You are on page 1of 27

Scandinavian Journal of Metallurgy 2000; 29: 232258

Printed in Denmark. All rights reserved

Copyright C Munksgaard 2000

SCANDINAVIAN
JOURNAL OF METALLURGY
ISSN 0371-0459

Review Article

Heat transfer in steelmaking ladle refractories and steel


temperature
A literature review
Tom P. Fredman
Heat Engineering Laboratory, bo Akademi University, Biskopsgatan 8, FIN-20500 bo, Finland*

Key words: ladle, refractory, lining, steel, heat transfer


c Munksgaard, 2000

Introduction
The word ladle has been used in the English language since the 12th century to denote a deep-bowled, long-handled spoon designed to convey liquids,
or a resembling instrument [1]. Although this type of
device has been present in metallurgical operations
since the discovery of iron, functioning as a means of
transporting molten metal in the casting procedure, it
is only recently that attention has turned towards the
loss of heat from the molten contents of the ladle. This
is due to the rapid development of contemporary
steelmaking and new ladle designs.
The purpose of this paper is to review some of the
literature dealing with heat transfer in steelmaking
ladle refractories and mathematical models thereof.
The reviewed works are classified into experimentallyand theoretically-focused investigations. Related
topics, e.g., measurement and modeling of steel temperature in the ladle or tundish and temperature control in casting, are included. Most contributions to the
field are conference papers and plant-specific studies,
although there are a small number of refereed journal
papers and some scientific theses and course material.
In most steelmaking facilities, the temperature evolution of the heat during casting is determined by the
tapping temperature of the converter. Therefore, in or* e-mail: tfredman/abo.fi

232

der to control the casting temperature properly, it is


very important to predict the loss of energy from the
heat at all process steps from the converter to the casting machine. This requirement can be relaxed and the
operating conditions of the converter can be standardized by introduction of melt reheating equipment
along the process route to ensure a sufficiently high
casting temperature. Reheating should be done as
close to the caster as possible (e.g., by plasma or induction heating in the tundish), to minimize temperature variation during casting. However, as reheating
facilities tend to be energy and space consuming as
well as expensive, it is still adopted practice to use
tapping temperature as a control variable for the temperature trajectory at casting.
Traditionally, the estimation of heat losses was
done by the caster foreman on the basis of long-term
professional experience. Recent developments in continuous casting practice, multiple ladle operations
with limited transfer facilities, new steel grades, new
refractory materials and ladle/tundish designs and
man-power reduction due to automation of the processes have led to a need for more accurate heat loss
estimation. The molten metal looses heat during holding, transportation, pouring and stirring stages between the converter and casting machine. Cooling is
also performed, when necessary, by scrap additions
and at ladle treatment when adding alloying elements
to the melt. Modeling the influence of stirring, pouring and alloying on heat temperature is mostly done

Heat transfer in steelmaking ladle refractories

using empirical expressions formed on the basis of


previously obtained measurement data. First principles have mainly been employed to describe the
heat losses due to holding and transportation.
It is evident that more refined methods of heat loss
estimation will be required in future steelmaking as
ladle treatment processes will become more important
[2] and the casting temperature window for most steel
qualities will be narrower. Process logistics is also
likely to become more complex with higher demands
on productivity and energy economy.
There are a number of frequently-used decision
variables in temperature control, for cooling; idling
the heat, stirring and scrap addition, for heating; reheating if equipment for this is available, either chemically or electrically. If the temperature of a heat declines too rapidly during casting, the casting speed
can be increased. On the other hand, casting speed
can be lowered if there is insufficient shell thickness.
In severe cases, when it is anticipated that the heat is
likely to be interrupted, despite high casting speed,
the heat can be partly or entirely rejected and recycled
to the process as scrap. For the heat loss to the ladle
refractory, the most important factors for temperature
control are the thermal state of the lining at tapping
from the converter and the thermal properties of the
refractory. Continuous measurement of the thermal
state being inconvenient and difficult, estimation appears a possible application for first-principles
modeling. Other variables affecting heat loss are the
extent of refractory wear and scull (solidified metal
residue in the ladle). Heat loss is also a function of the
ladle holding time, stirring time, nature and extent of
ladle treatment (alloying etc.), slag amount and composition as well as casting time [3, 4].
Strategies for improving temperature control can be
divided into 2 approaches; the hardware and the software approach. Hardware improvement of temperature control includes use of insulating cover agents on
free slag surfaces to inhibit radiation and convection
heat transfer, ladle lids for the same purpose and for
decreasing radiation heat transfer from the free inner
wall surfaces of ladle lining. Another frequently-used
measure at revamping of steel plants is to increase
ladle capacity, reducing the heat loss per mass of
metal. Recently, new refractory materials have appeared on the market, giving improved mechanical
stability and improved wear resistance. Thus, more
heats can be cast with the same refractory and energy
savings due to fewer preheatings and thermal startups of newly assembled ladle refractories can be
achieved. Unfortunately, most of these new materials
store more energy and conduct heat better. This will

