Numerical and Experimental Study on Fluid Dynamic Features of Combined Gas and Electromagnetic Stirring in Ladle Furnace

Value Paper Authors: Ulf Sand, Hongliang Yang, Jan-Erik Eriksson, and Rebei Bel Fdhila

2 Fluid Dynamic Features of Combined Gas and Electromagnetic Stirring

Basic fluid dynamic features of combined electromagnetic stirring, EMS, and gas stirring (EMGAS) have been studied in the present work. A transient and turbulent multiphase numerical flow model has been built. Simulations of a real size ladle furnace have been conducted for 7 cases, operating with and without combined stirring and varying the argon gas inlet plug position. Results of these simulations are compared considering melt velocity, melt turbulence, melt/slag-interface turbulence and dispersion of gas bubbles. An experimental water model has also been built to simulate the effects of combined stirring. The water model has been numerically simulated and visual comparison of the gas plume shape and flow pattern in the numerical and in the experimental model has also been done for 3 flow situations. The results show that EMGAS has a strong flexibility regarding the flow velocity, gas plume, stirring energy, mixing time, slag layer, etc.

IntroductionThe demands on the flow control in the ladle vary for different functions and stages in a ladle refining cycle. With the development of metallurgical knowledge, some common requirements on the fluid control in the ladle have been set. A strong turbulence or mixing at the slag/steel interface is required to achieve good desulphurization result. A short mixing time in the bath melt can lead to quick chemical homogeneity. Mixing time is affected by flow pattern, ladle size, turbulence intensity, etc. A relative calm flow is good for the inclusion floatation and avoiding re-oxidation, slag entrapment, etc.

Gas stirring and electromagnetic stirring (Figure 1) have been the two dominant stirring methods in a ladle furnace [1]. The strong slag/metal interaction under a high gas flow rate is preferable for desulphurization. The bubble cleaning effect with a low gas flow rate has also been widely utilized to remove the inclusions. An Electromagnetic Stirrer (hereafter EMS) can create a vigorous stirring in the melt while keeping the whole melt surface covered by the slag layer. As a result of this feature, the yield of alloy additions and steel cleanliness is

higher with electromagnetic stirring. The strength of EMS can be accurately controlled which leads to high reproducibility and operational flexibility. The stirring directions by EMS can also be changed between upward and downward stirring. The combination of electromagnetic and gas stirring (hereafter EMGAS) provides the possibility to utilize the advantages of both methods and further improve the performance of a ladle furnace. The basic idea of EMGAS is to control the gas bubbles in the melt, slag/metal interface, melt mixing, etc by means of the combined effect of EMS and gas stirring.

The optimal porous plug positions in the ladle bottom relative to the electromagnetic stirrer and the operating parameters of both gas and electromagnetic stirring are the primary questions to be answered in this project. These questions can only be answered when the fluid dynamic features and their metallurgical potentials are fully understood. The final targets of EMGAS are to be able to shorten the refining time in the ladle furnace, improve the steel cleanliness and/or reduce the energy consumption.

A transient and turbulent multiphase numerical flow model has been built and is used to simulate both real plant ladle furnaces as well as an experimental ladle furnace water model. Real plant simulations have been conducted for 7 flow cases, running with and without combined stirring and varying the argon gas plug position, considering a full size ladle. In the model, melt, slag and gas phases are included. Numerical results of these simulations are compared considering melt velocity, melt turbulence, melt/slag-interface turbulence and dispersion and

Figure 1. Fluid flow features of gas, electromagnetic and combined stirring.

Argon Argon

LF-EMS

LF-EMS

Numerical and Experimental Study on Fluid Dynamic Features of Combined Gas and Electromagnetic Stirring in Ladle Furnace

ABB Value Paper | 3

contact time of gas bubbles. The experimental water model has also been numerically simulated and visual comparison of the gas plume shape and flow pattern in the numerical and in the experimental model are presented for 3 different stirrer configurations. New features associated with the EMGAS regarding the gas plume, stirring energy, velocity, mixing time etc are introduced and discussed in the present work. Metallurgical potentials are also discussed.

