Improved Slag Formation in the Blast Furnace by Co-injection of Basic

Fluxes, Together with Pulverized Coal, through the Tuyeres

Jan-Olov Wikström1, Peter Sikström1, Lena Sundqvist2 and Guangqing Zuo3

1. MEFOS, Luleå, Sweden

2. SSAB Tunnplåt AB, Luleå, Sweden
3. Luleå Technical University, Luleå Sweden

ABSTRACT Excellent BF performance was achieved during the

When operating a blast furnace with 100% pellets as
ferrous burden, basic fluxes are normally charged
together with the pellets. Even if the pellets are
fluxed, there is normally a need for basic fluxes, to
neutralize the acid slag generated in the tuyere area,
when coke and coal are combusted. The fluxed
pellet itself has an early melting slag, but when it is
charged together with basic fluxes, excessive basic
slag can form in the cohesive zone where the top-
charged fluxes partly dissolve into the primary pellet
slag. The newly formed slag can have a very high
melting point and a low fluidity, which can even
result in re-solidification of the slag. As a result, the
permeability of the cohesive zone and of the bosh
region is reduced. Injection of basic fluxes together
with pulverized coal, instead of charging them
through the top, has been recognized as a possible
solution to the problem.
trial. The operation was very smooth and stable.
The slag volume was around 100 kg/t-HM and the
reductant rate was lowered by 11 kg/t-HM,
compared to the reference period. A significant
stabilisation and reduction in hot metal silicon
content was reached. Sulphur distribution, as well
as output of K2O by the slag, was improved.
Quenching and excavation of the BF showed that the
formation of a birds nest can be avoided when the
acidic slag, normally generated in the raceway, is
fluxed with BOF slag. Injection of fluxes opens up
new opportunities to further develop the blast
furnace process.
Successful trials with co-injection of BF flue dust
and pulverized coal in the experimental BF is also
reported. Industrial trials have started at SSAB
Luleå Works No 3 BF.

Fundamental research at laboratory scale at Luleå

Technical University has shown that mixing of
pulverized fluxes with ash from coke and coal
reduces the melting point of the tuyere slag by more
than 200°C. At the same time slag viscosity is
highly improved.
During a two-week period, co-injection of BOF slag
and pulverized coal was tested in LKAB’s
experimental blast furnace at MEFOS. The furnace
has a working volume of 8.2 m3 and a hearth
diameter of 1.2 m, and is equipped with three
tuyeres. The furnace is operated in campaigns of
four to ten weeks, at a productivity around
3.5 t/m3day. Each campaign is normally finalized by
quenching and dissection. The experimental BF has
earlier proved to be a very sensitive tool for
detecting differences in the properties of the ferrous
burden.
During the trial the ferrous burden consisted of
100% pellets. The objective was to improve slag
formation by fluxing the tuyere slag and to avoid
excessive basicity in the lower shaft. All fluxes, eg
BOF slag, were injected through the tuyeres.
INTRODUCTION
At SSAB Luleå Works, blast furnace operation with
100% pellets, as ferrous burden, has been the
practice since 19782. From 1982 olivine pellets
from LKAB have been the only ferrous raw
material, except for some short periods for testing
alternative new pellet types. A quite small amount
of basic fluxes is charged together with the pellets.
When using self-fluxed pellets instead of olivine
pellets of basicity ~1, the fluxes charged are used for
neutralisation of coke and coal ash which are mainly
released in the combustion zone.
According to laboratory standard tests, a type of
self-fluxed pellets with a high Fe-content has
excellent metallurgical properties, such as high
reducibility and a narrow softening and melting
temperature interval. However, industrial trials in
the early nineties using this type of fluxed pellets
have not revealed the expected improvements in BF
performance. One reason for this has been
theoretically attributed to a problematic slag
formation process inside the furnace. The fluxed

International BF Lower Zone Symposium, Wollongong 2002 - Preprint

18 - 1
 J.-O. Wikström, P. Sikström, L. Sundqvist and G. Zuo

pellet itself has an early melting slag, but when

charged together with basic fluxes, an excessively
basic slag can form in the cohesive zone where the
Primary slag Primary slag
basic fluxes partly dissolve into the primary slag
from the pellets. The newly formed slag can have a
very high melting point and a low fluidity, which
can even result in re-solidification of the slag. The
melting properties of the slag become even poorer
when it is further reduced and the FeO-content
decreases. As a result, the permeability of the
cohesive zone and of the bosh region is decreased, Bosh slag Bosh slag
causing irregular blast furnace operation. A
proposal to inject basic fluxes via the tuyeres instead Tuyere slag Tuyere slag
of charging from the top has been made to enhance
the slag formation process at low slag volumes1.
Injection of different types of material as basic
fluxes, sinter fines and iron oxide has been tested to
achieve lower Si and/or S content of the hot metal10, Final slag Final slag
11, 12, 13, 14, 15
. Injection of BF flue dust has also been
reported16. The results from these studies show that
basic fluxes, as well as iron oxide, have significant
effect on the Si content of hot metal. The effect is
B2= CaO/SiO2 B2= CaO/SiO2
higher when using sinter fines containing basic
constituents compared to when pure iron oxide is No flux injection BOF slag injection
used. Basic components also decrease the S content
of hot metal. The content of iron oxide and basic Fig. 1 Basicity of slags at different locations in
fluxes in injected BF flue dust can be expected to the blast furnace.
have a similar effect. In addition. the reported tests
of injection of BF flue dust indicate improved The primary slag is formed from the gangue content
combustion as well. in the pellets. The bosh slag is considered to be
formed by the primary slag, all additives and fluxes
As mentioned above, when self-fluxed pellets are and some ash from the coke that is consumed in the
used, the top-charged fluxes that are supposed to shaft by solution loss reaction. The ash released in
neutralize the acid slag from coke and coal ash, the raceway, when coke and coal are combusted,
being generated in the raceway zone, dissolves in the forms the tuyere slag, which also contains the flux
bosh slag and increases its basicity, as well as constituents , when they are injected. The bosh slag
melting point. Locally the slag basicity can be and tuyere slag are mixed and after some additional
extremely high because of an uneven distribution of reduction the final slag is formed. As can be seen,
the fluxes charged into the blast furnace. If the the excessive basicity of the bosh slag and the
fluxes, instead of being charged through the top, are formation of a very acid tuyere slag can be avoided,
injected through the tuyeres, the excessive basicity by the injection of fluxes.
of the bosh slag is avoided and the acid slag
generated in the raceway is fluxed. In Fig. 1 the
basicity of primary slag, bosh slag, tuyere slag and FUNDAMENTAL STUDIES
final slag are compared for self fluxed pellets. In
both cases shown, the basic flux used is BOF-slag, Selection of basic slag formation injectant
being charged through the BF top (left hand side) or
injected through the tuyeres (right hand side)1. Possible fluxes to be injected into the blast furnace
were considered to be burnt lime, burnt dolomite or
BOF slag. Limestone or dolomite have a strong
cooling effect on the raceway flame temperature,
which would limit the injection rate of pulverized
coal too much. The effect on tuyere slag formation
from addition of burnt lime, burnt dolomite and
BOF-slag was studied, by using a high temperature
microscope with a maximum working temperature
of 1640°C3-9, see Fig. 2.

