6872 Ind. Eng. Chem. Res.

2003, 42, 6872-6879

Investigation of a Mineral Melting Cupola Furnace. Part I.

Experimental Work
R. Leth-Miller,*,†,‡ A. D. Jensen,‡ P. Glarborg,‡ L. M. Jensen,† P. B. Hansen,† and
S. B. Jørgensen‡
Rockwool International A/S, Hovedgaden 584, DK-2640 Hedehusene, Denmark, and Department of Chemical
Engineering, Technical University of Denmark, Building 229, DK-2800 Lyngby, Denmark

Full-scale measurements on a mineral melting cupola furnace for stone wool production have
been made to investigate the behavior and increase the understanding of mineral melting cupolas.
The work includes probe measurements of gas concentration, solid/melt phase and gas
temperature, and melt composition at several vertical and radial positions in the cupola. Four
types of experiments were conducted: quenching of an operating cupola, sample collection from
the hot part of the cupola using probes through the walls, sample collection through a probe
inserted from the top of the cupola, and collection of input and output data. The experiments
showed the position of the melting zone, porosity, and other characteristics. The probe
experiments indicate that the oxygen is consumed approximately 0.3 m above the tuyeres. The
top probe experiments gave the temperature profile in the cupola.

Introduction
The production of stone wool products involves melt-
ing of rock materials and subsequent spinning of the
melt into fibers (wool). The melting of the rocks is
usually carried out in a cupola furnace. A few experi-
mental investigations of the conditions inside cupolas
have previously been reported in relation to the foundry
industry.1-3 The results of these measurements are not
directly comparable to stone wool cupola operation,
however, because the CO concentration is higher in the
foundry cupolas and the properties of the iron charged
in foundry cupolas are different from the properties of
the rocks used in stone wool cupolas. Nevertheless, the
information provided by these investigations does pro-
vide indications of the conditions in a stone wool cupola.
Experimental data for fixed-bed coal gasifiers, found in
ref 4, can also be used for qualitative comparisons with
stone wool cupola measurements.
In stone wool cupolas, the reduction of iron oxides to
metallic iron is unwanted, in contrast to the case in the
foundry cupolas. The location and mechanism of the
reduction of the iron oxides contained in the raw
materials in cupolas are not well-known. Mass transport
of FeO in foundry slag has been reported to be rate-
controlling,5-8 although others have found that the
reduction of iron oxide with CO at the slag-gas inter-
face is rate-controlling.9 Yet other authors report that
the CO formation rate in the Boudouard reaction
[C(s) + CO2(g) f 2CO(g)] is rate-controlling.10 In all of
the proposed mechanisms, the iron reduction can take
place only after the iron has melted, i.e., in the lower
first meter of the cupola and in the melt bath.
Through experimental investigations, this work aims
to obtain new insights about the conditions inside the
cupolas used for stone wool production. The thermal
efficiency of stone wool cupolas is only approximately
* To whom correspondence should be addressed. Tel.: +45
4656 0300. Fax: +45 4655 5990. E-mail: rasmus.leth.miller@
rockwool.com.
†
Rockwool International A/S.
‡
Technical University of Denmark.
10.1021/ie020449w CCC: $25.00 © 2003 American Chemical Society
Published on Web 10/03/2003
50%, i.e., about 50% of the heat released from combus-
tion of the coke is used for heating and melting the
rocks, and the rest is lost to the surroundings, mainly
to the cooling water and as CO in the flue gas. Thus,
there is a great potential for improvement of the energy
efficiency of the cupola.
To achieve the goals of this work, an elaborate set of
experimental investigations was carried out both on the
coke11 and raw material12 properties and on full-scale
operating cupolas. The latter experiments are reported
here after the mineral melting cupola is described. The
experimental data were also used to develop a math-
ematical model of the cupola (see ref 13).