Process modellingPhysical system. The ladle system simulated in this paper is shown in Figure 2. The ladle inner diameter is 2.85 m, the melt height is 2.80 m and the initial slag layer thickness is 0.1 m.

Numerical methods. The modelling of the melt, slag and the gaseous atmosphere in the ladle top is conducted using the Volume-Of-Fluid, VOF, approach. This formulation relies on the fact that the fluids are not interpenetrating. The fields for all variables and properties are shared by the phases and represent volume-averaged values, as long as the volume fraction of each phase is known at each location. The mathematical model of the VOF approach consists of the continuity equation for each phase, the momentum equations based on one velocity and pressure field. In the present model there are 3 phases: melt, slag and gas as indicated in Figure 2. The turbulence of the flow system is modelled using the Reynolds stresses model which accounts for six equations and the turbulent dissipation equation.

Bubble jet modelling. The argon gas injected through the porous plug in the ladle bottom is modelled using the Lagrangian particle tracking method where every individual bubble trajectory is calculated and source terms are taken into account to properly represent the interaction between the bubbles and the continuous phases for the momentum and heat. In the present study, a bubble size distribution is

considered at the argon gas inlet. The bubble drag and added mass forces as well as the bubbles hydrodynamic expansion are taken into account. The drag coefficient, CD, model of Morsi and Alexander [2] is used and the bubbles turbulence the Random Walk Model is applied. The bubble compression or expansion due to hydrodynamic pressure is modelled by adjusting the density and the volume of the bubble while maintaining constant bubble mass.

Simulated cases. The case list of the real plant numerical simulations is shown in Table 1. For comparative reasons constant argon gas flow rate is used for gas stirring and constant current is used for EMS stirring, for all cases. Two different gas plug positions have been considered through this experimental program.

Case: EMS direction: EMS current: Argon gas flow rate: Gas plug position, angle: Gas plug position, radius:

[-] [A] [dm3STP/min] [°] R

1 - 200 180 R/2

2 Upward 1350 - - -

3 Downward 1350 - - -

4 Upward 1350 200 180 R/2

5 Downward 1350 200 180 R/2

6 Upward 1350 200 28 R/2

7 Downward 1350 200 28 R/2

Figure 2. Ladle geometry and physical system.

Figure 3. Experimental facility of water model

Case: Water pump direction: Air flow rate: Gas plug position, angle: Gas plug position, radius:

[-] [dm3STP/min] [°] [R]

1W - 25 180 R/2

2W Upward 25 180 R/2

3W Downward 25 180 R/2

Table 2. Case list of compared experimental and numerical modelling.

Table 1. Case list of numerical simulations.

4 Fluid Dynamic Features of Combined Gas and Electromagnetic Stirring

Water model. The ladle furnace water model experimental set-up consists of a cylindrical vessel of 1.5 m height and 1.0 m internal diameter, filled with water until 0.82 m. A porous plug is positioned at the bottom of the tank to facilitate gas stirring and eight immersed pumps, located near the wall, to simulate the EMS effect, four placed at the bottom to pump the liquid upwards and four placed under the water free surface to pump the liquid downwards. Figure 3 shows a schematic sketch of the experimental facility.

As a first step, the water model experiments are mainly based on visualization using an advanced video camera. The water model case list is shown in Table 2. The visualization investigation that we present is focusing mainly on the flow pattern and the gas plume structure.