International BF Lower Zone Symposium, Wollongong 2002 - Preprint

18 - 2
 J.-O. Wikström, P. Sikström, L. Sundqvist and G. Zuo

Thermometer

Tube furnace Magnifier Frosted

glass
Lens Grid
Bulb

Sample Mirror Camera

Ocular
Thermocouple Sample holder
Fig. 2 Schematic drawing of high temperature microscope.

In this study9 the tuyere slag is considered to be
formed from the mixture of ash released from coke
and coal combusted in front of the tuyeres, together
with the fluxes being injected. The specimen tested
was made from a mixture of coke ash, coal ash and
fluxes from SSAB Tunnplåt in Luleå. Coke and
coal ash are obtained by combustion of coke and
coal in a furnace slowly heated to 800°C. The ratio
of coke ash, coal ash and fluxes used in the
specimen varies depending on coke rate, and the
amount of coal and fluxes injected. Table 1 shows
the amounts of fluxes to be injected corresponding
to different injection levels. To make a specimen
representing the tuyere slag at a certain level of flux
injection, corresponding amounts of ash mixture and
slag former are ground in a mortar.
A small specimen, about 2 mm square and 3 mm
high, is used for each test. The sample is placed on
an alumina plate on the sample holder. The sample
is gradually heated in a furnace and a heating rate of
10°C/min is used at temperatures exceeding 700°C.
Sketches of the sample are drawn to show the
changes in the shape of the samples during heating.
Fig. 3 shows an example of the course from
softening to flow out of a specimen. The results
show that the presence of BOF-slag, burnt lime or
burnt dolomite in the specimens significantly
decrease the melting temperature. The BOF slag
also decreases the softening and melting temperature
interval, Fig. 4.

Table 1 Amounts of fluxes injected (kg/thm) corresponding to different injection levels assumed.

Injection Coal injection rate

levels of 110 kg/thm 123 kg/thm 150 kg/thm
fluxes BOF slag Lime Dolomite BOF slag Lime Dolomite BOF slag Lime Dolomite
None 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
Low 8.3 4.2 6.8 8.8 4.4 7.1 9.5 4.7 7.1
Medium 16.7 8.3 13.4 17.5 8.8 14.1 18.9 9.5 14.1
High 33.5 12.5 20.1 35.2 13.2 21.2 37.8 14.2 21.1
Very High 50.1 16.7 26.7 52.8 17.5 28.2 56.4 18.9 28.1

a b c d

Fig. 3 Contour of sample at start of test (a), softening (b), active melting (c) and flow point (d).

International BF Lower Zone Symposium, Wollongong 2002 - Preprint

18 - 3
 J.-O. Wikström, P. Sikström, L. Sundqvist and G. Zuo

1550 Softening, BOF slag

Study of addition of BOF slag combined with
Temperature, °C

Flow point, BOF slag a) different pellet types

1450
DTA
1350
In order to reveal the influence of fluxes on the slag
1250
formation process in the blast furnace, softening and
1150 melting tests of four types of pellets, as shown in
0.0 0.5 1.0 1.5 2.0
Table 2 (self-fluxed - A and B; olivine pellets –
Basicity, B4
MPBO and KPBO.) with and without the addition of
Softening, burnt lime
BOF slag, were carried out. The experimental
Temperature, °C

1550
Flow point, burnt lime b)
1450 DTA apparatus (Fig. 5) has a maximum working
1350 temperature of 1600oC. The graphite crucible used
1250 (Fig. 6 40 mm i.d. × 85 mm), has 6 holes of 6 mm
1150 diameter on the bottom, allowing the dripping of the
0.0 0.5 1.0 1.5 2.0 molten materials and the inlet of gas through the
Basicity, B4
sample bed.
Temperature, °C

1550 Softening, burnt dolomite

Flow point, burnt dolomite c)
1450
DTA
1350 Pneumatic Position
cylinder transducer
1250
1150 pipe
PC 1
0.0 0.5 1.0 1.5 2.0
Thermocouple
Basicity, B4 Differential pre-
Rod ssure transducer

