6. Conclusion

The behavior of blasted desulfurization flux powder was investigated, and the blasting conditions for achieving the full effect of powder blasting were discussed. Powder blasting tests were carried out under the obtained blasting conditions at a 300 t-scale hot metal desulfurization facility, and desulfurization flux consumption decreased by 19% compared with conventional top addition.

(1) The jet velocity and particle velocity were measured in order to evaluate the behavior of the blasted powder. Particles were accelerated by the surrounding jet, depending on the jet velocity. Acceleration was achieved with a high jet velocity.

(2) The velocity of the blasted powder calculated by model equations showed good agreement with the measured results. The particle velocity was successfully calculated from the jet velocity.

(3) The behavior of blasted powder was analyzed by using the above-mentioned model. Under the condition of a nitrogen gas flow rate of 200 NL/min, at which the effect of blasting was observed in a previous hot model experiment, the velocity of the particle at the molten hot metal surface was approximately 40–200 m/s. However, at the nitrogen gas flow rate of 100 NL/min, at which the effect of blasting was not observed, the velocity at the molten hot metal surface was approximately 30–95 m/s, which is not sufficient for smaller particles to disperse directly into the molten hot metal . This is considered to be the reason why only the effect by continuous addition was observed when powder blasting was performed at this relatively low velocity.

(4) Blasting conditions for practical hot metal pretreatment were designed based on the model, and the effect of blasting was confirmed in a 300 t-scale plant test. Improvement of desulfurization was observed, and flux consumption decreased by 19%. It is considered that the effects of both promotion of flux dispersion and suppression of flux aggregation were achieved under the designed conditions.

Symbols

Nomenclature

a: Acceleration of particle (m/s²)

Cd: Resistance coefficient

d_p: Diameter of particle (m)

g: Gravitational acceleration (m/s²)

m: Weight of particle (kg)

Re: Reynolds number (–)

r: Radius of particle (m)

r_o: Radius of hole (m)

u: Velocity (m/s)

u_m: Maximum velocity of jet at central axis (m/s)

u_o: Jet velocity at nozzle outlet (m/s)

V: Volume of associated film (m³)

V_p: Volume of particle (m³)

v_C: Critical velocity (m/s)

W_p: Weight of particle (kg)

We_c: Critical Weber number (–)

x: Distance from nozzle outlet (m)

Δr: Thickness of associated film (m)

η: Viscosity (Pa·s)

θ_c: Contact angle between particle and liquid (°)

ρ_g: Gas density (kg/m³)

ρ_l: Liquid density (kg/m³)

ρ_p: Particle density (kg/m³)

${\rho }_p^{}\prime$: Apparent density of particle (kg/m³)

ρ^*: Density ratio (–)

ρ^*2: Density ratio (–)

σ: Surface tension of liquid (N/m)

References

• 1)   K.  Kanbara,  T.  Nisugi,  O.  Shiraishi and  T.  Hatakeyama: Tetsu-to-Hagané, 58 (1972), 34.
• 2)   K.  Nakanishi,  N.  Bessyo,  S.  Takada,  A.  Ejima,  M.  Kuga,  J.  Katsuki and  M.  Kawana: Tetsu-to-Hagané, 64 (1978), 1528.
• 3)   N.  Kurokawa,  S.  Matsuo,  H.  Jouguchi,  K.  Yamada and  Y.  Watanabe: Sumitomo Met., 45 (1993), 52.
• 4)   N.  Kikuchi,  S.  Nabeshima,  S.  Takeuchi,  T.  Yamauchi,  Y.  Kitano and  S.  Ogura: Tetsu-to-Hagané, 90 (2004), 322.
• 5)   N.  Kikuchi,  S.  Nabeshima and  Y.  Kishimoto: ISIJ Int., 52 (2012), 1809.
• 6)   Y.  Nakai,  I.  Sumi,  H.  Matsuno,  N.  Kikuchi and  Y.  Kishimoto: ISIJ Int., 50 (2010), 403.
• 7)   Y.  Nakai,  I.  Sumi,  N.  Kikuchi,  Y.  Kishimoto and  Y.  Miki: ISIJ Int., 53 (2013), 1411.
• 8)   Y.  Nakai,  I.  Sumi,  N.  Kikuchi,  Y.  Kishimoto and  Y.  Miki: Tetsu-to-Hagané, 101 (2015), 1.
• 9)   Z.  Tanaka and  K.  Iinoya: J. Chem. Eng. Jpn., 3 (1970), 261.
• 10)   S.  Ito and  I.  Muchi: Tetsu-to-Hagané, 55 (1969), 1152.
• 11)   I.  Sumi,  Y.  Kishimoto,  Y.  Kikuchi and  H.  Igarashi: ISIJ Int., 46 (2006), 1312.
• 12)   T.  Shakouchi: Jet Flow Engineering, Morikita Publishing, Tokyo, (2004), 34.
• 13)   Y.  Ozawa,  K.  Suzuki and  K.  Mori: Tetsu-to-Hagané, 69 (1983), 753.
• 14)   K.  Ogawa and  H.  Matsumoto: CAMP-ISIJ, 10 (1997), 80.