surely present a challenge to heat loss estimation and


is, in fact, the background to much of recent work in
this area. As mentioned, an effective but expensive
route to improved temperature control is installation
of reheating facilities. Since this requires reliable online temperature measurement for control of the heating input and major investments as well as increased
energy input it has not been very popular.
Software improvement of temperature control includes systematic heat loss estimation and process
scheduling, e.g., through use of expert systems, simulation and tuning caster operation for coherence to converter output. Attempting to estimate the heat losses
along the route between converter and casting machine
and to correct the influence of these on the casting temperature by altering the converter tapping temperature
resembles the feedforward approach of control engineering. However, there is also feedback represented by
the thermal state of the ladle brought to tapping.

Thermal control in continuous casting


In a series of articles, [57], Cramb outlines different
control strategies for continuous casting. Emphasis is
on time and temperature control. The author defines time
control as the problem of targeting each operational
step on a specified time plus or minus an acceptable
variation. Time control, in continuous casting, is dependent upon temperature control. To improve time
control, the temperature can be manipulated if the right
process hardware is available. There are 2 operating
strategies for achieving high productivity on casting
machines. One is to maintain steady state operation
through process consistency, reproducibility, statistical
analysis and research to deepen the understanding of
each process step. The alternative is biasing the operation to deal with poor coordination. An example of
this approach is use of ladle metallurgy as a buffer between steelmaking and the caster when synchronization between these is poor. Whereas the consistency
strategy requires extended research and development
of the process, the latter coordination approach emphasizes quick decisions. In this strategy, optimization of
steelmaking tends to become more important than that
of continuous casting. Temperature control in caster
operation has 3 objectives. Tundish superheat is to be
controlled, a sufficiently high heat content of the liquid
steel must be ensured to allow for the intermediate operations before the tundish and, finally, slab enthalpy is
to be maximized. These can be met by proper control
of liquid steel temperature, of which Cramb notes three
main strategies.