Fig. 4 Measured softening and flow out Thermocouple PC 2

temperatures of the specimens based on no Crucible Water
injection and injection of BOF slag (a), Gas
Support
burnt lime (b) and burnt dolomite (c). The tube
pipe

basicity B4 on the x-axis corresponds to the Rotating con-

tainer of drops
N2

different injection levels. Motor Balance

According to the laboratory results obtained, an

improvement of slag formation in front of the Fig. 5 Schematic view of the experimental
tuyeres can be expected, if injection of fluxes is apparatus.
made. It seems that the injection of BOF-slag
improves the properties more than burnt lime or Thermocouple tube
dolomite, in terms of lower melting point and more Rod
narrow temperature interval between softening and Ram
flow out, and possibly also a lower viscosity, at all
injection levels being studied. Sample

The explanation is as follows: Crucible

BOF slag contains a larger amount of

components such as FeO, MnO, TiO2, V2O5, etc
that are included neither in the basicity formulae
nor in the phase diagrams. These components
could further decrease the melting point and the
viscosity of the slag formed.
The BOF slag injected is a slag already formed
in the converter, having a relatively low melting
point, while the burnt lime and dolomite are
more pure materials with much higher melting
temperatures. Therefore, BOF-slag is more
easily re-melted and dissolved into other
components.
The alumina rate in the tuyere slag is lower in
the case of BOF-slag injection.
Support
tube
Fig. 6 Placement of the samples in the furnace
tube.
Samples of self-fluxed pellets for the tests were pre-
reduced to reduction degrees of 60, 75 and 90%
under conditions of 850oC and 55%N2-40%CO-
5%H2 reducing gas. Olivine pellets were pre-
reduced at a temperature of 950oC with the same
reducing gases. Accordingly, BOF slag used was
also pre-reduced under the same conditions.

International BF Lower Zone Symposium, Wollongong 2002 - Preprint

18 - 4
 J.-O. Wikström, P. Sikström, L. Sundqvist and G. Zuo

Table 2 Main chemical components of pellets and BOF slag.

CaO CaO+MgO CaO+MgO
Sample CaO SiO2 MgO Al2O3 Fe2O3 FeO Tfe SiO SiO2 SiO2+Al2O3
2
Pellet A 1.33 1.33 0.35 0.32 95.63 0.44 67.23 1.00 1.26 1.28
Pellet B 1.22 1.31 0.76 0.38 95.16 0.55 66.99 0.93 1.51 1.59
MPBO 0.28 1.97 1.48 0.36 94.54 0.77 66.72 0.14 0.89 1.15
KPBO 0.40 2.15 1.47 0.23 94.77 0.37 66.57 0.19 0.87 1.15
BOF-slag 41.09 9.84 12.97 0.95 8.08 16.40 18.40 4.18 5.49 6.09

Even mix of
Coke pellets and
BOF-slag
Pellets BOF-slag
BOF-slag Uneven mix- Pellets
Coke ture of pellets
and BOF-slag

a b c
Fig. 7 Placement of BOF slag in the pellets bed.

Considering the amount used and segregation of the
BOF slag in the blast furnace, 5 or 10% of BOF slag
by weight of pellets were added into the pellets
sample bed. Three different placements of BOF slag
in the pellets bed were used, as shown in Fig. 7, by
adding BOF slag to the self-fluxed pellets bed, as
one layer beneath the pellets (Fig. 7-a), evenly
distributed in the pellets bed (Fig. 7-b) or arbitrarily
mixed with pellets (Fig. 7-c). The samples were
sandwiched by two layers of coke in the crucible
during the softening and melting tests.
The experimental results showed that without the
addition of the BOF slag these pellets have rather
good softening and melting properties and can
satisfy the demands of blast furnace operations.
Fig. 8 shows the softening and melting temperatures
of these pellets with a pre-reduction degree of 75%.
Adding BOF slag to the pellets bed could give
different effects on the softening and melting
properties of the pellets, depending on the pellet
types and the placement of the BOF slag in the
sample bed. For self-fluxed pellets, when BOF-slag
was evenly distributed in the pellet bed, it did
increase the melting temperature by more than 80oC.
No dripping material was collected up to a test
temperature of 1500oC. It seems that the BOF slag
can block the dripping process completely during the
tests. When BOF slag was unevenly placed in the
sample bed, it increased the melting temperature by
about 50oC, and reduced the amount of the dripped
metallic iron. When it was uniformly placed
beneath the pellet layer, no change of melting
temperature of the pellets could be observed. Fig. 9
shows the change of the softening and melting
temperature of pellet-A (75% pre-reduction degree)
with the different placement of the BOF-slag in the
sample bed.

Fig. 8 Softening and melting temperature of

different pellets used. Fig. 9 Changes of the softening and melting of
pellets with the placement of BOF-slag
(5%).

