ISIJ International, Vol. 51 (2011), No. 2, pp.

193–202

Modelling the Tapping Process in Submerged Arc Furnaces Used

in High Silicon Alloys Production

Mehdi KADKHODABEIGI,1) Halvard TVEIT1) and Stein Tore JOHANSEN2)

1) Department of Materials Science and Engineering, Norwegian University of Science and Technology (NTNU), P.O. Box
7491, Trondheim, Norway. 2) SINTEF Materials and Chemistry, Flow Technology Group, P.O. Box 7465, Trondheim,
Norway.
(Received on May 31, 2010; accepted on October 4, 2010 )

The tapping process is an important step in the silicon and ferrosilicon production process. Tapping is sim-
ply how to transfer the melt from the furnace into the ladle. The tapping process has always been a chal-
lenging industrial operation where the metal flow rate is influenced by many different phenomena. In this
present work we present a model for the tapping of the submerged arc furnaces. Using the model the ef-
fects of furnace crater pressure, metal height and permeability of the different internal zones have been
studied. The model is based on computational fluid dynamics (CFD) where the geometry is taken from in-
dustrial furnace geometry. The internal zones with individual permeabilities are defined based on informa-
tion from furnace excavations. From the model we extract new information about the process and explain
phenomena which control the tapping flow rate. It was found a very good agreement between the model
predictions and industrial measurements.
KEY WORDS: submerged arc furnace; tapping process; furnace crater pressure; bed permeability; compu-
tational fluid dynamics (CFD).

In the present work the main issues which impact the tap-
1. Introduction
ping process are considered to be the furnace crater pres-
High silicon alloys are industrially produced in the sub- sure, the metal height over the furnace bottom and the per-
merged arc furnaces. The furnace casing is made of sheet meability of the packed beds in different zones inside the
steel, the lower part lined with hard blocks of strongly cal- furnace.
cined carbon and the upper part with firebrick. The furnace The full 3D flow model of the two-phase gas–liquid flow
contains three electrodes which are situated at the corners inside the furnace is using full scale industrial furnace
of a regular triangle and they are submerged into the charge geometry. The permeabilities of the zones inside the fur-
materials. Conversion of the electrical energy into heat in nace were tuned based on the results of furnace excava-
the bulk of the charge materials results in carbothermic re- tions, data available from the literature and operational in-
duction of the ore. As the smelting process proceeds the formation. In order to validate the results of the model sev-
molten ferrosilicon together with process gases mainly eral industrial tests have been done and compared to the
composed of SiO and CO are produced. Ferrosilicon melt is model results.
accumulated over the furnace bottom and process gases
move through charge particles towards the furnace top
2. The Issues Affecting the Tapping Process
where they are sucked into the furnace off-gas system. The
melt is after tapping from the furnace taphole refined or al- Industrially it is very important to control the tapping
loyed before the final casting. process, in particular represented by the metal flow rate and
If tapping due to any reason fails the furnace process tapping time. The underlying physical phenomena affecting
must be stopped and the tapping must be completed before the tapping flow rate are believe to be the furnace crater
the process can proceed.2) Tapping flow rate and hence tap- pressure, metal height and the permeability of the packed
ping time are the most important operational parameters. In beds inside the furnace.
practice the melt inside the furnace is never tapped com-
pletely. It means that some melt is always remaining inside 2.1. Furnace Crater Pressure
the furnace when the taphole is closed. The tapping flow Inside the furnace different chemical reactions take place
rate and tapping time are by large controlled by the condi- and as the result of the reactions process gases mainly com-
tions inside the furnace as well as different operational de- posed of SiO and CO are generated. The reactions can be
cisions.1) Therefore identification and investigation of the formally treated as a combination of gross three reactions,
parameters affecting the tapping process is of significant in- Eqs. (1) and (3).1)
dustrial importance.