Process Description
A cupola furnace for melting rocks is a vertical shaft
furnace, 4-6 m tall, with a circular cross section with
a diameter of 1-2.5 m (cupolas in the foundry industry
are often larger); see Figure 1. Preheated air is blasted
through a number of tuyeres (4-20) at the bottom of
the furnace, and coke and raw materials (rocks, bri-
quettes, limestone) are fed from the top. Some cupolas
are charged with mixed charges and others alternating
with coke and raw materials. However, because the coke
percentage is low (10-15%), the layers mix as they move
down through the cupola, and in practice, there is no
difference in the two modes of operation. The coke burns
at the bottom of the furnace, and the combustion
proceeds until oxygen is consumed approximately 0.5
m above the tuyeres. The hot combustion products flow
upward through the cupola, heating the raw materials
and coke, which move down during operation. Melting
of the rocks takes place in a zone about 0.5-1 m above
the tuyeres (the melting zone), and the melt is collected
in a melt bath. The level in the melt bath is maintained
by a simple siphon. The furnace is made of steel and is
cooled with water in a cooling jacket to prevent the steel
from melting. Evaporation of the cooling water in the
jacket circulates the cooling water from the jacket to a
tank above the cupola. Steam leaves at the top of the
jacket, and water from the tank enters the jacket at the
bottom. The cupola operates at approximately 10 kPa

Figure 1. Cupola furnace.
overpressure at the level of the tuyeres and a slight
underpressure at the top.
Coke is a fuel made through the pyrolysis of coal at
high temperatures (∼1000 °C) for several days. The
remaining char contains only very little volatiles.11 The
coke used in stone wool cupolas is foundry coke, i.e., coke
with an average diameter of ∼0.2 m. This coke typically
contains between 3 and 15% ash.
The raw materials fed to a mineral melting cupola
can be rocks, briquettes, and limestone. The rocks are
typically formed in volcanos and could be diabase,
gabbro, or basalt. The briquettes are made of various
minerals, such as olivine or basalt, diabase, and gabbro.
Waste stone wool can also be recycled in the briquettes.
Figure 2. Probes used to measure gas concentrations, temperature, and melt composition in the cupola.
Ind. Eng. Chem. Res., Vol. 42, No. 26, 2003 6873
The limestone is mostly added to adjust the viscosity of
the melt to the requirements of the spinning process.
The raw material particles typically have an average
diameter of 0.05-0.15 m.
Approximately 30% of the iron oxides in the minerals
is reduced to metallic iron in the cupola. The metallic
iron sinks to the bottom of the melt bath, where it
accumulates and must be tapped off before the iron level
reaches the siphon. The iron reduction is an undesired
reaction because metallic iron is a waste material, and
the production must be stopped two or three times a
day for iron tapping.
Experimental Section
In this section, four experiments are described that
aim at measuring the internal state of the cupola.
Measurements of the inputs and outputs of the cupola
are also described.
Gas concentrations of O2, CO, and CO2; melt composi-
tion; and gas and solid/liquid temperatures were mea-
sured using several different probes. Two types of probes
were used: a set of probes inserted horizontally through
the wall (wall probes) and probes inserted vertically
from the top of the cupola. The location of the melting
zone inside the cupola was explored through quenching
of an operating cupola by sudden purging with nitrogen.
Melt sample collection through the tuyeres and at the
siphon outlet was conducted to determine the location
of the iron reduction. The full-scale experiments were
performed on Rockwool cupolas.
Wall Probes. The gas concentrations, melt composi-
tion, and solid/liquid temperature in the hot part of an
operating cupola were measured using wall probes
located from 0.3 to 0.9 m above the tuyeres.
Procedure. To perform a measurement, the probe is
placed on a stand and driven horizontally into the
cupola with an electrical motor. The probe is then left
in the cupola for 1-2 min while the samples are
collected. A picture of the three types of probes is shown
in Figure 2. The gas probe is a steel tube with a tip
welded to the end of the pipe and with a hole on the
bottom side close to the tip through which the gas is
drawn. The temperature probe is a steel tube connected
to an IR pyrometer. The melt sample probe is a solid
steel rod with three pockets on the top side distributed
along the length of the probe. Five gates were installed
on the cupola at 0.3, 0.45, 0.6, 0.75, and 0.9 m. The
measurements were made at three radial positions: at
the wall, at the half-radial position, and at the center
of the cupola.