International BF Lower Zone Symposium, Wollongong 2002 - Preprint

18 - 5
 J.-O. Wikström, P. Sikström, L. Sundqvist and G. Zuo
Furthermore, white-grey powders could be found in
the crucible after the tests for the latter two cases.
X-ray diffraction analyses showed that the white-
grey powder contained di-calcium silicate, and other
complex compounds. TGA tests showed that the
melting temperature of the powder was higher than
1600oC. These results indicate that compounds with
high melting points, e.g. di-calcium silicates, can
form in the course of slag formation. The
mechanism of generation of these powders and their
influences on the slag properties remain to be further
studied.
Fig. 10 Variation of the softening and melting
temperatures of MPBO with the addition of
BOF-slag; Pre-reduction degree of pellets –
75%.
The reason for the troublesome melting process
when adding BOF slag to the self-fluxed pellet bed
can be attributed to the formation of basic slag of
high basicity and the reduction of FeO in the slag.
With dissolution of BOF slag into the primary slag
of self-fluxed pellets, the basicity of slag formed
increases accordingly. When the inevitable
segregation of BOF slag is considered, the slag
formed could have very high basicity locally.
Considering 5 and 10 % addition of BOF slag, B2 of
the bosh slags formed can reach up to 1.87 and 2.36
respectively. It is the same as or very close to the
basicity 2 of dicalcium silicate (2CaO.SiO2, B2=
1.87).
In addition, reduction of FeO can significantly
change its melting point at this range of slag
basicity. With decrease of FeO content, the high
melting point of the bosh slag can further increase its
melting point dramatically. Therefore, adding BOF
slag into the samples will bring twofold effects on
melting properties: 1) Increase melting point by an
increased basicity; 2) Further increase melting point
by reduction of FeO in this region. In this case,
formation of a very high melting point slag of di-
calcium silicate is quite possible.
The improvement of the melting process when
adding BOF slag to the olivine pellets bed may be
due to the formation of the favourable bosh slag.
International BF Lower Zone Symposium, Wollongong 2002 - Preprint
18 - 6
Both MPBO and KPBO pellets contain high MgO,
but with a low ratio of CaO/SiO2, quite different
from that of self-fluxed pellets. Its primary slag has
an estimated melting point of about 1600oC. When
adding 5 or 10% BOF slag to the pellets bed, the
basicity of the slags formed increases to a level of
about 0.8 and 1.2 respectively. This type of slag has
a melting point of about 1400oC, which is quite
suitable for blast furnace operation. The test results
indicate that the melting behaviour is slightly better
with the lower addition of BOF slag.
These results indicate that when using self-fluxed
pellets as the burden of the blast furnace process,
top-charging BOF slag is not a favourable
alternative for improving the control of the blast
furnace operation.
PILOT PLANT TRIALS WITH CO-
INJECTION OF BOF SLAG AND
PULVERIZED COAL
Experimental BF
To verify the positive effect of co-injection of BOF
slag and pulverized coal, a test campaign6 was
performed in LKAB’s Experimental BF, located at
MEFOS premises in Luleå. A simplified lay out of
the experimental BF is shown in Fig. 11.
Fig. 11 Illustration of the BF and its auxilliary
equipment.
The furnace has a working volume of 8.2 m3 and a
hearth diameter of 1.2 m. To ensure realistic
conditions heat losses must be kept at a minimum.
For this reason insulating refractory was chosen, and
only the bosh area and the 3 tuyeres are water-
cooled. The blast is preheated to 1200°C in pebble-
bed stoves. The furnace is equipped with a bell-type
top with moveable armour for burden distribution
control. Probes for temperature measurement, gas
analysis and solid sampling over the blast furnace
diameter are installed at three different levels 7.
The experimental BF is operated in campaigns of
normally 4-10 weeks at a productivity ranging from
3.2 to 3.8 t/m3day. Normal tap-to-tap time is 60
 J.-O. Wikström, P. Sikström, L. Sundqvist and G. Zuo

minutes and normal tapping duration 5–15 minutes. The trial was carried out with a mixture of 70%
Process data are logged continuously and stored in a Pellet A (fluxed), and 30% Pellet B (olivine), except
database. The experimental BF has proved to be a for the last period, which was operated with 100 %
sensible tool for detecting differences in the Pellet A. The reason for mixing the two types of
properties of different ferrous burden. The response pellets was to achieve a suitable slag composition
time is much shorter for the experimental BF and therefore the excessive basicity of bosh slag
compared to a commercial furnace. mentioned in the introduction was not achieved. The
quite high Si content of hot metal in the
The experimental BF is equipped with a lock-hopper experimental BF demands that quartzite is added to
coal injection system, see Fig. 12. Below the the burden. This addition also decreases the bosh
injection vessel a cylindrical fluidisation chamber is slag basicity. The last period was operated with
fitted. For BOF-slag injection a separate vessel is only Pellet A as ferrous burden, in order to reach an
connected to the fluidisation chamber, via a acceptable slag basicity without charging any fluxes
volumetric screw feeder. The BOF slag is totally through the top. Chemical analysis of ferrous raw
mixed with the pulverized coal in the fluidisation materials and fluxes are shown in Table 4.
chamber, before pneumatic transportation to the 3
tuyeres6. Table 4 Chemical analysis of the raw materials.
Pellet A Pellet B BOF Lime- Quart-
slag Stone zite
Fe 66.6 66.7 15.67 0.28 0.83
CaO 1.60 0.35 38.2 53.4 0.60
SiO2 1.60 2.10 12.3 1.91 93.0
MgO 0.35 1.45 9.47 1.12 0.55
Al2O3 0.40 0.20 5.55 0.07 3.34

The coke was crushed and sieved to a fraction of 15-

Fig. 12 Injection system.
Test conditions
During a two-week period in November 2000, a trial
with simultaneous injection of pulverized coal and
BOF-slag was carried out. The trial was divided
into five periods, the first one being a ”reference”
period, without any BOF-slag injection. The
original test plan is shown in Table 36.
30 mm, i.e. standard coke for the experimental BF.
The top-charged slag forming additives were lime
stone, quartzite and BOF-slag. The particle size of
the slag formers was 9-15 mm, and the injected BOF
slag particles were slightly smaller than those of the
pulverized coal, i.e. 80 % < 100 µm. The reason for
choosing BOF slag as tuyere injectant is a result of
the fundamental studies earlier described. Besides
that, BOF slag is also a material that is easy to
handle in a pneumatic system.

Table 3 Original trial plan.