193 © 2011 ISIJ

 ISIJ International, Vol. 51 (2011), No. 2

SiO 2 (s, l) ⫹ 2C(s) ⫽ Si (s) ⫹ 2CO (g) ..............(1) from industrial measurements show that the furnace crater
pressure has a dynamic nature. Several parameters affect
SiO 2 (s, l) ⫹ C(s) ⫽ SiO (g) ⫹ CO (g) ..............(2) the furnace crater pressure.1,2) The dominating parameters
Fe 2 O 3 (s, l) ⫹ 3C(s) ⫽ Fe (l) ⫹ 3CO(g) ............(3) are the type of charge materials, rate of chemical reaction in
the crater zone, electric load, electrode heights, reduced
The process gases produced by these reactions, flow to- permeability of the charge materials due to condensation of
wards the furnace top by passing through semi-porous re- process gases in the upper part of charge materials, while
gions of charge materials. The results of the furnace exca- secondary effects, such as softening and melting of charge
vations and different scientific investigations regarding sili- particles in the regions near by the crater zone, affects the
con and ferrosilicon furnaces3–5) have shown that in the re- local permeabilities. Avalanches, released naturally or by
gion around and beneath the electrode tips, very close to the stoking cause significant variations in the crater pres-
furnace bottom, a cavity is formed. Formation of these cav- sure.6) Figure 2 shows the time variation of the furnace
ities is mainly due to the fact that conversion of electric en- crater pressure and its relation with the furnace electric load
ergy to heat happens inside these cavities through electric in a silicon producing furnace.6)
arcs. The size of these cavities increases as the production
process in the furnace proceeds.1) During the process these 2.2. Permeability of the Packed Beds in the Furnace
cavities are filled with the process gases. The production of Excavations of the submerged arc furnaces3–5) show that
process gases together with the reduced permeability of inside the furnaces there are different zones with different
charge materials around theses cavities leads to creation of physical properties. Due to different chemical reactions, un-
very high pressures in the furnace crater zone. A simplified even heat distribution, softening and melting of charge par-
representation of the conditions inside the furnace is shown ticles and condensation of process gases in the bulk of
in Fig. 1. charge materials, formation of these zones is inevitable.
The gas pressure inside the furnace crater zone has been During tapping the melt is drained from a porous bed
measured and reported in some previous works.6,7) Results which lies on the furnace bottom. The bed is mainly com-
posed of the unreacted charge materials and solid SiC
which are produced by the furnace process.1)
The physical properties of each zone in the furnace, such
as porosity and the particles size, are scaled to give the ex-
pected flow resistance for each zone of the charge. The
porosity distribution in the charge material is influenced by
proper chemical reactions happening in each zone. En-
dothermic and exothermic reactions determine the heat dis-
tribution and hence the porosity distribution in the charge
packed bed. Reduced permeability in the charge materials
creates higher resistance for the flow of process gases and
hence increases the crater pressure. On the other hand, the
permeability at the furnace bottom bed has a crucial effect
on the melt flow rate from the taphole. Both these phenom-
ena underline how the permeability of the packed beds can
affect the tapping process.

2.3. Metal Height in the Furnace

The metal height in the furnace is expected to be a func-
Fig. 1. The symbolic representation of the conditions inside a
silicon/ferrosilicon furnace.1) tion of the melt production rate and the porosity of the
packed bed at the furnace bottom. According to the
Bernoulli’s equation for fluid flow the metal height will in-
fluence the tapping flow rate from any vessel, including our
furnace.
Knowledge of the metal height in the furnace is always a
challenging issue. If the level of the melt inside the furnace
becomes too high the liquid silicon comes in contact with
the molten quartz according to Eq. (4).2)
Si (l) ⫹ SiO 2 (l) ⫽ 2SiO(g) ......................(4)
This reaction takes place at temperatures above 1 800°C
and is strongly endothermic. Reaction (4) may disturb the
process in three ways.2) Firstly, the reaction will normally
increase production of silica-dust and hence lower the sili-
Fig. 2. Time variation of the furnace crater pressure and its rela- con yield. Secondly, this will have an impact on the carbon
tion with the furnace electric load in a silicon producing balance. An increase in the SiO-loss from the furnace will
furnace.6) reduce the amount of carbon needed by the process.