233

Fredman

O Tap every heat at the highest possible superheat,


allowing for delays etc., correct the temperature of
(cool) the heat at the ladle metallurgy station
O Tap every heat with the minimum possible superheat, which will enable meeting the aim tundish
temperature without correcting actions
O Tap every heat at a predetermined temperature, reheat or cool in the ladle or tundish if necessary.
The 1st strategy is typical for coordination time control, resulting in long ladle heat times and frequent
ladle cooling. Furnace and ladle refractory life is generally poor due to high tapping temperatures. Moreover, the use of scrap in the BOF is minimized, which
limits production. In the 2nd approach, energy cost is
minimal but extended knowledge of heat loss mechanisms is fundamental. No room for process irregularities can be allowed and the strategy is not very flexible. Successful implementation of this approach requires an on-line computer program for calculation of
furnace tap temperature based upon the caster scheduling. Strategy no. 3 maximizes furnace and refractory
life as well as simplicity of operation, however, costing more in installation and operation of special ladle
and tundish heating devices. Depending on the composition of the specific grade to be cast, optimal operation can be achieved with a mixture of the 2nd and
3rd strategy or, if reheating facilities are there, with
strategy no. 3. Relying completely on the 2nd strategy
is possible only with an advanced on-line computer
simulation system, properly tuned to the actual casting operation and accounting for all heat loss mechanisms of the molten metal during its residence in the
ladle and tundish. Accurate modeling of the preheating and waiting empty phases is also essential. In
any case, calculation of the temperature at tapping or
on exit from the ladle station must be done. This is
the topic of [7], where a simple flowsheet for the calculation is presented. As a starting point, the liquidus
temperature for the specific steel grade to be cast is
taken. To this the desired superheat in the tundish is
added, giving the aim tundish temperature. If the calculated heat losses during residence in the tundish
and at draining the ladle are added, the aim temperature on departure from ladle treatment is obtained.
Adding the standard loss during ladle treatment and
a calculated loss due to holding will give the aim steel
temperature on arrival at ladle treatment station. Finally, adding the tapping losses brings us to the predicted tapping temperature. Reliable estimation of the
different heat losses generally requires statistical
analysis of measurement data from the specific caster.
Some empirical knowledge can be omitted if relevant

234

theoretical knowledge is incorporated into models for


different heat losses. However, no matter how advanced a model, tuning to the actual process will always be needed. This is understandable considering
the large number of variables influencing the steel
temperature. Tap temperature estimation should be
based on timing, ladle and tundish lining status, projected ladle metallurgy operations and projected casting time. Furthermore, if the steelmaking shop is
closely synchronized to the caster, tap temperatures
can be lower than if there are bottlenecks in the process logistics. If so, steel reheating might be necessary
which, apart from easing temperature control, gives
the steelmaker the ability to recover cold heats and
allows for extended degassing and ladle metallurgy.

Experimental studies of steelmaking


ladle systems
Overview
Many of the reviewed articles with an experimental
approach to temperature control feature temperature
measurements in steelmaking ladles in circulation.
The most common method is to insert thermocouples
into the refractory lining at different distances from
the ladle bottom and hot face to monitor the temperature of the lining during the ladle cycle [816]. In an
early work, [17], the temperature of ladle hot face was
recorded by immersion of a thermocouple, in contact
with the refractory, into the molten metal. A non-contact method, use of a radiation pyrometer, is described
by Ettwig [18]. This method was used by various investigators [14, 15, 19]. In a few works [13, 14, 19] the
thermal stratification of the molten metal was investigated by immersion of a lance fitted with thermocouples at different heights. Other experimental techniques include tracer experiments to determine the
residence time distribution of the steel in the tundish
[12]. The tare weight of ladles can also be measured
systematically [19] to track sculling and refractory
wear, and tundish steel temperature may be measured
by thermocouple immersion [13, 14, 17]. A physical
modeling approach was adopted by Hlinka & Miller
[20], who simulated the filled ladle using a vessel
made from acrylic plastic containing hot water.

Reviewed investigations
Samways & Dancy [17] measured steel temperature
during normal operation of a 285 t open hearth and
an 84 t basic oxygen furnace in an ingot casting process. These measurements were done with Pt-Pt 10%
Rh thermocouples, calibrated against the melting
point of pure iron. Tap temperatures were recorded

Heat transfer in steelmaking ladle refractories

Fig. 1. Schematic of the ladle-tundish model. Experimental apparatus


used by Hlinka & Miller [20]. Courtesy of the Association of Iron and
Steel Engineers.