Duration Slag volume Pellets type PCI BOF slag inj. Top charged
Period Hours kg/t-HM A/B kg/t-HM kg/t-HM slag formers
Reference 48 140 70/30 100 0 Yes
Period 1 72 140 70/30 100 25 Yes
Period 2 24 125 70/30 100 25 Yes
Period 3 96 115 70/30 100 25 Yes
Period 4 120 100 100/0 100 36 No

International BF Lower Zone Symposium, Wollongong 2002 - Preprint

18 - 7
 J.-O. Wikström, P. Sikström, L. Sundqvist and G. Zuo

Results of pilot plant trials
The operational results for the periods with co-
injection of BOF-slag and pulverized coal were
excellent, and the trial was performed without any
major disturbances6. Burden descent and
permeability, as well as other process parameters,
were smooth and stable. Mass balances for each
period are shown in Table 5.
The reductant rate was totally decreased by 11 kg/t-
HM in the last period compared to the reference
period. The change was primarily in coke rate, as
The trial showed that the hot metal silicon content
significantly decreased, with small changes in the
carbon content, see Table 7. During period 4,
without any top-charging of additives, the standard
deviation in hot metal silicon stabilized at a much
lower level, which indicates that the blast furnace
process was stable. The correlation between silicon
content and hot metal temperature is shown in
Fig. 13, which indicates a reduction of silicon
content by 0.3 %.
[% Si] REF Si Per 4 Si
2

the coal injection rate was kept constant. The main 1,8

reasons for the lower reductant rate was the fact that 1,6

it was possible to operate the furnace with a lower 1,4

and more stable hot metal silicon content, as well as 1,2

1
a lower slag volume. 0,8

0,6

The blast volume, as well as flame temperature, was 0,4

kept relatively constant during the whole trial, see 1350 1370 1390 1410 1430 1450 1470 1490
Temperature °C
Table 6. The gas utilization was roughly unchanged Fig. 13 Silicon in relation to hot-metal
during the whole trial, approximately 47 %, which is temperature.
a normal level for this kind of ferrous burden
materials in the experimental BF.

Table 5 Results of calculated mass balance for each period.

Hours Prod. Slag vol. Coke Coal inject Fuel-rate BOF slag inj Top-charged
total tonne/h kg/t-HM kg/t-HM kg/t-HM kg/t-HM kg/t-HM slag formers
Ref. 34 1.33 136 439 98 537 0.0 Limestone, Quartzite, BOF slag

Period 1 57 1.39 138 444 93 538 23.3 Limestone, Quartzite, BOF slag

Period 2 20 1.38 129 446 99 545 24.3 Limestone, Quartzite, BOF slag

Period 3 92 1.43 110 440 93 532 24.3 Limestone, Quartzite, BOF slag

Period 4 110 1.42 101 428 98 526 36.9 None

Table 6 Tuyere and top-gas parameters for each period.

Tuyere parameters Top-gas parameters
Blast Temp Flame Temp EtaCO H2
°C °C % %
Reference 1129 2158 0.466 2.78
Period 1 1129 2164 0.470 2.70
Period 2 1129 2148 0.466 2.76
Period 3 1129 2147 0.473 2.74
Period 4 1130 2157 0.461 2.78

International BF Lower Zone Symposium, Wollongong 2002 - Preprint

18 - 8
 J.-O. Wikström, P. Sikström, L. Sundqvist and G. Zuo

Table 7 Average hot-metal component concentrations for each trial period.

Trial Temp C Si Si C/Si Mn P S (S)/[S]
Period °C wt-% wt-% stdv wt-% wt-% wt-% %/%
Reference 1416 4.12 1.28 0.31 3.17 0.16 0.044 0.120 11.5
Period 1 1418 4.31 1.09 0.27 4.01 0.17 0.043 0.095 12.5
Period 2 1431 4.51 1.16 0.25 3.91 0.16 0.040 0.057 26.0
Period 3 1426 4.37 1.04 0.19 4.20 0.13 0.037 0.078 21.1
Period 4 1423 4.24 1.00 0.18 4.09 0.14 0.035 0.082 22.8

During the trial the slag volume was decreased from
140 to 100 kg/t-HM. The average slag composition
is shown in Table 8. The low slag volume, and
accordingly higher Al2O3 content, did not cause any
problems regarding slag viscosity. The change in
slag composition is shown by the arrow in Fig. 14,
indicating that a further increase in Al2O3 or CaO
content might cause a problem by increasing the
melting temperature. Pellets included in material
samples taken with the material sampling probe at
the upper level of the cohesive zone were studied by
SEM analysis. Condensed slag phases were found
on the pellet surface as well as in areas of high
porosity. One example is shown in Fig. 16. The
slag often contains oxides with a high content of K
in combination with Si and Al. These oxides were
found in all samples studied from the reference
period as well as from the test periods.

Table 8 Average slag component concentrations for each trial period.

Period CaO SiO2 Al2O3 MgO K2O K2O- S S-distr. MnO Basicity Bells
Wt-% wt-% wt-% wt-% wt-% yield wt-% (%) /[ %] wt-% CaO/SiO2 Basicity
Reference 32.1 38.6 14.7 11.6 0.59 0.64 1.28 11.5 0.42 0.83 1.04
Period 1 34.1 37.5 14.6 11.5 0.43 0.50 1.31 12.5 0.34 0.91 1.12
Period 2 34.3 34.0 16.0 11.7 0.34 0.41 1.63 26.0 0.22 1.01 1.23
Period 3 33.4 33.6 17.7 11.8 0.41 0.46 1.70 21.1 0.22 0.99 1.21
Period 4 34.5 33.3 18.9 9.5 0.51 0.54 1.96 22.8 0.20 1.04 1.20

REFK2O Per 4K2O

1
0,9
0,8
0,7
% K2O in slag

0,6
0,5
0,4
0,3
0,2
0,1
0
0,90 1,00 1,10 1,20 1,30 1,40
Bells Ratio
Fig. 15 K2O and basicity in slag.

