Decarburization and Inclusion Removal Process in Single Snorkel Vacuum Degasser

Fluid flow model

The argon-steel two-phase flow in RH and SSF was assumed to be a Newtonian incompressible steady isothermal flow. Besides, the gas bubbles are assumed to be spherical and their interactions leading to coalescence and breakup were neglected. And the effect of the fluctuation of free surface on fluid flow is also negligible [2, 3, 11, 12, 13, 14, 15, 16]. The key governing equations for fluid flow consist of the continuity equation and Navier–Stokes equation for gas–liquid phases, the turbulence energy and dissipation rate equations.

Besides, the boundary conditions for fluid flow model are described as follows. For all nodes at the refractory walls in RH degasser, the wall function method was applied, and the normal gradients of pressure and gas volume fraction were also set to zero. For the free surfaces in ladle and vacuum chamber, the symmetry boundary condition was imposed. Furthermore, the gas bubbles reaching the free surface were assumed to escape at the flotation velocity.

Inclusion removal model

The inclusions are assumed to be spherical alumina, and begin to have a uniformly distributed in the liquid steel. The total inclusion mass fraction is so small that the effect of inclusions on fluid flow is negligible. The industrial trial results show that the fractional inclusion number density had an exponential relationship with the inclusion radius [17, 18, 19] and can be expressed as: fr=Ae−Br. The inclusion number density, the inclusion volume concentration and the characteristic inclusion radius can be expressed as: N∗=∫0∞frdr=AB, C∗=∫0∞43πr3frdr=8πAB4 and r∗=63/B respectively. So the inclusion transport equations which describe collision and aggregation among inclusions can be expressed as follows [13, 20]:

Here, ρl and ρp are the density of liquid steel and inclusion, kg/m3; νl is the kinematic viscosity of liquid steel, m2/s; g is the gravitational acceleration, m/s2; Deff is the effective diffusion coefficient, m2/s; the source term SN∗ accounts for the effect of bubble adhesion and coalescence among inclusions on the inclusion number density, while the source term SC∗ accounts for the effect of bubble adhesion on the inclusion concentration [2, 13]

Here, the value of Hamaker constant A∗ is 0.48×10−20J for alumina inclusion [21]; r is the initial size of the monomer particle, m; rb is the bubble radius, m; ug is the bubble flotation velocity, m/s; αg is the gas volume fraction.

At the top slag, 80 % of inclusions reaching top slag are assumed to be removed while the remaining 20 % of inclusions are entrained into the liquid steel [20]. The inclusion adhesion to the refractory wall can be treated as the mass diffusion of boundary layer [2, 13, 20, 21]. Moreover, at the ladle bottom, the reverse effect of inclusion floatation velocity on the inclusion adhesion has also been taken into account. Thus, the boundary fluxes for inclusion number density and concentration can be found in our previous work [2, 13, 22].

Decarburization model

The decarburization in RH and SSF is assumed to be taken place at the free surface of vacuum chamber, the inner site of liquid steel and the bubble surface of gas–liquid plume [23, 24, 25, 26]. Besides, the decarburization rate is controlled by the mass transfer of carbon and oxygen in liquid steel [24, 25]. And the concentration of carbon and oxygen at the gas–liquid interface are in equilibrium with CO partial pressure in gas phase [23, 24, 26]. The governing equations for calculating the concentration distribution of carbon and oxygen can be represented as [24],

where ϕ represents the mass fraction of carbon and oxygen; μeff is the effective viscosity, Pa⋅s; Sc is the Schmit number; Sϕ is the source or sink of carbon and oxygen and the detailed expression can be found in our previous work [2, 22].