6874 Ind. Eng. Chem. Res., Vol. 42, No. 26, 2003
Figure 3. CO profiles at the center of the cupola measured with
the probes.
Figure 4. Average CO profile measured with the probes. (Vertical
bars indicate the standard deviations of the measurements.)
Results. Gas Concentrations. Figure 3 shows the
results of the CO concentration measurements made at
the center of the cupola plotted against vertical position.
The measurements were made on several days. The
concentration is less than 1% at 0.3 m above the tuyeres
and then increases to between 5 and 15%. The mea-
surements are very scattered, even though the cupola
operation did not change significantly between the
measurements on the individual days. The concentra-
tions of O2 and CO2 are not shown because they reflect
the same information as the CO concentrations.
For clarity, the average of the CO concentration
measurements at each position was calculated. Figure
4 shows these results as a function of the vertical
position at the wall, halfway from the center, and at
the center. The reaction between CO and oxygen is rapid
at high temperatures; hence, CO and oxygen cannot
coexist in the hot part of the cupola if perfect mixing is
assumed. The CO concentration is therefore most likely
very close to zero until the oxygen has been consumed.
Reactions in the probe might disturb the measurements,
i.e., CO and oxygen that might coexist because of
nonideal mixing might react in the probe. Above the
position where all oxygen is consumed, the CO concen-
tration increases as a result of gasification. At gate 1,
0.3 m above the tuyeres, the CO concentration is close
to 0%, indicating that the oxygen level has just dropped
to 0% (however, this point is based on only one mea-
surement and is therefore associated with higher un-
certainty). At gates 4 and 5, the results indicate that
the gasification rate is significantly reduced, probably
caused by the decrease in temperature. Figure 4 shows
that the concentration of CO rises faster at the center
than toward the wall, indicating that the gasification
of coke is fastest at the center. The difference in
Figure 5. Temperature profile measured with the probes. (Verti-
cal bars indicate the standard deviations of the measurements.)
gasification rate between the wall and the center is
probably caused by the lower temperature at the water-
cooled wall.
Temperature Profiles. Figure 5 shows the average
temperature profiles. The temperature is between 1400
and 1650 °C at 0.45 m above the tuyeres and then
decreases upward through the cupola. At 0.9 m above
the tuyeres, it is between 1250 and 1400 °C. There are
two points in the plot that deviate from the expected
behavior. At gates 4 and 5 (0.75 and 0.9 m above the
tuyeres, respectively), the temperature is rising at the
wall. Also, it is surprising that the temperature is lower
at the center than halfway to the center. Two factors
can be suggested to explain these observations: (1) The
high rate of gasification of coke due to less oxygen at
the center requires energy, thereby lowering the tem-
perature. (2) The high melting rate at the center leads
to faster movement of new colder raw materials down
through the cupola.
In addition, the measurements represent a mix of the
temperatures of coke and raw materials in both solid
and liquid form, which makes the interpretation of the
temperature measurements more difficult.
For all measurements, there were large variances or
disturbances. The difficulty of obtaining reliable tem-
perature measurements using the wall probes might be
associated with the fluctuations in the temperature
readings, which indicate the presence of high-frequency
noise. To account for this noise, a reliable filtering
procedure should be used in future measurements
rather than visual averages. The plots show only an
average of the measurements, which is a rather crude
way to treat the data, as the precise conditions of the
operation were not identical during all the measure-
ments. The operation changed mostly in terms of the
melting rate, i.e., the charge composition was relatively
unchanged, but the blast air flow rate changed. Fur-
thermore, a cohesive zone might influence the measure-
ments, even though the extent of the cohesive zone, and
thus the importance of the so-called windows, is smaller
than that found in foundry cupolas as the raw materials
melt over a wide temperature range than iron. Never-
theless, the average values presented here still provide
valuable information.