With the injection of BOF slag the tuyere slag is

almost neutral. That causes the sulphur released
from the combustion of coke and coal to more easily
Fig. 14 Phase diagram showing the change in slag be absorbed by the slag. The trial showed that the
composition. sulphur distribution was improved when BOF slag
was injected, see Fig. 17.
The absorption of alkalis by the slag was also
slightly improved during the trial, as shown in
Fig. 15. No problems associated with the
accumulation of alkalis were noted in spite of the
fact that the slag basicity was higher at the end of the
trial.

International BF Lower Zone Symposium, Wollongong 2002 - Preprint

18 - 9
 J.-O. Wikström, P. Sikström, L. Sundqvist and G. Zuo

70 µm Experimental BF quenching and excavation

During the last eight hours of operation of one

c campaign in the experimental BF, different injection
was performed in each tuyere, i.e. tuyere 1 without
injection, tuyere 2 with coal injection and tuyere 3
c with co-injection of BOF slag and pulverized coal.
Injection rates per tuyere were at the same level as
during the trial. After the injection period, the
furnace was stopped and quenched with nitrogen.
c After cooling down, the burden was excavated from
the top to the tuyere level, and a further examination
of each raceway area was done6.
Fig. 16 Slag phases (C) with high content of K, Si
and Al were found at the surface of the In Table 9, the chemical composition of bosh slag is
pellets during the reference period, period 3 estimated, from mass balances. As can be seen, the
and period 4. basicity (B2) values are quite similar, but the Bells
basicity is much lower for the injection period. The
30,00
30
table also shows an estimated chemical composition
of the tuyere slag. The basicity is significantly
25,00
25 increased and alumina content decreased, when BOF
slag is injected.
20,00
20

15,00
15

10,00
10

5,00
5

0,00
0
Ref period Period 1 Period 2 Period 3 Period 4

Fig. 17 Sulphur distribution (%S)/[%S].

Fig. 18 From left: tuyeres and raceway view from above; raceway size from above; bottom shape of raceway
view from side.

The difference between the raceways was significant Discussion

(Fig. 18). The raceway corresponding to “normal”
conditions, i.e. coal injection, had a characteristic
“birds nest” formation. The raceway with “all coke
operation” and the raceway with co-injection of
BOF-slag and coal showed a porous end of the race
ways. In normal operation a “birds nest” is formed
by the acidic slag from coal and coke, which has not
been drained from the raceway area, and the char
particles that accumulate at the end of the raceway.
The consumption of reducing agents was lowered by
11 kg/t-HM when all BOF-slag was injected through
the tuyeres. The slag volume was lowered by
35 kg/t-HM, which corresponds to ~7 kg of coke.
The decreased silicon content in the hot metal
corresponds to a decrease in consumption of
reducing agents of ~10 kg/t-HM. On the other hand,
hot metal temperature as well as hot metal carbon

International BF Lower Zone Symposium, Wollongong 2002 - Preprint

18 - 10
 J.-O. Wikström, P. Sikström, L. Sundqvist and G. Zuo

content was slightly higher during the injection over the carburisation reaction. The amount of Si
period. reduced from SiO2 in the slag is also expected to
decrease because of the lower slag volume.
Table 9 Estimated chemical component However, the results indicate that the effect of
concentrations (by weight) of the bosh changed raceway conditions is the main reason for
and tuyere slag formed during the trial. the decreased Si content of the hot metal. The large
Bosh slag Tuyere slag drop in Si content of hot metal occurs when injection
of fluxes has started, a further decrease occurs when
Ref. Period 4 Ref. Period 4
the slag volume is decreased, and when slag
CaO 35.9 33.1 1.0 28.7
injection is increased there is less effect on the Si
MgO 11.2 7.2 0.9 7.4
content.
SiO2 39.9 38.5 59.9 39.0
Al 2O3 5.5 10.1 27.6 15.5
Na2O 0.7 1.6 0.7 0.4 INJECTION OF BF FLUE DUST
K2O 0.5 0.6 1.5 0.8
MnO 2.1 1.0 0.2 2.6 Blast furnace flue dust contains valuable amounts of
B2 0.90 0.86 0.02 0.74 carbon and iron. The content of harmful tramp
Bells Ratio 1.64 1.19 0.03 0.95 elements is normally low enough for recycling into
the blast furnace. Since 1993 SSAB Tunnplåt in
The melting behaviour of the tuyere slag, shown in Luleå has recycled flue dust by charging it into the
Fig. 19, was estimated by calculations using the blast furnace in the form of a cold bonded dust
computer program Chemsage. This program uses a briquette8. As the briquette quality as well as
thermodynamic data file for multi-component slags. productivity are negatively affected by flue dust,
The temperature for formation of the first melt is tuyere injection was considered as an alternative
increased, and the temperature at which >80% of the solution. The possibility of injecting flue dust was
slag is melted is decreased. This results in a more first evaluated by characterisation of the flue dust
narrow softening and melting temperature interval and studies on the raceway conditions. Two
for the tuyere slag, and it is expected that the successful pilot-scale trials (2 days and 5 days,
viscosity also decreases, when BOF slag is injected. respectively) were carried out on the experimental
This might be the main reason why no birds nest is BF. The amount of reducing agents could be
formed in the end of the raceway. decreased and the blast furnace operation was stable
and smooth. The test results implied that tuyere
100 injection is a suitable method for recycling of flue
% slag melt formed