Solution method and parameters

The solution procedure can be described as follows. Firstly, the computational fluid dynamics package, CFX, was used to calculate the steady gas–liquid flow. Then the unsteady conservation equations were solved by CFX in order to obtain the spatial distribution of carbon concentration, oxygen concentration, inclusion number density and volume concentration. The convergence criteria is that the value of the root mean square normalized residual for variables was less than 1×10−5 and the global imbalances, which means the ratios of the difference between the total input mass flux and the total output mass flux to the total input gas mass flux was less than 0.1 %. As shown in Table 1, except the snorkel diameter, the dimension parameters of ladle and vacuum chamber in RH were identical to those in SSF. And the related calculation and property parameters were listed in Table 2.

Turbulence kinetic energy in single snorkel degasser

Figure 3(a) shows the turbulence kinetic energy distribution in RH degasser with 1400 NL/min of gas flow rate. The maximum of turbulence kinetic energy in RH appears near the free surface of the vacuum chamber. And the turbulence kinetic energy at the up and down snorkels is much greater than that in the ladle, which means that the mixing in the ladle is relatively weaker in contrast with that in the snorkel and the vacuum chamber. Figure 3(b) shows the turbulence kinetic energy distribution in SSF with 1400 NL/min of gas flow rate. The maximum of turbulence kinetic energy in SSF also appears near the free surface of the vacuum chamber, while the value is less than that in RH degasser on the condition of the same gas flow rate. Moreover, the turbulence kinetic energy of ladle in SSF becomes much greater than that in RH degasser, which leads to a better mixing.

Figure 3:

Distribution of turbulence kinetic energy at the main plane. (a) RH (b) SSF.

Decarburization in single snorkel degasser

Figure 4 shows that the decarburization rate in SSF is much greater than that in RH degasser. The key reason is that the turbulent flow in snorkel and vacuum chamber is more drastic in SSF than that in RH degasser, so the mass transfer coefficient in snorkel and vacuum chamber in SSF is much greater than that in RH. Besides, the circulation flow rate in SSF is greater than that in RH degasser [10]. And the mixing time decreases with the increasing circulation flow rate, which is in favor of decarburization. Moreover, the residence time of argon bubbles injected from ladle bottom is much greater than that from the up snorkel, so the gas–liquid interfacial area in SSF is also greater than that in RH degasser, which can promote the decarburization rate. Furthermore, for decarburization in RH and SSF, the decarburization rate is controlled by the mass transfer of carbon from liquid steel to the gas–liquid interface. The overall decarburization reaction rate can be described by the following expression, which is regarded as a first-order reaction,

Figure 4:

Variation of average carbon mass fraction with time.

Here, wav,C is the average carbon concentration, we,C is the carbon equilibrium concentration, kdec is the volumetric mass transfer coefficient for decarburization and can be obtained by regression analysis. As shown in Figure 4, it can be found on the condition of the same gas flow rate, the volumetric mass transfer coefficient in SSF is almost as that in RH, the volumetric mass transfer coefficient in SSF is almost twenty times bigger than that in RH, which leads to a much greater decarburization rate in SSF.

Figure 5 shows the distribution of carbon mass fraction at main section in RH and SSF. As the major decarburization site, the carbon mass fraction in vacuum chamber and snorkel is much less than that in ladle. Besides, because of the shorter residence time, the carbon mass fraction in vacuum chamber decreases along with the flow direction. Moreover, Figure 5(a) shows that the carbon mass fraction under up snorkel is great. The reason is that the static pressure in ladle is so great that the critical nucleation pressure of CO bubbles can hardly reach. Consequently, the decarburization reaction can not occur. Figure 5(b) shows that the carbon mass fraction in SSF is much less than that in RH. And the distribution of carbon mass fraction in SSF is quite different with that in RH. It should be noted that the greater mass fraction of carbon at the recirculation zone under up snorkel in RH do not appear in SSF.

Figure 5:

Distribution of carbon mass concentration at the main plane after 240 sec. (a) RH (b) SSF.