within 1 min of tap and teeming (casting) temperatures were taken by immersing a silica sheath slowly
into the teeming stream as close to the nozzle as possible. Ladle temperatures were monitored with a contact chromel-alumel thermocouple to give an indication of ladle brick temperature. Ladle additions,
tapping, holding and teeming times were also recorded in an effort to formulate a statistical model for
the overall drop in steel temperature between tapping
and teeming. The tapping time and temperature, steel

grade chemical composition, refractory thermal state,


holding and casting time and temperature were identified as the most important variables affecting the
temperature drop.
Data were gathered from 73 open hearth heats and
25 basic oxygen heats and a computer program based
on expressions linear in tapping, holding and teeming
times was designed. The effect of ladle additions was
estimated by considering different reaction mechanisms and their consequences for the temperature
drop. Ladle heat losses from conduction through the
walls and radiation as well as convection from the
melt surface were correlated with tapping, holding
and teeming times. Surface losses were considered
only during tapping because of the insulating effect
of the slag, reducing convection losses. An interesting
observation is the accelerated temperature drop at the
final stages of teeming, as the relative importance of
conduction losses through the bottom increased. Although the results may not be valid as such for modern continuous casting processes, the work is a good
example of how empirical models can be formulated
and evaluated.
Hlinka & Miller [20] developed a method for
modeling liquid steel-refractory systems by means of
hot water and acrylic plastic, see Fig. 1 for their experimental setup. This physical modeling approach
was tested at the Homer Research Laboratories of
Bethlehem Steel. The method allows both heat and
fluid flow phenomena to be studied simultaneously
and the observations can, with certain preconditions,
be scaled up to the ladle-tundish system used in
strand casting. The purpose of the work was to study
the effect of ladle and tundish preheat and casting
speed on steel temperature. As mathematical calculations gave misleading results, assumed due to thermal stratification in the molten steel, physical
modeling was adopted. Through dimensional analysis of the heat transfer from the steel to the refractory
it was found that containers made from acrylic plastic
containing hot water with a layer of mineral oil to
simulate slag could model the ladle and tundish. The
melt surface, of the model as well as the real system,
was considered completely insulated by the slag.
Temperatures were recorded at three positions in the
model system; by the ladle stopper rod on the bottom,
in the tundish and near the surface of the water in
the ladle. Tracer dyes were injected into the ladle to
visualize the convection currents. A marked difference was found between the fluid flow in a ladle with
thick and a ladle with thin slag, see Fig. 2. The temperature in a tundish teemed from a ladle with thick
slag leveled off and stayed almost constant during

235

Fredman

Fig. 2. Convection currents during teeming. Impact of slag layer on


stratification, according to the results of Hlinka & Miller [20]. Courtesy of the Association of Iron and Steel Engineers.

casting. Moreover, it was observed that stratification


in the ladle actually serves to even out the temperature trajectory of the teeming stream as the colder peripheral streams descend towards the nozzle and are
replaced by hotter steel from the bulk zone. This phenomenon was pronounced especially during prolonged casts and was attributed to the insulating effect of the slag layer. When water model observations
were scaled up and compared with measurements on
a 7.5 t heat made in the lab melt shop, agreement was
within 12C. The effect of preheat of ladle and tundish
was investigated by keeping hot water in the vessels
prior to the experiments and the results were scaled
up to 7.5 t and 75 t heats, respectively. Tundish leveloff temperature as fuction of casting rate was also obtained from the model trials and a slight increase in
tundish temperature was detected for high casting
speeds. A probable cause was thought to be the storage of heat in tundish walls/refractory at the initial
stages of casting. As a higher level-off temperature,
which occurred for higher casting rates, allows a
lower tap temperature, the model experiments confirmed that casting speed can be increased to manage
cold heats. Finally, scale factors were calculated from
similarity considerations of material properties, temperature, time, heat and fluid flow and geometry.
Ameling et al. [8] outline temperature control of
die Hamburger Stahlwerke GmbH. After the ladle
fleet had been revamped from schamotte to dolomite
linings with tighter requirements on temperature control, it was necessary to obtain more accurate on-line
process information. The ladles were permanently fitted with two thermocouples each, one measuring the
temperature of the steel shell safety lining interface
and one the temperature at the interface between the
working and safety linings. The difference between
these readings was used to quantify the thermal state