80 dust.
Characterisation of flue dust
60
40 Reference Blast furnace flue dust contains fine particles of the
20
burden materials, ie mainly pellet and coke fines. It
Period 4 contains normally 20-30 % Fe and 40-50 % C. In an
0 earlier injection trial the flue dust was ground in a
1000 1100 1200 1300 1400 1500 rod mill to 80% < 80 µm. In that case it was very
Temperature, °C
sticky and clogged in the pipework. The reason is
that particles that are easy to grind become very fine,
Fig. 19 Estimated melting behaviour of the
before the more hard particles have reduced their
tuyere slag formed during the trial.
size.
The Si content in hot metal is significantly decreased
For the experimental BF trial, unground flue dust
when BOF slag is injected through the tuyeres. One
was chosen, which means that it was taken directly
important mechanism providing the hot metal with
from the cyclone at BF No. 3 at SSAB in Luleå
Si is considered to be SiO gas generation at the high
(Fig. 20). When the dust is taken from the cyclone,
temperatures in the flame zone and reduction of SiO2
it has gone through a windscreen when it follows the
in the slag by carbon in the coke at temperatures
off gas out of the blast furnace. This means that the
above 1700oC. When BOF slag is injected, the
particles have the same terminal velocity,
basicity of the tuyere slag increases significantly,
independent of density. For fluidisation of flue dust
and thereby the activity of SiO2 in the tuyere slag is
together with pulverized coal, this is a big
decreased. The MgO content also increases and as a
advantage.
result the wettability of the slag decreases. These
facts reduce the generation of SiO gas. If the partial
pressure of SiO gas is very high at the tuyere level,
decarburisation of the hot metal can occur due to the
reaction SiO + C = Si + CO, and can predominate

International BF Lower Zone Symposium, Wollongong 2002 - Preprint

18 - 11
 J.-O. Wikström, P. Sikström, L. Sundqvist and G. Zuo

100 Table 10 Raw materials used during the test,

kg/t-HM.
80
Reference Test
60
Top Pellets 1368 1368
charging Limestone 30.2 35.6
%

40 Quartzite 23.1 20.2

Pulverized coal BOF slag 64.6 52.4
20
Flue dust 1st test Coke 444 412
Flue dust 2nd test
Injection Coal 101 103
0 Flue dust 0 23.4
1 10 100 1000 O2 enrich. 2.71% 2.75%
µm
Fig. 20 Particle size distribution of injected Table 11 Chemical composition of hot metal
materials. and slag (wt%).
Reference Test
Laboratory tests – The effect of flue dust Hot Temp. 1461ºC 1448ºC
injection on coal combustion Metal C 4.49 4.45
Si 2.18 1.78
Laboratory tests, to study the effect of flue dust
S 0.030 0.037
addition on the combustion of pulverized coal, have
Slag CaO 30.7 30.6
been performed in a fixed bed in cooperation with
SiO2 34.3 35.3
University of Science and Technology in Beijing.
The test results shows that the blast temperature, the MgO 18.9 19.3
coal injection rate as well as the amount of flue dust Al2O3 15.3 14.1
addition have a significant effect on combustion CaO/SiO2 0.87 0.89
efficiency. The effect of increased blast temperature
and dust amount on combustion efficiency is Both the reference period and the test period were
positive, but an increased coal injection rate has a characterized by stable blast furnace operation.
negative effect. The change in particle size of blast However, the variation in burden descent rate
furnace flue dust, in the range 100 to 200 mesh had decreased significantly when flue dust was injected.
no significant effect on coal combustion efficiency. The burden resistance index is higher during the test
period, but with less variation.
Injection of flue dust into the experimental BF
Discussion
The same equipment and operational procedure were
used as for the trials with BOF slag injection. The The consumption of reducing agents decreases when
trial worked out very well, without any major flue dust is recycled, because the carbon in the flue
problems. However, as a result of the layout of the dust replaces carbon in coke and coal. The
injection system, the control of coal injection rate consumption decreased by 21 kg/t-HM when an
was affected by the flue dust addition. This resulted amount of flue dust corresponding to a coal injection
in slightly increased variation in gas utilization. The rate of 14 kg/t-HM was injected. The gain of carbon
average injection rate during the test was was very high, and the replacement ratio is estimated
23 kg/t-HM8. to be approximately 1.5, mainly due to decreased
silicon content in the hot metal.
As can be seen from Table 10, the consumption of The decreased silicon content in hot metal is
reducing agents based on mass balances was considered to be a result of changed conditions in
decreased by 21 kg/tHM during the trial. The the raceway. The basicity of the tuyere slag is
injected flue dust had a carbon content of 47 %. The slightly increased, as can be seen in Table 12. The
C content in the injected amount of flue dust Fe2O3 corresponding to 23-wt.% of Fe in the flue
corresponds to approximately 14 kg of pulverized dust is supposed to be reduced in the raceway. This
coal. No significant difference in the heat level of will influence the oxygen potential in the raceway.
the blast furnace was noticed, except for reduced
silicon content in the hot metal (see Table 11).