Sampling of the melt showed that the melting zone
extends as far up as over 0.9 m above the tuyeres, but
the probes cannot determine the height at which all of
the raw material has melted. The melt composition
results are described in the following section.
Iron Reduction. Iron oxides are present in most raw
materials and are essential for stone wool properties,
 Ind. Eng. Chem. Res., Vol. 42, No. 26, 2003 6875

Table 1. Chemical Composition (in % w/w) of the Melt Samples Collected to Investigate the Iron Reduction in the
Cupola Furnace
composition excluding the free iron
samplea Fe SiO2 Al2O3 TiO2 FeO CaO MgO Na2O K2O P2O5 MnO
siphon 0.21 42.6 19.2 1.6 5.9 19.7 7.8 1.8 1.0 0.3 0.1
tuyere 3.6 43.5 18.3 1.7 7.5 17.5 7.7 2.0 1.0 0.5 0.1
gate 2 0.4 40.2 18.2 1.6 10.4 18.4 7.4 1.9 1.3 0.4 0.1
gate 3 0.7 40.5 18.0 1.6 9.4 19.3 7.4 1.9 1.4 0.5 0.1
gate 4 0.6 37.5 17.1 1.5 15.1 18.0 6.7 2.0 1.5 0.5 0.1
gate 5 7.2 42.0 20.6 1.8 3.9 20.1 7.3 2.1 1.6 0.5 0.1
a Probe samples are from a different cupola than the tuyere and siphon samples. The gate positions are as follows: 2, 0.45; 3, 0.6; 4,

0.75; and 5, 0.9 m above the tuyeres.

because they provide fire resistance to the wool. In the

cupola, approximately 30% of the iron oxides are
reduced to metallic iron. The metallic iron is immiscible
with the glass melt and sinks to the bottom of the melt
bath at the bottom of the cupola. As a result, the level
of metallic iron in the melt bath increases during
operation, and the metallic iron must be tapped out of
the bottom of the cupola before it reaches the outlet of
the siphon.
As a first step toward understanding the iron reduc-
tion, which is a prerequisite to avoiding or reducing it,
melt samples were taken out through the tuyeres and
from the siphon. The chemical compositions of the
samples were analyzed using a magnet to determine the
content of metallic iron and X-ray analysis to determine
the remaining composition. Melt samples from the probe
measurements described above were also analyzed. The
values were averaged over two samples for the siphon
and the tuyere samples. The results for the probes are
averages of the analyses of six samples, two at each of
the three radial positions of the pockets in the probe.
Results. The results of the chemical analyses are
shown in Table 1. The analyses show that the composi-
tions of the samples from the tuyeres and from the
siphon are similar when the metallic iron is removed
from the tuyere sample. Slightly higher contents of SiO2
and FeO were found in the tuyere sample, whereas the
content of Al2O3 was somewhat lower. The reason for
the observed differences might be that the melt was not
well-mixed above the melt bath, but iron reduction in
the melt bath might also have an effect. However, it
appears that most of the iron reduction occurs above
the melt bath.
The compositions of the melt samples taken with the
probes show that some metallic iron is formed at all of
the positions investigated. The amount is, however, only
slightly above that in the siphon, except for probe 5. The
high content of free iron in the sample from gate 5 was
actually found in only one of the five samples (not six
because one of the pockets was empty after one of the
probings), while the remaining values were below 1. It
is most likely that the one measurement with the
extremely high content was an unsuccessful measure-
ment because it was seen only once. This means that
the major part of the iron reduction probably occurs
below 0.45 m above the tuyeres, which is the lowest
position of the wall probes used for collecting melt
samples.
The conclusion from these measurements is that the
iron reduction occurs from the melt bath and ap-
proximately 0.5 m up. Reduction above 0.5 m above the
tuyeres can be neglected, and the iron reduction in the
melt bath is limited and can probably be neglected. The
Figure 6. Gas temperature profiles measured on two different
days with the top probe and the temperature profiles obtained with
the wall probes.
results are surprising as the reduction was expected to
takes place in a zone with a high CO concentration and
low O2 concentration, but the measurements show the
opposite. This indicates that the mechanism of iron
reduction in stone wool cupolas involves reaction of the
molten iron oxide with the coke surface, which is the
only reducing reactant in the zone from the tuyeres and
0.5 m up.