© 2011 ISIJ 194

 ISIJ International, Vol. 51 (2011), No. 2

Thirdly, the huge amount of energy needed for Reaction (4) The model is a generalization of Darcy’s Law, commonly
may reduce the process temperature in the hot zone and in- used for flows in porous media, and of the Reynolds-aver-
crease the need for SiO circulation in the furnace, which is aged Navier–Stokes equations, with an eddy viscosity ac-
very unfavorable for the total process performance. counting for the turbulent effects. The last term in Eq. (6)
Therefore it is unfavorable to have a very high metal represents the resistance to flow in porous media. Based on
height in the furnace before start of tapping. As a result the the well-known Ergun’s equation,10) the resistance force for
optimal time intervals between the taps are affected by the flow through a bed of particles is given by:
metal height inside the furnace.  150 μ (1⫺ γ )2  1.75ρ (1⫺ γ )  
R⫽ v⫹ v v .....(11)
Dp2 γ3 Dp γ3
3. Development of a Model for Tapping Process
The defined resistance in the Ergun’s equation for each
In order to model the flow of gas and liquids out from the component can be written as:
heart of the furnace we need a multidimensional multiphase
model that can handle the flow of gas and liquid though ⎛μ  1  ⎞
Ri ⫽⫺⎜ ⋅ vi ⫹ C2 ρ v vi ⎟ ................(12)

porous materials. In addition we need to keep track of the ⎝α 2 ⎠
sharp interface between gas and liquid inside the furnace.
The modeling choice was therefore to apply volume aver- Dp2 γ2
aged Navier–Stokes equations in which the flow resistance α⫽ .........................(13)
150 (1⫺ γ )3
due to the more or less transformed charge materials can be
expressed by general friction laws. When it comes to the 3.5 (1⫺ γ )3
gas liquid interface it must be tracked as sharply as possi- C2 ⫽ .........................(14)
Dp γ3
ble. As the needed models concepts already were available
in the commercial Fluent code we used Fluent’s Volume- In order to model the turbulent flow of the gas and liquid
Of-Fluid (VOF) model to keep track of the gas–liquid inter- phases available in the furnace, a modified version of the
face. k–e model for fluid flow in porous beds8) is applied. In this
In order to simplify the model we have limited the analy- turbulence model an extra source term due to turbulence
ses to isothermal but high temperature conditions. production due to large solid particles is added to both the
kinetic energy and the dissipation rate model equations.
3.1. Model Governing Equations The turbulent viscosity is then determined from these mod-
As the gas–liquid interface is tracked by the VOF model ified turbulent kinetic energy and dissipation rates. Accord-
we assume that gas and liquid have locally identical veloci- ingly, the model allows unified treatments for the flow of
ties. Hence, mass and momentum conservation is expressed process gas and silicon melt over the entire furnace volume.
using a common fluid velocity and using mixture proper-
ties. Accordingly, the conservation equations for mass and 3.2. Geometry and Inside Conditions of the Furnace
momentum in an isotropic porous medium is written in co- Geometry of the furnace used in the modelling has been
ordinate free tensor notation as: chosen based on an industrial ferrosilicon furnace. The
schematic of the furnace geometry and a part of the compu-
∂γρ
⫹ ∇ ⋅ (γρu ) ⫽ 0 ..........................(5) tational grid used in the modelling process are shown in

∂t Figs. 3(a) and 3(b).
∂γ u The furnace diameter is about 8 m, the furnace height is

⫹ ∇ ⋅ (γρuu ) ⫺ ∇ ⋅ [ μeff γ (∇u ⫹ (∇u )T )]
  
∂t 3 m and the taphole diameter is 10 cm. Inside the model dif-
ferent zones have been defined. The physical properties of
⎛ 2 ⎞
⫽ γ ∇ ⎜ p ⫹ ρk ⎟ ⫹ γρ gβ (T ⫺ Tref ) ⫺ γ R ......(6) each zone have been selected based on the results of fur-