ResultsGas plume. The plume shape of gas stirring can be greatly influenced by EMS. In Figures 4, 5 and 6, the water model cases are plotted. Comparing the results from the numerically simulated water model to the physical water model we see similar flow pattern. The three cases show an excellent agreement in terms of plume shape and bubble distribution in the ladle. Similar to what we observe in the experiment, cases 2W and 3W show a lot more bubbles circulating far from the main gas plume, especially in case 3W. In the numerical simulation the total number of bubbles in the water is approximately twice as many in case 3W than in case 1W. This indicates that combined stirring can have a huge positive effect in terms of cleaning the melt from impurities, judging solely on the number of simulated bubbles in the water. In reality the bubbles may break-up or coalesce under the effect of the external flow. Bubble break-up leads to significantly increase the number of bubbles and consequently their interfacial area. However, coalescence decreases the interfacial area and can also have a direct influence on the bubbles jet structure as can be seen in Figure 5 where the curvature of the gas plume is smaller in the experimental case than in the simulation due to the influence of the larger bubbles created by coalescence. Coalescence and break-up are not considered in the present model.

Flow field. The flow velocity field and the contours of the gas volume fraction are plotted in Figure 7, for the real plant simulations, i.e. case 1 to 7. It is found that the average melt velocity is mainly governed by the EMS and showed in Figure 8 the cases including EMS take approximately 25 seconds after start of stirring reach fully developed flow with an approximate velocity of 0.5 m/s for upward stirring and 0.7 m/s for downward stirring. In case 1, only utilising gas stirring, the calculated final velocity is significantly smaller, around 0.05m/s. This is a clear illustration that the mixing induced by the EMS is much more intense. When using the EMS, the presence of the argon gas and the position of its injection seem to have small

Figure 4. Water model plume shape, case 1W, numerical (left) and physical (right) | Figure 5. Water model plume shape, case 2W, numerical (left) and physical (right). | Figure 6. Water model plume shape, case 3W, numerical (left) and physical (right).

4

5

6

Figure 7. Final melt velocity, slag interface and gas plume distribution at 40s.

ABB Value Paper | 5

influence on the mean velocity in the melt. Similar conclusions are drawn from the results shown in Table 3 where the mean turbulent kinetic energy created by the downward stirring is significantly higher than the other cases. Solely operating with gas stirring creates a turbulence level one order of magnitude lower than the cases with EMS.

Stirring energy. Stirring energy is equivalent to the energy input rate into the liquid vessel. Specific stirring energy (stirring energy per unit mass) has been widely employed to evaluate the mixing process in the melt. For gas stirring system, the stirring energy is an energy balance with potential energy of the rising bubbles and the thermal expansion work done by the bubbles. Several empirical formulas for gas stirring have been proposed based on this principle [3]. For electromagnetic stirring system, the stirring energy can be calculated by the mechanical power input rate

(1)

Where is the stirring force per unit volume, is the velocity of liquid steel, is the weight of liquid steel, is the unit volume.

In theory, the specific stirring energy input rate shall be equal to the dissipation rate of mechanical energy per unit mass of the system. The dissipation rate of mechanical energy can be calculated as:

(2)

Where:

(3)

(4)

is the mean rate of the strain, is the fluctuating rate of the strain, is the viscous dissipation rate of the mean flow, is the dissipation rate of the turbulent energy. represents the i-th component of the mean melt velocity and is its fluctuating term that represents the turbulence. The specific stirring energy can be calculated as:

(5)

Equation (5) has been used in this work to calculate the stirring energy in gas stirring system, electromagnetic stirring system, and combined stirring system.

The advantages of the Equation (5) are that the stirring energy is calculated through the integration in the whole liquid volume, the effects of all the factors including slag layer, free surface, stirring force, etc are accounted through their influence on the flow velocity and turbulence.

Figure 8. Time history of average melt velocity.

Table 3. Time average turbulent kinetic energy in the melt, considering fully developed flow.

Case 1 2 3 4 5 6 7

Turbulent kinetic energy, k , [m2/s2] 0.005 0.050 0.089 0.054 0.092 0.057 0.085

Figure 9. Time history of stirring energy in the melt.

6 Fluid Dynamic Features of Combined Gas and Electromagnetic Stirring

Figure 9 shows the calculated stirring energy for cases 2 and 3, as a function of time. It clearly shows the increase of stirring energy during the accelerating stage. For case 3 there is a time delay, compared with case 2, before reaching the initial peek. The main reason for this is due to the fact that in case 3 the accelerated fluid first has to pass by the bottom and the opposite side wall before reaching the slag layer whilst in case 2 the accelerated fluid immediately reaches the slag. This implies that a big contribution to the stirring energy comes from the interaction between the melt and the slag and thus the deflection of the slag layer. Table 4 shows the time average stirring energy for all cases, considering the fully developed flow.