236

of the lining and fed to the process computer together


with other important variables, such as estimated
time to start of casting and amounts of ladle additions, to calculate the required tapping temperature.
Portable ladle covers were in continuous use and integrated lids were on trial. A prototype ladle had lining
bricks laid in a spiral like manner to prevent joint
opening and cracking. Although the response of the
temperature measurements to irregularities in the
ladle cycle can be questioned, the initial results from
the trials were characterized as promising.
Widdowson [21] presents a summary of methods
and aims for control of temperature and chemical
composition in the ladle. An assessment of existing
procedures and future prospects is given, with emphasis on composition chemistry. The benefits of inert
gas bubbling as a means of avoiding thermal stratification and minimizing idle time between gas stirring
and tapping from the converter are outlined on basis
of previous reports. As key variables for temperature
control Widdowson lists tapping time and temperature, ladle additions, scull, slag load, lining and melt
surface properties, holding and stirring times and tundish state.
Kitamura et al. [10] briefly report on a measurement
campaign with six thermocouples in a ladle lined with
roseki firebrick. The ladle was monitored during ladle
treatment, degassing and teeming. From the data, correlations were obtained for the temperature drop of
the molten steel as a function of time elapsed from
end of tap. Covered as well as uncovered ladles were
included in the study.
Saunders [16] gives guidelines for preheating and
thermal control developed after switching from fireclay to high-alumina ladle linings. The studied plant
had ingot as well as continuous casting process
routes. Considering the 25% increase in density and
the doubling of lining thermal conductivity, a measurement campaign spanning 80 heats for a ladle with
70% alumina lining was implemented. It was found
that to avoid excessive sculling and refractory damage
a revision of practice would be necessary. Preheat efficiency was described by considering waste gas temperature, excess combustion air and existence of heat
recovery. Saunders recommends that preheat facilities
be specified so as to deliver around 70% of the thermal energy to the lining required for steady state in
the nominal preheat time. The initial thermal state of
the lining was discovered to be a major factor affecting heat loss during the ladle cycle. Measurements indicated that the lining reached thermal quasi steady
state in five regular ladle cycles. A computer model
for the refractory thermal state was developed for use

Heat transfer in steelmaking ladle refractories

Fig. 3. Calculated and measured temperatures, by Saunders [16], as


functions of preheat time at different positions in the ladle lining.
Courtesy of the Iron & Steel Society.

in preheating control and estimation of steel heat loss


between tapping and casting. In Fig. 3, the temperatures at various monitored positions in the lining of a
ladle are compared with calculated results obtained
with the model. When the ladle cycle was regular,
both preheat and tap-teem simulations agreed well
with the measurements. The model was also used for
design purposes, to compare thermal efficiency of different refractory materials and to calculate the required thickness of insulating linings. Insulating
linings were found to be cost-effective in rapid continuous casting cycles where there is risk of exceeding
the recommended maximum ladle shell temperature
of 370C.
Rieche et al. [14] studied heat losses from the 220 t
ladles of plant no. 1 of Mannesmannrohren-Werke
AG. In particular, it was desired to determine the influence of thermal state at tapping and use of insulating covers on the steel temperature drop. The ladle
fleet had dolomite working linings and schamotte
safety linings with insulation bricks between the
safety lining and ladle shell. The ladle lining and steel
temperatures were monitored with thermocouples,
tundish steel temperature was measured with an immersion probe and the exposed hot face temperatures
with a radiation pyrometer. The measurements compared reasonably well with calculations of steady
state heat conduction through the ladle wall, despite
variation of thermal conductivity with temperature
not considered in the model and irregularities in the