⎝ 3 ⎠ nace excavation and the data available in the literature.
Where g is the fluid fraction and r is the fluid bulk density, Over the furnace bottom, where the melt is accumulated,
with k representing gas and liquid respectively. three different zones have been defined. Difference in phys-
ical properties of these zones is directly related to the un-
ρ⫽ ∑γ
k
k
ρk ∑γ
k
k .........................(7) even temperature distribution in the existent packed bed.
The flow resistance is designed to increase as we move out
from the furnace centre and approach the taphole. High
μ⫽ ∑γ
k
k
μk ∑γk
k
.........................(8) flow resistance is created partially by a high solids fraction.
Hence, the volume fraction of the melt is highest in the fur-
The scalar equation for the propagation of the liquid frac- nace centre and decreases by increasing the distance from
tion has similar form as Eq. (5). the furnace centre. The porosity both in the upper and lower
The effective and turbulent viscosities are defined as fol- parts of the furnace ranges between 10% up to 60% de-
lows: pending on the position of the packed beds. While studying
μeff ⫽ μ ⫹ μ t ................................(9) the effect of metal height and furnace crater pressure on the
tapping flow rate the porosity and permeability distribution
k2
μ t ⫽ ρCμ .............................(10) in the lower part of the furnace is the same in all simula-
ε tions. The 2D view of the defined zones in the model is pre-
sented in Fig. 4.

195 © 2011 ISIJ

 ISIJ International, Vol. 51 (2011), No. 2

at atmospheric pressure. The fluid properties are based on

the furnace operating temperature, which is around
1 800°C, while local temperatures in the crater zone may
reach levels above 2 000°C. Over the internal surface of the
furnace walls we apply a no-slip condition for the velocity
of both liquid and gas.
In order to simulate the generation of process gases due
to chemical reactions in the crater zone, a mass source term
for gas production was defined inside the crater zone. The
production rate of the melt which is around 70 ton/d for the
considered ferrosilicon furnace in this study is introduced
via defining a mass source in the metal zone of the model
and it is fixed in all case studies. The numerical value for
the mass source terms of process gas and melt was based
on information from industrial operation.

4. Results
Simulations have been performed for a large number of
operational conditions. This allows the investigation of how
furnace crater pressure, metal height and the permeability
impact the tapping flow rate, the total tapping weight and
the tapping time. The results from the model are presented
in different sections, according to the phenomena under ex-
amination. In addition to the model results, for sake of
comparison and validation, the results of industrial meas-
urements are also presented.
Fig. 3. 3D schematic of a ferrosilicon furnace (a) and a slice of
the computational grid used in the modelling process (b). 4.1. Investigation of the Effect of Crater Pressure
Measurements of the crater pressure in the silicon and
ferrosilicon producing furnaces have been reported in the
literature.6,7) The crater pressure can be measured by mak-
ing a hole inside and parallel to the electrodes. As the hole
extends all the way to the electrode tip pressure contact
with the crate zone is obtained. In this manner the crater
pressure can be measured at the top of the electrode where
temperature is low. The measured furnace crater pressure
shows a dynamic behavior due to multiple interacting
mechanisms. However, still we can define representative av-
erage values. In the furnace under consideration in this
study the reported crater pressure ranges from 30 mbar up
to 200 mbar. The results of the model for the tapping flow
rate in the case where the initial metal height inside the fur-
nace is 10 cm is presented in Fig. 5.
Results show that the tapping flow rate is initially quite
high, depending on the crater pressure. After a while the
Fig. 4. Different zones defined inside the submerged arc fur-
tapping rate decreases significantly and reaches to a level
naces for ferrosilicon production.
which is almost the same in all cases. The results in Fig. 5
show that the increased crater pressure leads to increased
The liquid phase in the model is a special ferrosilicon tapping flow rate and decreased waiting time for the drop in
alloy with the density of 4 300 kg/m3. The gas phase is a tapping flow.
mixture of process gases made of SiO and CO gas which in The comparison between the result of the CFD model
the furnace operating temperature the gas mixture density and the industrial measurements for both the tapping flow
is 0.23 kg/m3. rate and the total tapping weight versus tapping time is pre-
sented in Fig. 6.
3.3. Numerical Method and Boundary Conditions The drop in flow rate drop during tapping is an interest-
In order to solve the set of governing model equations ing phenomenon which is a result of the melt flow pattern
the commercial CFD software Fluent 6.3.26 was used. The inside the furnace, caused by the crater pressure.9) When
furnace operates at atmospheric pressure condition. There- the tapping flow rate drops gas starts to emerge from the
fore the boundary conditions at the furnace top and at the taphole. After this initiation of two phase flow from the tap
taphole outlet were defined as constant pressure boundaries hole, the gas flow rate increases as the tapping process pro-