International BF Lower Zone Symposium, Wollongong 2002 - Preprint

18 - 12
 J.-O. Wikström, P. Sikström, L. Sundqvist and G. Zuo
Table 12 Estimated chemical composition of
tuyere slag. All amounts in wt.%.
Reference Test
CaO 0.95 6.21
MgO 0.83 1.86
SiO2 59.7 55.8
Al2O3 27.6 25.1
TiO2 1.58 1.53
Na2O 0.65 0.58
K2O 1.45 1.33
MnO 0.16 0.93
B2 0.02 0.11
Bells Ratio 0.03 0.14
CONCLUSIONS
Co-injection of BOF slag as well as blast furnace
flue dust together with pulverized coal has been
investigated through laboratory scale experiments
and test campaigns in LKAB’s Experimental blast
furnace, located at MEFOS’ premises in Luleå,
Sweden.
By injection of BOF-slag through the tuyeres,
instead of charging fluxes through the top of the
blast furnace, formation of an acid tuyere slag can be
avoided, while in the same time formation of a slag
with an excessive basicity in the shaft is also
avoided. Trials in the experimental BF with co-
injection of BOF slag and pulverized coal resulted in
a smooth and stable operation. The slag volume was
around 100 kg/t-HM and the reductant rate was
lowered by 11 kg/tHM. A significant stabilization
and reduction of hot metal silicon content was
reached. Sulphur distribution, as well as the output
of K2O by the slag, was also slightly improved.
Quenching and excavation of the blast furnace
showed that the formation of a birds nest can be
avoided when the acid slag, normally generated in
the raceway, is fluxed by the BOF slag.
Successful trials with co-injection of blast furnace
flue dust and pulverized coal have also been
performed in the experimental BF. The injection of
flue dust improves the coal combustion efficiency as
well as the slag formation in the raceway. The
improved raceway conditions resulted in a
significant stabilisation and decrease in hot metal
silicon content. As a result the gain of carbon was
very high and the replacement ratio was estimated to
be approximately 1.5, when injecting 23 kg/t-HM of
blast furnace flue dust. Industrial trials have started
at BF No. 3 at SSAB Tunnplåt in Luleå.
International BF Lower Zone Symposium, Wollongong 2002 - Preprint
18 - 13
ACKNOWLEDGEMENTS
The authors wish to thank Jernkontoret (The
Swedish Steelmakers Association) for financial
support for the two-week campaign in the
experimental BF. We also thank LKAB and SSAB
for support in the form of equipment, raw materials,
analysis, manpower, etc during the trial. Finally we
thank the Swedish National Energy Administration
(STEM) for financial support for the fundamental
studies.
REFERENCES
1. J. Ma, (1999), ”Injection of flux into the blast
furnace via tuyeres for optimising slag
formation ”, ISIJ International, Vol 39, No.7, p.
679-704.
2. M. Brämming, J-O. Wikström, (2002), ”A blast
furnace view on slags”, Scandinavian Journal of
Metallurgy, Vol 31, p. 88-99.
3. L. Sundqvist Ökvist, G. Zuo, J. Ma, (2000),
”Influence of basic fluxes on slag formation in a
blast furnace using LKAB pellets”, 6th
International Conference on Molten Slags,
Fluxes and Salts, Stockholm-Helsinki, Sweden-
Finland.
4. L. Sundqvist Ökvist, G. Zuo, (2000), ”Effect of
fluxes on slag formation in a blast furnace by
top charging and tuyere injection”, 7th Nordic-
Japan Symposium, Nagoya, Japan.
5. G. Zuo, (2000), ”Softening and melting
characteristics of self-fluxed pellets with and
without the addition of BOF slag” ISIJ
International, Vol 40, p. 1195-1202.
6. P. Sikström, L. Sundqvist Ökvist, J-O.
Wikström, (2002), ”Injection of BOF-slag
through blast furnace tuyeres – trials in an
experimentel blast furnace”, ISS 61st
Ironmaking Conference, Nashville, Tennessee,
USA.
7. A. Dahlstedt, M. Hallin, J-O Wikström, (2000),
”Effect of raw materials on blast furnace
operation - the use of an experimental blast
furnace”, European Ironmaking Congress, Paris,
France.
8. L. Sundqvist Ökvist, P. Sikström (2002),
”Recycling of flue dust into the blast furnace”,
TMS Conference, Luleå, Sweden.
9. L. Sundqvist Ökvist, (2001), ”The melting
properties of tuyere slag with and without flux
injecion into the blast furnace”, ISIJ
International, Vol 41, No 12, p. 1429-1436.
10. T Sawada, R. Nakajima, S. Kishimoto, H. Hotta
and K. Ishii (1990), "Flux injection from the
tuyere for low silicon operation", 49th
Ironmaking Conference, 25-28 March, Detroit,
Michigan, USA, p. 309-315.
 J.-O. Wikström, P. Sikström, L. Sundqvist and G. Zuo

11. C. Yamagata, Y. Kajiwara, S. Suyama, K. Sato

and S. Komatsu (1990), "Simultaneous Injection
of Pulverized Coal and Dolomite into Blast
Furnace Tuyeres", ISIJ International, Vol 30,
No 5, p. 362-371.
12. K. Kushima, M. Naito, K. Shibata, H. Sato, H.
Yoshida and M. Ichida (1988) "Iron ore
injection into blast furnace raceway", 47th
Ironmaking Conference, 17-20 April, Toronto,
Canada, p. 457-466.
13. K. Takeda, Y. Sawa, S. Taguchi, N. Takashima,
T. Matsumoto and H. Obata (1990), "Iron ore
and flux injection through blast furnace tuyeres
and its effect on raceway phenomena", The
Sixth International Iron and Steel Congress,
Nagoya, Japan.
14. G. Brun, R. Nicolle, J.M. Steiler, J.L.
Bouttement, J.L. Eymond and J.P. Menaut
(1988), "Injection de minerai au haut fourneau
et amitrise de la teneur en silicium", La rrevue
de Metallurgie CIT, janvier , Vol 85, No 1,
p.21-32.
15. C. Yamagata, Y. Kajiwara, S. Suyama and
T.Miyake (1990), "Desiliconisation Reaction of
Pig Iron with High FeO containing Blast
Furnace Slag under Prezzurised and Coke.
coexisting conditions", ISIJ International, Vol
30, No 9, p. 740-747.
16. H.W. Gudenau, H. Denecke and H. Wipperman
(1998), "Iron Ore and Iron Containing Dust
Injection into the Blast Furnace", 2nd
International Congress on the Science and
Technology on Ironmaking and 57th
Ironmaking Conference, Toronto, Canada.

International BF Lower Zone Symposium, Wollongong 2002 - Preprint

18 - 14