ladle cycle. In Fig. 5, a comparison of the measured


and calculated temperature profiles is presented. As
preheating capacity was inadequate, the results
showed clear benefits of using ladle lids and insulating agents on the slag surface. After a disturbance,
such as prolonged waiting, it took at least two ladle
cycles to re-establish quasi steady state. When a newly
lined ladle went into service, approximately four ladle
cycles were needed to arrive at steady conditions.
This is illustrated in Fig. 6. The time from end of cast
to begin of tap and casting time were detrimental for
steel temperature drop in the next cycle. An interesting detail is the jump downwards in temperature at
the safety lining shell interface indicated by the
measurements (cf. Fig. 5). A possible cause of this is
lining separation caused by thermal expansion of the
shell. It was also observed how the temperature continued to decrease for a short while after tapping of
steel into the ladle had begun, (see Fig. 6 stage d) in
the ladle cycles. From measurement data, it was estimated that the steel cooling rate reached a stable
value of about 0.42C/min 50 min after end of tapping. The temperature measurements of stratification
in the molten steel gave a maximum difference between top and bottom of 14C, diminishing after gas
flushing for 1 min. When stratification re-occurred,
the temperature difference between top and bottom
was only 4C. Results from lance measurements intended for examination of stratification should, however, be viewed critically as immersion of the lance
always more or less disturbs the flow field, which in
this type of system is strongly coupled with the temperature field.
Cardouat [22] reports on how tighter requirements
on temperature control and lack of reheating equipment initiated investigation of heat losses at the Solmer works. Thermal development during ladle treatment, alloying, waiting periods and pouring were
studied. A model dependent on idle time of the heat
prior to casting, ladle treatment time and nature of
additions and tare weight of ladle before tap was formulated. No measurement results are given, neither
is the mathematics behind the model described. A
computerized system for ladle management, including facilities for follow-up on the use of ladle preheat
and maintenance, is reported to be under development.
Minion & Leckie [11] investigated the impact of a
novel ladle lid system on temperature control, using
thermocouple measurements from the ladle linings. A
new lid handling system was developed at the Stelco
Hamilton plant, featuring attachment of the lid to the
ladle with hinges, making overhead cranes for this

237

Fredman

purpose obsolete and bringing down uncovered time


to a minimum. The lids were removed from the ladles
only at tapping from the converter. Measurements indicated that the safety lining and shell reached thermal quasi steady state in a few cycles, the maximum
shell temperatures being 220C for a ladle with alumina working lining. An alternative cycling practice
with preheating instead of ladle lids was evaluated
and compared with the lid practice with respect to hot
face temperature, steel temperature drop, amount of
cooling scrap and refractory wear and durability. By
all of these criteria, the strategy using the novel lid
system was clearly superior to preheating. The variation in hot face temperatures after casting decreased
to 30C with the lids and this helped to reduce slag
line refractory joint opening and spalling. Thus, good
temperature control and ladle practice, apart from delivering ladles to tap in a sufficiently hot state and
using suitable lining materials, must also include use
and rational handling of ladle lids. The measured
steel temperature drops during the campaign were
compared with simulations done at the University of
Toronto and the calculations proved that lid use consistently lowered tapping temperatures. Moreover, a
flatter temperature trajectory of the steel during casting could be achieved.
Perkins et al. [12] summarizes temperature control
at the Lackenby works of British Steel plc. A variety
of experimental techniques were used, including thermocouple monitoring of ladle linings, tracer experiments to investigate tundish residence times and
physical scale modeling with mercury and water to
study thermal effects of gas flushing. The concepts
mean working lining temperature and cooling rate
of working lining were used for classification of the
thermal state of ladle linings into 7 disjoint classes.
Preheating rules for the different classes were formulated and a computer code for ladle scheduling was
developed. The computer program keeps track of
ladle availability, thermal status and identifies the
ones suitable for tapping. Results from the residence
time experiments were scaled up and an empirical
model for tundish output temperature was set up
with the help of calculated wall and surface heat
losses and a simplified model of ladle fluid dynamics.
Steel heat loss in the tundish was considered a function of residence time only. Unfortunately, initial difficulties blocked on-line use of the model in the process computer. The ultimate aim of the project was
design of an on-line system for temperature control.
It was discovered that improvements would be
needed in modeling heat losses during gas flushing
and in temperature measurement prior to tapping be-