© 2011 ISIJ 196

 ISIJ International, Vol. 51 (2011), No. 2

Fig. 5. The tapping flow rate at different crater pressures inside Fig. 7. Comparison between the results of the model and indus-
the furnace, initial metal height is 10 cm. trial measurements for the total tapping weight, the initial
metal height in the model is 10 cm.

Fig. 8. Tapping flow rate only due to hydrostatic pressure created

by the metal height, initial metal height is 12 cm.

duced in order to keep a constant maximum metal height.

In order to show that the increased flow rate at the start
of tapping and significant flow rate drop are due to exis-
tence of the high pressure crater zone a simple study was
done. In this case the initial metal height inside the furnace
was considered to be 12 cm but there was no additional
pressure due to process gases released from the crater zone.
In fact, the only parameter, driving the flow, was the metal
height. The result of this simulation is shown in Fig. 8.
Fig. 6. Comparison between industrial measurements and results
This result contains two important points. The first point
of the model for tapping flow rate (a) and total tapping is that the tapping flow rate is very low, even with higher
weight (b) from a ferrosilicon producing furnace, the ini- initial metal height than the cases shown in Fig. 5. In fact,
tial metal height is 8 cm and the crater pressure is to have such high tapping flow rates without the crater pres-
80 mbar. sure the initial metal height must be up to 60 cm. Such high
metal height is neither acceptable nor possible from the in-
ceeds. Still the gas mass flow is very small compared to the dustrial point of view. The second point is that when only
melt flow. the hydrostatic pressure due to the metal height drives the
Figure 7 shows the comparison of the model results for flow, the flow rate drop observed in the industrial measure-
the total tapping weight with initial metal height of 10 cm ments can not be explained. These two points substantiate
versus the industrial measurements. The points in the graph that the crater pressure due to limited flow permeability is
show the total tapping weight before tapping flow rate the cause of a significant change in tapping rate, appearing
drops. after a certain mass of metal is tapped.
As it can be seen from Fig. 7 that a high crater pressure
results in less metal tapped, even if the metal flow rate is 4.2. Investigation of the Effect of Metal Height
higher and tapping time shorter. The consequence of this is In order to investigate the effect of metal height on the
that for an operation of the furnace at higher crater pres- tapping parameters, different metal heights have been stud-
sures the time intervals between taps will have to be re- ied. The range considered for the metal height starts from