Mixing time. The effectiveness of homogenisation of heat or any concentration during ladle operations has been evaluated

by adding a scalar concentration for 10 s in a small region in the numerically simulated melt. Sampling the transported concentration in a number of evenly distributed positions through out the whole melt and tracking the signal convergence then gives an estimate of the effectiveness homogenisation of the stirring operations. The values of the mixing time of the seven simulated real plant cases are presented in Table 5. Utilising EMS creates a much shorter mixing time than pure gas stirring [4], the combined stirring can lead to even shorter time comparing to cases operating only with EMS. From the simulations it is indicated that pure gas stirring is by far less efficient than stirring with EMS.

Slag eye area. The slag eye opening has comparable size as the published and other reported ones from the real process data [5]. From the results plotted in Figure 10 we can conclude that the slag eye opening is mainly due to the EMS. The smallest slag eye in the present study is obtained with only gas stirring. It must however also be noted that the slag eye area is sensitive to the slag layer thickness, thus a much thicker initial slag layer may give different results than reported in this investigation.

In Figure 11 the slag eye evolution is shown for the cases operating with downward EMS mode. Initially we see huge fluctuations of the slag eye area, however settling to somewhat more stable values after the initial acceleration time of approximately 25 s. The mean slag eye area at fully developed flow is presented for all the real plant cases in Table 6.Figure 10. Slag eye opening at final simulation time (bottom view), 40s, for

all cases, filled black circles indicate the porous plug.

Figure 11. Change of slag eye area with time. Figure 12. Time history of surface average turbulent kinetic energy at the melt/slag interface.

Case 1 2 3 4 5 6 7

Stirring energy [W/ton] 117 200 205 197 235 223 140

Table 4. Time average stirring energy, considering fully developed flow.

Case 1 2 3 4 5 6 7

Mixing time [s] >500 160 95 190 80 175 85

Table 5. Mixing time for simulated cases.

ABB Value Paper | 7

Melt/slag interface turbulence. Local turbulent mixing is the governing process to enhance the reactions at the melt/slag interface. Gas stirring [6] without EMS induces a high turbulence at the jet impact zone on the interface. This effect remains local and the turbulent kinetic energy averaged on the overall melt/slag interface is small. The EMS, however, induces progressively a high level of distributed turbulence which stabilizes at a certain level when the developed flow is reached. Figure 12 shows the surface average interface turbulence as function of time for the real plant cases operating with downward EMS stirring. Combined stirring with the gas plug at the 180° position seems to enhance the interface turbulence and positioning the plug at the 28° position decreases the level of turbulence. Having the plug positioned close to the stirrer will make the gas plume act as a obstacle reducing not only the liquid velocity as shown in Figure 8 and overall liquid turbulence shown in Table 3 but also the level of turbulence at the melt-slag interface. Table 7 shows the turbulent kinetic energy, k, for fully developed flow for all simulated real plant cases.

Melt cleaning.An important aspect of ladle furnace operations is the removal of impurities in the melt. Traditionally pure gas stirring have been used to facilitate this. Impurities may be trapped by the gas bubbles interface and wake and transported to the slag layer

by the bubbles movement due to buoyancy. The gas plume is then directed straight upwards as in Figure 4 which leaves the main volume of the melt largely unaffected by the cleaning by bubbles. By applying the EMS the gas plume will be altered and the distribution of the bubbles may be controlled by the level of EMS. By the positioning of the gas plug and the direction of the EMS, the combined stirring may be optimised for cleaning operations. Table 8 shows the final bubbles volume fraction in the melt for the real plant simulations.