Top Probes. Procedure. In addition to the measure-
ments at the wall probes, the temperature was also
measured using a 6-m-long steel probe that was inserted
from the top of the cupola a short distance into the rocks
and coke packing. The burden of the charge then pulled
a probe with a thermocouple inside it downward.
The top probe was inserted near the wall, i.e., less
than 0.3 m from the wall (cupola diameter was 1.8 m).
The results are, therefore, affected to some extent by
the water-cooled wall. The precise location of the probe
tip cannot be determined as it is the charge that pulls
the probe down through the cupola and the flow pattern
of the charge is not known.
Results. The temperature profiles are shown in Figure
6 for two experiments on the same cupola. The temper-
ature profiles show that, 1 m above the tuyeres, the
temperature is approximately 1300 °C. The temperature
then decreases almost linearly to approximately 150 °C
at the top of the cupola. The two profiles are very
similar, and the difference between them is probably
due to slightly different paths that the probe traced
through the cupola. The temperature profiles obtained
with the wall probes are also plotted in Figure 6. The
profiles of the different probes are in qualitative agree-
ment, but the wall and top probes were used on different
cupolas, and the temperatures obtained with the wall
probes are for solid/liquid phase, so the temperature
measurements are not immediately comparable.
6876 Ind. Eng. Chem. Res., Vol. 42, No. 26, 2003
Figure 7. Cross section of the open quenched cupola. (Picture height and width are 2.5 and 1.8 m, respectively.)
Quenching of a Cupola. Given the high tempera-
tures and the tight packing of the coke and rocks, it is
difficult to obtain an overview of the conditions inside
the cupola in terms of parameters such as the bulk
density and location of the melting zone. Therefore, an
operating cupola was quenched to obtain such an
overview.
Procedure. The quenching of the cupola was carried
out by interrupting the blast air and injecting nitrogen
to stop the combustion and other chemical reactions and
to cool the furnace. The nitrogen was injected from the
top of the cupola to prevent the nitrogen from gaining
heat from the hottest region and then continuing to melt
the rocks above even after the combustion was stopped.
Just before the quenching, the cupola was tapped for
all iron and melt in the melt bath, and the hole in the
bottom was used as the outlet for the nitrogen. The
injection of nitrogen was continued until the tempera-
ture in all regions in the cupola was below the self-
ignition temperature of the coke (600 °C). Subsequently,
the cupola was allowed to cool to room temperature over
several days. The void in the cupola was then filled with
cement, and when the cement had cured and dried, the
large lump was cut open to provide a picture of the state
inside the operating cupola. The quenched cupola was
charged with basalt, briquettes, and coke during the
period before in the quenching.
Results. A cut through the cupola is shown in Figure
7. It can be seen that the melting zone is located slightly
higher near the tuyeres than near the center. The extent
of the melting zone is not distinctly related to the
position, i.e., near the wall or the center. The briquettes
melt over an interval of 0.35 m, from 0.4 to 0.75 m above
the tuyeres, and the basalt melts over an interval of 0.25
m, from 0.45 to 0.70 m above the tuyeres. Figure 8
shows an example of a basalt rock in the melting
process. A droplet was frozen just before dripping off
and was captured in the concrete. Another indication
of melting is when the surface edges are not sharp and
bubbles are present.
There is evidence of iron accumulation in the melt
bath caused by iron that solidified during the week of
production. Apparently, the iron never melted again,
thus decreasing the volume of the melt bath and thereby
increasing the need for more frequent tappings. The
accumulation is lowest near the tapping holes and
highest furthest away.
The melt zone is located a little higher near the wall
than near the center. This is surprising because it was
expected that the furnace would be hottest at the center
furthest away from the water-cooled wall. The extent
of the melt zone is thus apparently not dependent on
the radial position (i.e., whether at the wall or at the
center).