197 © 2011 ISIJ

 ISIJ International, Vol. 51 (2011), No. 2

very low level around 4.5 cm up to 12 cm, the latter being

higher than the taphole level. The effect of metal height on 4.3. Investigation of the Combined Effect of Crater
the tapping flow rate is for a crater pressure of 90 mbar pre- Pressure and Metal Height
sented in Fig. 9. Industrial measurements prove that the furnace crater
It is observed that the increased metal height leads to in- pressure has a dynamic behavior during the smelting
creased tapping flow rate. In addition we see an increase in process. In the industrial operation the metal production
the tapping time before the flow rate drops. It is obvious rate is not constant and different phenomena can change the
that the effect of metal height on the tapping flow rate is not production rate and hence the metal height in the furnace
as significant as the effect of furnace crater pressure is. varies by time. Therefore investigation of the most impor-
The most important result from the investigation of the tant parameters over a larger span of values is necessary. In
metal height effect is that in all different range of metal this research we have investigated the furnace crater pres-
heights, the tapping rates before the metal flow rate drops sure range from 45 to 180 mbar and for the metal height
are virtually the same. It means that the metal height is not range from 4.5 to 12 cm. The tapping time is also a key pa-
the factor which determines at which tapping rate the metal rameter in modeling the tapping process and therefore we
flow rate will drop. In other words, it is the furnace crater explore how tapping time is affected by crater pressure and
pressure which determines at which tapping rate the metal metal height. Note that from now on we define the tapping
flow rate drops. The effect of metal height is to determine time as the time until the tapping flow rate suddenly drops.
when the tapping flow rate drops. It is obvious that high The reason for this definitions is that after gas break
metal height and corresponding large metal mass will lead through in the taphole the tapping rate drops down to very
to larger tapping time. Furthermore, we see that for a given low level which is almost the same in all cases, and has a
crater pressure the metal mass tapped before reduction in low metal flow rate that is in good agreement with the in-
tapping rate is a clear indication of the total amount of dustrial measurements.
metal in the furnace. Figure 11 shows the effect of furnace crater pressure on
Figure 10 shows the same results as Fig. 9, but now as the average tapping flow rate and the total tapping weight
the total tapped weight. It is seen how the total tapping for different metal heights inside the furnace. It is observed
weight from the furnace increases with increased metal that furnace crater pressure has a positive effect on the tap-
height. ping flow rate for different metal heights but the total tap-
ping weight from the furnace is lower in the cases with
higher crater pressure.
The simulation results can alternatively be presented as

Fig. 9. The effect of metal height on the tapping flow rate. Crater
pressure is 90 mbar.

Fig. 11. The effect of crater pressure on the average tapping rate
Fig. 10. The effect of metal height on the total tapped mass. The (a) and the total tapping weight (b) for different metal
crater pressure is 90 mbar. heights inside the furnace.

© 2011 ISIJ 198

 ISIJ International, Vol. 51 (2011), No. 2

Fig. 12. The effect of metal height on the average tapping rate Fig. 13. The relation between the crater pressure, the metal
(a) and the total tapping weight (b) for different furnace height and the average tapping flow rate (a) and the rela-
crater pressures. tion between the crater pressure, the metal height and
the tapped metal mass (b).
the average tapping flow rate or the total tapping weight,
both versus metal height (see Fig. 12).
As it is seen from Fig. 12(a) increase in the metal height
leads to increased tapping flow rate for all crater pressures.
Furthermore, and as expected, the total mass of tapped
metal increases as the metal height and hence metal volume
in the furnace is increased, as observed from Fig. 12(b).
As tapping time expresses the time we can tap before
gassing from the taphole starts, tapping time is the most im-
portant operational parameter. The effects of metal height
and furnace crater pressure on the tapping flow rate and the
total tapping weight as a function of time is presented in
Fig. 13.
Figure 14 represents the percentage of the melt which is
tapped from the furnace inside as a function of tapping time
for different crater pressures and metal heights. As it can be Fig. 14. Percentage of the melt which is tapped from the furnace
seen when the crater pressure is lower the amount of the inside as a function of tapping time for different crater
pressures and metal heights.
metal tapped from the furnace is higher and the tapping
time is longer. The general picture is that short tapping
times correspond to low metal height and high crater pres- flow rate decrease leads to a very interesting and valuable
sures. However, in these cases with the shortest tapping result about prediction of the metal height inside the fur-
times the results are telling that more than 90% of the nace. Figure 15 shows both the results of the CFD model
metal remains inside the furnace when gassing from the and industrial measurements. It is observed that the range
taphole is initiated. of industrial data falls between the metal height of 6 cm and
9 cm predicted by the model. A point that should be noticed
4.4. Prediction of the Crater Pressure and the Metal is that the metal height considered in the modeling and the
Height in the Furnace Using the Results of CFD metal volume are not linearly related as the furnace bottom
Model contains a packed bed of solid particles which displaces the
Comparison of the results of the model with the indus- melt and has a given porosity distribution.
trial measurements for the total tapping weight before the The same method can be applied for the furnace crater