Figure 13 shows the bubbles volume fraction for all cases with combined stirring, indicating that either positioning the gas plug opposite to the stirrer operating upwards (case 4) or positioning the plug close to a downward EMS (case 7) would give the most favourable case for cleaning. This may however not be fully true since it may not say anything concerning the dispersion of the bubbles, the bubbles residence time or the distance travelled by the bubbles.

Figure 14 shows the bubbles average residence time for the combined cases and Figure 15 shows the bubbles total volume travelled in the melt. Both figures indicate that positioning the gas plug near the downward EMS is the best case for cleaning. This is also indicated in Table 9 showing the total bubbles surface area in the melt. Exposing the gas flow to strong

Figure 13. Time history of bubbles volume fraction in the melt. Figure 14. Bubbles average residence time in melt.

Case 1 2 3 4 5 6 7

Slag eye area [m2] 0.37 1.06 1.66 1.00 1.55 1.13 1.12

Table 6. Time average slag eye area, considering fully developed flow.

Case 1 2 3 4 5 6 7

Turbulent kinetic energy, k , [m2/s2] 0.031 0.125 0.130 0.111 0.133 0.105 0.113

Table 7. Time and surface averaged turbulent kinetic energy at the melt/slag interface, considering fully developed flow.

Case 1 2 3 4 5 6 7

Bubbles volume fraction in melt, [m3/m3] 0.0036 - - 0.0066 0.0044 0.0040 0.0081

Table 8. Final bubbles volume fraction in melt, i.e. at 40s.

8 Fluid Dynamic Features of Combined Gas and Electromagnetic Stirring

shearing liquid flow should also increase the bubbles total area by intense bubbles break-up.

ConclusionsBasic fluid dynamic features of combined EMS and gas stirring have been studied in the present work. A transient multiphase numerical model has been built. Visual investigation of the gas plume shape and flow pattern in a water model has also been done. The gas plume distribution observed in the experimental set-up is in good agreement with the simulated flow cases. The results show that EMGAS has a strong flexibility regarding the flow velocity, gas plume, stirring energy, mixing time, slag layer, etc. This study suggests that there is a big potential to implement EMGAS in the ladle refining process to shorten the production time and/or make cleaner steel.

Desulphurization in the ladle is enhanced by a strong turbulence at the slag/metal interface. The present study shows that this flow situation can be obtained by combining EMS and gas stirring as we can see in the results for cases 5 and 7, presented in Table 7. We have observed that the mixing time is mainly governed by the EMS, however it can be further shortened when combined with gas stirring, as seen in Table 5. For the cleaning stage in the ladle furnace, one can use the EMS to control the shape of the gas plume to make full use of the bubble cleaning effect. EMGAS cases of short mixing time are also favourable for an efficient heating and alloying process.

References − [1] A. Thrum, P.J. Hanley: AIST Steel Technology, 80 (2003),

No. 3, 35-39. − [2] S. A. Morsi, A.J. Alexander: J. Fluid Mech., 55 (1972), No.

2, 193-208. − [3] D. Mazumdar, R. I. L. Guthrie: ISIJ Int., 35 (1995), No. 1,

1-20. − [4] Y. Sundberg: Archiv Eisenhüttenwessen, 55 (1984), No.

10, 463-470. − [5] B. E. Gabrielsson, Steve Lubinski: The 9th Int. Vacuum

Metallurgy Conference, Apr, 11-15, 1988, San Diego, Califor-nia, pp. 1-16.

− [6] L. Jonsson, P. Jönsson: ISIJ Int., 36 (1996), No.9, 1127-1134.

Case 1 2 3 4 5 6 7

Bubbles total surface area in melt, [m2] 5.26 - - 9.67 6.33 5.94 12.01

Table 9. Final bubbles total surface area in melt, i.e. at 40s.

Figure 15. Bubbles total travelled volume in melt.

ABB Value Paper | 9

Contact us

3BU

Sxx

xxxx

A

BB

US

Cre

ativ

e S

ervi

ces

xxxxABB Ltd

Affolternstrasse 44 CH-8050 Zurich, Switzerlandwww.abb.com