238

fore the model could become reliable enough for online operation.
Thermal behavior of the refractories of the 300 t
andalusite-schamotte ladles in the Rheinhausen
works of Krupp Stahl AG was studied by Hoppmann
et al. [9]. The usual observation method, with thermocouple measurements over a campaign consisting of
several ladle cycles, was employed. The safety lining
and shell reached quasi steady state in four ladle
cycles after a newly lined ladle was taken from preheating into operation. As the working lining wore
down toward the end of the campaign, the heat loss
from the steel increased. Shell temperature never exceeded 250C when ladle lids were used periodically
during waiting times. Theoretical results for the refractory thermal state prior to tapping were obtained
by solution of the dynamic heat conduction equation.
Radiation losses from the slag surface of the steel were
calculated and partial solidification of the slag as well
as re-melting on stirring of the heat were considered.
The greatest discrepancy between theoretical estimates and measurements was observed at the thermocouple positions in the vicinity of the hot face, i.e.,
where the dynamics are most important, and the error
was estimated to be equivalent to a temperature difference in the steel of 0.8C. For preheating, it was
found that only 23% of the input energy to the gas
burner increased the energy content of the refractory,
0.3% was stored in the ladle lid, 6% was conducted
through the refractory walls and 70.5% was lost with
the combustion exhaust gases. Thermal behavior of
the andalusite-lined ladles was compared with the results from a similar study of 100 t dolomite lined
ladles. In a 60 min waiting period, when the steel in
the larger ladle cooled 72C (0.65C/min), the steel cooled 95C (0.85C/min) in the smaller one.
Christensen [23] assessed heat losses for two different ladle configurations, a tapping ladle lined with
carbon paste and fireclay brick and one lined solely
with fireclay brick. The distribution of the heat loss to
the surroundings between the slag surface (and cover
when in use), ladle bottom and sidewall is outlined
for these two systems, as well as losses at tapping,
refining, pouring and while the ladles stood empty.
Here, heat radiating from the top of the ladle
amounted to most of the overall heat loss. The temperature drop of the molten metal arising from conduction to the lining is not indicated. Only shell temperatures were measured and, from these, experimental correlations yielded estimates of the heat flux to
the ambient. Radiation losses were studied using an
interesting measurement setup consisting of a small
copper plate water-cooled on one side. Maintaining

Heat transfer in steelmaking ladle refractories

water flow through the cooling circuit at a constant


rate and measuring the in- and output temperatures
for the water gave a hint of the absorbed radiation
heat flux.
Petegnief et al. [13] report on experiments carried
out at the Hagondange and Neuves-Maisons plants of
the French steel industry. Although these plants had
ladle furnaces there was a need for improved temperature control. Temperatures were measured in the
refractory of a 135 t ladle over a campaign of 30 heats,
ladle stratification and tundish temperatures were
measured with lance and immersion thermocouples.
For comparison, one-dimensional temperature profiles were obtaine by solving the heat conduction
equation using a finite difference method. Agreement
with measurements was within 50C. When steel temperatures in the ladle were estimated from the tundish
measurements and the temperature drop was calculated, the heat balance model relying on calculated
temperature profiles failed. A probable cause of this
was temperature stratification and sculling. To study
stratification, the two-dimensional Navier-Stokes
equations were solved numerically for the flow field,
which was combined with the energy balance equation for the molten steel. Based on the theoretical calculations, minor adjustments could be made to the
heat balance model for steel in the tundish. When the
thermal effects of ladle lid use were studied, it appeared that lid practice would reduce preheating time
by 1 h 40 min for a natural gas burner consuming
180 normal m3/h of fuel. The information from the
experiments and modeling was used in an operators
guide for the ladle furnace at Hagondange. As a result
of the introduction of the new decision support, standard deviation in tundish steel temperature was reduced by 50%. It was concluded that good temperature control can be achieved with relatively simple
models, provided the ladle cycle is sufficiently regular.
Rutqvist et al. [15] found that the most important
factor influencing casting temperature is ladle thermal
state prior to tapping, provided the ladle cycle is sufficiently regular and rapid. For quantification of the
thermal state, measurement of the hot face temperature of the empty ladle was proposed. Remote measurement of this temperature was possible through the
use of a radiation thermometer (pyrometer). To investigate whether this type of measurement could be
used to unambiguously classify the possible internal
thermal states of the refractory, a thermocouple measurement campaign was undertaken. It turned out that
if, in addition to the temperature, its time derivative
would be known, the character of the internal thermal