199 © 2011 ISIJ

 ISIJ International, Vol. 51 (2011), No. 2

Fig. 15. Comparison between the model results and industrial

measurements for the total tapping weight versus tap-
ping time, for different metal heights and crater pres-
sures.

Fig. 17. The effect of bed permeability on the tapping flow rate
and the total tapping weight, the initial metal height is
12 cm.

erties on the tapping process, two different beds with differ-

Fig. 16. Comparison between the model results and industrial
measurements for the average tapping rate versus tap- ent permeability (and hence resistance) were considered at
ping time, for different metal heights and crater pres- the furnace bottom. The permeability difference was con-
sures. sidered to be only due to different particles size inside the
beds and not due to porosity change. The first bed is the
pressure. Figure 16 shows the comparison between the main bed of the model and in the second bed the permeabil-
model results and the industrial measurements. ity has been reduced down to 40% of the base case. The
Comparison of the industrial measurements with the comparison of the tapping parameters between the low and
model results support that variations in tapping rate ob- the high permeable beds for the crater pressures ranging be-
served industrially may be caused by the dynamic behavior tween 90 mbar up to 180 mbar is presented in Fig. 17. The
of the furnace crater pressure. From Fig. 15 we see that the initial metal height in all these cases is constant and fixed at
industrially observed variations in tapping rate can be ex- 12 cm.
plained by large with industrially observed variations in As it can be seen from Fig. 17 the tapping flow rate de-
crater pressures while the initial metal level before tapping creases by decreasing the bed permeability. The tapping
starts is around 8 cm. From Fig. 16 we can conclude that time is also influenced by the bed permeability although in
metal height variations cannot explain the industrially ob- the cases with higher crater pressure the difference in the
server variations in tapping rate and tapping times. Based tapping times is very small. It is clear that the effect of bed
on these results it is likely that the model will be able to permeability on the total tapping weight is insignificant as
give indications of the furnace crater pressure by having the crater pressure increases.
just the information about the total tapping weight. Figure 18 shows that how the bed permeability affects
the average tapping rate from the furnace and the total tap-
4.5. Investigation of the Effect of Bed Permeability ping weight before the flow rate drops in different range of
The packed bed at the furnace bottom has two important crater pressures. It is obvious that the higher the bed perme-
effects. Firstly it influences the metal height for a given ability the higher the average tapping flow rate.
metal volume. Secondly the packed bed creates a resistance Investigation of the effect of bed permeability on the av-
against the flow of melt towards the taphole and leads to re- erage tapping flow rate and the total tapping weight versus
duced tapping flow rate from the furnace. In fact the melt tapping time for different range of furnace crater pressures
should drain through this packed bed during tapping has been presented in Fig. 19.
process. As it can be seen from this figure as the crater pressure
In order to investigate the effect of the backed bed prop- decreases both the tapping time difference and the propor-

© 2011 ISIJ 200

 ISIJ International, Vol. 51 (2011), No. 2

Fig. 18. The effect of bed permeability on the average tapping Fig. 19. The effect of bed permeability on the average tapping
flow rate (a) and the total tapping weight (b) for differ- flow rate (a) and the total tapping weight (b) versus tap-
ent furnace crater pressures, the initial metal height is ping time for different furnace crater pressures, the ini-
12 cm. tial metal height is 12 cm.