An impression of the bulk porosity of an operating
cupola can be obtained from the cross section of the
quenched cupola. Prior to the quenching, the bulk
porosity was estimated to be 0.4-0.5, which the picture
in Figure 7 supports. The picture has not been system-
atically analyzed, e.g. by image analysis, so here, it can
only be concluded that there is no apparent contradic-

Figure 8. Basalt droplet captured in the quenched cupola. (Total picture width corresponds to 70 mm.)
tion between the perception of the bulk porosity before
and after quenching.
The quenched cupola was charged in alternate layers
with coke and raw materials, but there is no evidence
of layers in Figure 7. This is not surprising given that
the layers charged were relatively thin, e.g., the coke
layer could barely cover the cross section even if it were
perfectly distributed.
The interpretation of the quenching experiment should
be taken with some caution, as the quenching did not
happen instantaneously. It took at least 15 min before
the nitrogen injection was working properly, and after
this initial period, the entire melt was, of course, not
frozen instantaneously. In this period, part of the melt
might have escaped to the melt bath, in which case the
visible cross section would not represent the true picture
concerning the melted fraction in the operating cupola.
Measured Input/Output Data. Some of the input
to and standard output from the cupola were measured
at several occasions in connection with development
Ind. Eng. Chem. Res., Vol. 42, No. 26, 2003 6877
projects in Rockwool. The input parameters include the
amount and type of coke and raw materials charged and
the amount and properties of the blast air. The output
data are the temperature of the melt, the temperature
of the flue gas, the CO content in the flue gas, and the
cooling water loss. To determine the CO concentration
in the flue gas, the O2 concentration must also be
measured to determine the amount of false air that
dilutes the flue gas. The gas concentrations were
measured with an IR analyzer for CO2 and CO, and the
O2 was measured in a chemical cell (O2 and CO2 were
used to check the oxygen balance). The flue gas tem-
perature is measured with an ordinary thermocouple.
The temperature of the melt was regularly measured
manually with a thermocouple that was placed in the
melt jet as it left the siphon. The cooling water loss (i.e.,
the amount of water evaporated from the cooling water
tank) was not measured continuously, but could be
measured manually in terms of how much water was
added into the cooling water tank.
6878 Ind. Eng. Chem. Res., Vol. 42, No. 26, 2003
Table 2. Measured Melt Temperature (°C), Flue Gas
Temperature (°C), CO Content in the Flue Gas (%), and
Cooling Water Loss (MW) for Five Different Sets of
Cupola Operating Conditions
operating coke cal vala Vblast/ Tm Tflue COflue CW
conditions (%) (MJ/kg) Vblast,max (°C) (°C) (%)
1 13.5 28.465 0.802 1501 165 6.4
2 16.1 31.194 0.802 1487 158 9.6
3 17 31.194 0.813 1481 176 10.9
4 15 30.721 1 1501 195 6.5
5 14 30.721 0.879 1485 175 6.6
a Calorific valve.
The uncertainties of the measurements are estimated
as follows, based on many years of experience: (i) melt
temperature, (20 °C of ∼1500 °C; (ii) flue gas tem-
perature, (2 °C of ∼150 °C; (iii) CO in the flue gas,
(0.5% of ∼5; and (iv) cooling water loss, (0.1 m3/h of
∼4 m3/h.
The melt temperature is difficult to measure, because
the surface cools when the melt leaves the siphon. The
melt is very black, and thus, there is heat radiation loss
almost only from the surface and not from deeper in the
melt. This results in significant temperature gradients;
hence, the value read from the instrument is very
dependent on where in the melt the thermocouple is
placed. The flue gas conditions were measured where
the flue gas should be well-mixed and fluctuations in
the measurement are small. The cooling water loss is
constant (at least when averaged over 1 h), and thus
the precision is good.