Inclusion removal in single snorkel degasser

Figure 6 shows the inclusion removal rate in RH and SSF during the whole process. The inclusion removal rate in SSF is much greater than that in RH. It can be found that the inclusion removal rate is nearly 90 % after 1,000 sec in RH, while the inclusion removal rate is nearly 90 % only after 300 sec, which means SSF has an advantage in inclusion removal. The main reason is that the turbulence kinetic energy is greater in SSF than that in RH, which means that the collision and coalescence among inclusions in SSF is more effectively than that in RH. Besides, inclusions after coalescence can be removed with a larger flotation velocity. Moreover, the inclusion removal in RH and SSF can be assumed as a mass transfer process with a direction from liquid steel to slag, refractory wall and bubble surface. The overall inclusion removal rate can be described by the following expression, which is regarded as a first-order reaction,

Figure 6:

Variation of inclusion removal rate with time.

Here, Cav,C is the average inclusion volume fraction, Cav,C is the final inclusion volume fraction, kinc is the volumetric mass transfer coefficient for inclusion removal and can be obtained by regression analysis. As shown in Figure 6, the volumetric mass transfer coefficient for inclusion removal in SSF is about three times of that in RH, which leads to a much greater inclusion removal rate in SSF.

Figure 7 shows the distribution of inclusion volume fraction after 200 sec. By comparing Figures 3 with 7, it can be found that, in RH and SSF, the inclusion volume fraction is smaller in the regions where the turbulence kinetic energy is greater, while the inclusion volume fraction is greater in the regions where the turbulence kinetic energy is smaller. Such a phenomenon indicates the turbulence kinetic energy has a great effect on the inclusion removal. For example, Figure 3(a) shows that turbulence kinetic energy between sidewall and two snorkels and under the up snorkel is much weaker than other regions, which is not in favor of inclusion growth. So inclusions tend to accumulate in these regions, especially for those smaller inclusions. In this way, Figure 7(a) shows that the inclusion volume fraction between sidewall and up snorkel in ladle is greater, while the inclusion volume fraction is annular at the recirculation zone under the up snorkel. On the other hand, Figure 3(b) shows that the turbulence kinetic energy is very weak only near the ladle bottom. And the size of dead zone in SSF is much smaller than that in RH. Besides, since the inclusions attached to gas bubbles injected from ladle bottom in SSF can be removed directly by slag entrapment with a greater floatation velocity. The inclusion volume fraction fall away in two-phase plume zone, which leads to the disappearance of accumulation of inclusions in ladle of RH.

Figure 7:

Distribution of inclusion volume fraction at the main plane after 200 sec. (a) RH (b) SSF.

Figure 8 shows that the variation of inclusion number density of different sizes ranges with the time. For inclusions smaller than 5 μm, the number density in both RH and SSF decreases drastically during 0~1,000 sec. For inclusions in the range of 5~10 μm, the number density in SSF increases a little during the first 200 sec, and then decreases, while the number density in RH decreases during the whole 0~1,000 sec. For inclusions in the range of 10~15 μm, the number density in SSF increases by almost five times after about 300 sec, while the number density in RH increases a little during the first 100 sec, and then decreases. For inclusions larger than 15 μm, the number density in SSF increases by about three times after about 400 sec, while the number density in RH increases a little during the first 100 sec, and then decreases. The main reason for above phenomena is the collision and growth of inclusions. At the beginning, the increase of newly bigger inclusions is greater than the decrease of removed inclusions and the number of bigger inclusions increases. Then while the inclusions were removed gradually, the increase of newly bigger inclusions becomes slowly, the number of bigger inclusions also decreases.

Figure 8:

Variation of inclusion number density with time in RH and SSF. (a) 0~5 μm (b) 5~10 μm (c) 10~15 μm (d) 15~μm.

Besides, for inclusions larger than 15 μm, Figure 8(d) shows that the terminal number density in RH becomes even larger than the initial value, while the terminal number density in SSF is much smaller than the initial value, which means the 1000 sec is not enough for RH to remove inclusions larger than 15 μm, but is enough for SSF to remove inclusions larger than 15 μm. Thus, in order to removal inclusions larger than 15 μm, the SSF can reduce treatment time.