tional difference of the average tapping flow rate increases. total tapping weight before flow rate drops. The effect of
The results show that the tapping flow rate is higher in the metal height on the tapping flow rate is not as significant as
cases when the bed permeability is higher. It is also ob- the effect of furnace crater pressure. A trivial result is that
served that both the tapping time difference and the propor- the metal height, directly related to the total metal volume,
tional difference of the total weight of the melt tapped from is the key factor which determines when tapping flow rate
the furnace increases as the furnace crater pressure de- drops.
creases. The total tapping weight is higher in the case with The change of permeability in different zones inside the
more permeable bed. furnace influences the process gas flow resistance and pres-
sure drop as the gas flows from the crater to the top of the
charge. A low permeability in the charge results in a high
5. Conclusions
pressure inside the crater zone. The permeability of the
A comprehensive model for the tapping process in the packed bed formed over the furnace bottom, particularly
submerged arc furnaces used for production of high silicon near the taphole zone, affects the tapping flow rate. Re-
alloys has been developed. Different important issues which duced permeability of the bottom packed bed causes re-
can affect the tapping process have been investigated. The duced tapping flow rate and hence longer tapping time. In
most important parameters affecting the tapping are furnace addition low permeability result in reduces yield of the
crater pressure, metal height inside the furnace and perme- tapped metal.
ability of different packed beds inside the furnace. The results of the model are in very good agreements
The model results show that the furnace crater pressure with the industrial measurements and the model is able to
affects the tapping process through influencing the tapping explain the variations in industrial tapping rates as a result
flow rate and the fraction of the metal that can be tapped of variations in crater pressure.
before gassing from the taphole is initiated. The crater pres-
sure is the factor which determines the tapping flow rate by Nomenclature
the start of tapping as well as the tapping rate just before r : Fluid density (kg/m3)
flow rate drops. Higher furnace crater pressure leads to g : Porosity of the bed
higher tapping flow rate from the furnace at the start of tap- u, v : Fluid velocity (m/s)
ping process and at the same time it causes the faster tap- m : Dynamic viscosity (Pa · s)
ping flow rate drop. m t : Turbulent viscosity (Pa · s)
The metal height affects the tapping flow rate and the p : Pressure (Pa)

201 © 2011 ISIJ

 ISIJ International, Vol. 51 (2011), No. 2

k: Turbulent kinetic energy (m2/s2) 2) H. Tveit, T. Halland, K. I. Landrø, S. T. Johansen and B. Ravary:
e: Turbulent dissipation rate (m2/s3) Proc. Silicon for the Chemical Industry, Loen, Norway, (2002).
3) A. Schei: Proc. Tidsskrift Kjemi Bergvesen, Metallurgi 27, Trond-
g: Gravity acceleration (9.81 m/s2) heim, Norway, (1967).
T: Temperature (K) 4) Y. Otani, M. Saito, K. Usui and N. Chino: Proc. 6th Int. Cong. of
b: Thermal coefficient of volumetric expansion (K⫺1) Electro Heat, Brighton, England, (1968).
Dp : Particle diameter (m) 5) E. H. Myrhaug: PhD Thesis, NTNU, Trondheim, Norway, (2003).
Cm : Turbulent model constant 6) S. T. Johansen, H. Tveit, S. Grådahl, A. Valderhaug and J. Byberg:
Proc. 8th INFACON Conf., Beijing, China, (1998).
a: Permeability of the packed bed 7) H. T. Ingason, J. Halfdanarson and J. A. Bakken: Proc. 7th INFA-
C2 : Inertial resistance CON Conf., Trondheim, Norway, (1995).
k: Liquid and gas phase index 8) A. Nakayama and F. Kuwahara: J. Fluids. Eng., 121 (1999), 427.
9) M. Kadkhodabeigi, H. Tveit and S. T. Johansen: J. Prog. Comput.
REFERENCES Fluid Dyn., 10 (2010), No. 5–6, 374.
10) S. Ergun: J. Chem. Eng. Prog., 48 (1952), No. 2, 89.
1) A. Schei, J. Tuset and H. Tveit: Tapir Forlag, (1998).

© 2011 ISIJ 202