The flue gas temperature is between 100 and 200 °C,
which is a temperature range in which the temperature
can be measured very accurately. However, the tem-
perature is measured in a relatively large duct, so the
flow profile and heat loss can affect the precision of the
measurement. Also, false air will disturb the tempera-
ture measurement. Measurement of the CO concentra-
tion in the flue gas is also disturbed by false air.
The precision of the cooling water loss measurement
is affected by the heat loss from the pipes and tanks,
etc., in the cooling water system. The heat loss from the
surfaces was estimated using heat-transfer correlations
from ref 14, and it was found that approximately 1%
was lost this way. The total amount of new water filled
into the tank was recorded. The measurements usually
proceeded for more than 6 h, with the meter indicating
how much water had been added into the cooling water
tank being controlled regularly (i.e., once per hour), and
the average was then found.
It was seldom possible to close the energy balance
from the measurements. Typically, the energy measured
in the output streams was less than the energy supply
in the coke and the hot blast air. The reason for the
discrepancy is not clear, but can partly be attributed to
the uncertainty in the blast air flow rate. This value
affects the calculated energy in the flue gas, both
thermal and as CO. Other error sources include the
leaks in the system that are known to exist (the O2
measurements can only discover leaks in positions with
underpressure).
Results. The results of the input/output measure-
ments are shown in Table 2. The results provide
valuable information for the development a mathemati-
cal model of a mineral melting cupola, as presented in
the second part of this work.13 In themselves, the
measurements contribute little to the understanding of
the cupola.
Discussion of Experimental Results. The experi-
ments performed in this work to map the conditions
inside a mineral melting cupola furnace have provided
new information. The concentration and temperature
profiles inside the cupola have been measured, the
(MW) location of the melt zone has been determined, and the
1.64 location of the iron reduction has been indicated. These
1.34 results provide a better understanding of cupola opera-
1.52 tion.
1.78
1.61 The wall and top probe measurements show behavior
similar to those of foundry cupolas, blast furnaces, and
fixed-bed coal gasifiers. The measurements presented
in refs 1-4 show that oxygen is rapidly consumed and
then the CO concentration increases to a steady level
when the temperature has decreased so much that the
gasification reactions have virtually stopped. Also, the
temperature profiles found in gasifiers, blast furnaces,
and foundry cupolas qualitatively compare with the
profiles found in this work.
The location of the melting zone was indicated by both
the wall probes and the quenching experiment. There
is qualitative agreement, but quantitative differences,
between the two experiments, as the quenching indi-
cates that the melting zone lies below 0.75 m above the
tuyeres, but melt was collected in the wall probes 0.9
m above the tuyeres. The two experiments were per-
formed on different cupolas, so different operating
conditions could explain the discrepancy. It is also likely
that the position of the melting zone in the quenched
cupola moved downward somewhat during the tapping
and/or during the time when the nitrogen injection was
not working properly.
The results of measurements on full-scale cupolas
should be interpreted with caution as it is difficult to
determine what the exact conditions were during the
measurements. Also, the cupolas are not designed for
measurement purposes but for production, so no effort
was made, for example, to prevent air from coming in
through the charging system every time new raw
materials and coke are charged, and this disturbs the
flue gas measurements.
Conclusion
In situ measurements have been performed to inves-
tigate the internal conditions of a full-scale operating
cupola. The measurements provide new insight. Four
types of experiments were conducted: (1) Gas concen-
tration, solid-/liquid-phase temperature profiles, and
melt composition in the zone from the tuyeres and 1 m
up were measured by probing the cupola through the
wall. (2) Gas temperature from 1 m above the tuyeres
and up was measured with a probe inserted from the
top of the cupola. (3) Quenching of the cupola provided
insight into the state of the cupola charge and the
location of the melting zone. (4) Measurements of input/
output streams from the cupola provided an energy
balance that could be used for parameter estimation in
the model presented in the second part of this work.13
The following conclusions can be drawn from the
experiments:
(i) The oxygen is consumed 0.3 m above the tuyeres.
(ii) The temperature is between 1400 and 1650 °C at
0.45 m above the tuyeres and then decreases upward
through the cupola. At 0.9 m above the tuyeres, it is
between 1250 and 1400 °C.
(iii) Iron oxides are reduced on the coke surface in
the zone from the tuyeres and 0.5 m up.

(iv) One meter above the tuyeres, the temperature is
approximately 1300 °C. The temperature then decreases
almost linearly to approximately 150 °C at the top of
the cupola.
(v) The melting zone is located from 0.35 m above the
tuyeres to ∼1 m above the tuyeres.
(vi) The melt zone is located a little higher near the
wall than near the center. This is surprising because it
was expected that the furnace should be hottest at the
center, i.e., the position furthest away from the water-
cooled wall.
(vii) The bulk porosity is 0.4-0.5.
(viii) Charging the cupola with alternating layers of
coke and raw materials is no different than supplying
it with mixed charges.
The experiments with the probes and the quenching
provide new qualitative and quantitative knowledge
about operating mineral melting cupolas. The measure-
ments provide a good basis for calibrating and validating
a cupola model in the second part of this work.13
Note Added after ASAP Posting
This article was released ASAP on 10/03/03 with the
incorrect title. The correct version was posted on 11/21/
03.
Acknowledgment
This work is part of the research program of Rockwool
International A/S carried out in cooperation with CAPEC
(Computer Aided Process Engineering Centre) and
CHEC (Combustion and Harmful Emission Control) at
the Department of Chemical Engineering, Technical
University of Denmark. The project is funded by Rock-
wool and Erhvervsfremmestyrelsen (Danish Ministry of
Ind. Eng. Chem. Res., Vol. 42, No. 26, 2003 6879
Business and Industry), and the industrial Ph.D. pro-
gram is administrated by the Academy of Technical
Sciences.
Literature Cited
(1) Piwowarsky, E. Giessereiforschung 1973, 2, 1.
(2) Saxén, H. Steel Res. 1996, 3, 73.
(3) Viswanathan, N. N.; Srinivasanand, M. N.; Lahiri, A. K.
Ironmaking Steelmaking 1997, 24, 476.
(4) Hobbs, M. L.; Radulovic, P. T.; Smoot, L. D. Prog. Energy
Combust. Sci. 1994, 19, 505.
(5) Sarma, B.; Cramb, A. W.; Fruehan, R. J. Metall. Mater.
Trans. B 1996, 27, 717.
(6) Bafghi, M. S.; Fukuda, M.; Ito, Y.; Yamada, S.; Sano, M.
ISIJ Int. 1993, 33 (11), 1125.
(7) Paramguru, R. K.; Galgali, R. K.; Ray, H. S. Metall. Mater.
Trans. B 1997, 28, 805.
(8) Siddiqi, N.; Bhoi, B.; Paramgura, R. K.; Sahajwalla, V.;
Ostrovski, O. Ironmaking Steelmaking 2000, 27 (5), 367.
(9) Min, D.-J.; Fruehan, R. J. Metall. Mater. Trans. B 1992,
23, 29.
(10) Sasaki, Y.; Soma, T. Metall. Mater. Trans. B 1977, 8, 189.
(11) Leth-Miller, R.; Jensen, J.; Jensen, A. D.; Glarborg, P.;
Jorgensen, S. B. J.; Jensen, L.; Hansen, P. Ironmaking Steelmak-
ing, manuscript submitted.
(12) Leth-Miller, R.; Jensen, A. D.; Glarborg, P.; Jorgensen, S.
B. J.; Jensen, L.; Hansen, P. Thermochim. Acta, manuscript
accepted.
(13) Leth-Miller, R.; Jensen, A.; Glarborg, P.; Jorgensen, S. B.
J.; Jensen, L.; Hansen, P. Ind. Eng. Chem. Res. 2004, 43, 6880-
6892.
(14) McCabe, W. L.; Smith, J. C.; Harriot, P. Unit Operations
of Chemical Engineering (International Ed.), 5th ed.; McGraw-
Hill, 1993.
Received for review June 17, 2002
Revised manuscript received April 22, 2003
Accepted June 20, 2003
IE020449W