state could be determined. Formally, this can be explained by considering the boundary condition at the
hot face; the heat flux is proportional to the surface
temperature. Hence, the temperature and its time derivative determine the heat flux and its rate of change,
characterizing the thermal state of the dynamic part
of the lining. The internal profiles of the refractory
were simulated for a set of disjoint classes made up
of different temperature and time derivative values
for the hot face. The simulations were verified by the
thermocouple measurements. For each class, an empirical model for the steel temperature drop was
formed on the basis of campaign data.
The main topic of Grip [19] is radiation pyrometer
measurement of ladle hot face temperature on transfer
from continuous casting and during preheating. Brief
summaries are also given of thermal stratification
studies with thermocouple lance, impact of gas stirring on stratification, development of a thermal model
for molten steel in the ladle cycle and preheat burner
control. Interpretation of measurement data was similar to [15] and the main objective was to control converter tapping temperature so as to minimize scrap
cooling and deviation from aim casting temperature.
The positions of the pyrometers were at the sanding
station in ladle maintenance, at the ladle track before
the converter and at the ladle burners. Hot face physical properties could vary substantially within the
same ladle. This was due to slag residue, sculls, wear,
refractory cracks etc., and could give a dispersion of
up to 50C in the measurements. The sanding procedure resulted in sand clouds interfering optically
with pyrometer operation, which could be avoided by
extending the period of measurement beyond the
sanding time and retaining the maximum temperature value. Problems with pyrometer overheating at
the ladle burners also occurred. Ladle sculling was estimated by monitoring the tare weights of the ladles.
However, inconsistent results were obtained due to
calibration differences between the overhead cranes.
It was decided to use the same overhead crane every
time when weighing a specific ladle, if possible.
Ettwig [18] discusses non-contact measurement of
temperatures on hot surfaces using a polarization pyrometer for wavelengths shorter than 700 nm (visible
light lies between 390 nm and 770 nm). Temperature
measurement with this method is possible to an accuracy of 10 K above 800 K. The greatest error source is
background (reflected) radiation from the examined
surface. There are a number of methods to suppress
the reflected radiation, such as water cooled deflection
devices or polarization of the measurement beam.
Water cooling has some drawbacks: energy is needed

239

Fredman

Fig. 6. Temperature trajectories at the measurement points in the ladle


lining during various process stages, by Rieche et al. [14]. Calculations
are indicated by broken lines and measurements by full lines. Adapted
from data in [14].

Fig. 4. Reflectances for radiation incident on a metal.

ized component goes through a minimum at an incidence angle (measured from the surface normal vector) of approximately 1.2 rad. This result of Ettwig
[18] will be derived here by considering the Maxwell
equations for the electromagnetic field, [24], from
which the reflectances for the polarization components of the incident radiation on the surface can be
obtained. Expressed as functions of incident and
transmitted angles (qi, qt), real (nR) and imaginary (nI)
parts of the index of refraction (n), it is seen that
R (qi, qt)

|n|2 cos2qi2nR cosqi cosqtcos2qt


,
|n|2 cos2qi2nR cosqi cosqtcos2qt

(1)

for the parallelly polarized component, and


R (qi, qt)

|n|2 cos2qt2nR cosqi cosqtcos2qi


,
|n|2 cos2qt2nR cosqi cosqtcos2qi

(2)

for the perpendicularly polarized one. For the index


of refraction, obviously
|n| n2Rn2I .

(3)

The relation between incident and transmitted angles,


considering air as the surrounding medium, is Snells
law of ray optics
Fig. 5. Temperature profiles in the ladle lining calculated and measured
by Rieche et al. Adapted from data in [14].

sin (qi) |n|sin (qt) .

to sustain the cooling, measurement errors are introduced due to cooling of the studied surface, reflection from the water cooling equipment onto the surface and lower reliability of the measurement. For reflection and absorption of incident radiation at a metallic surface, the reflectances for the parallelly and
perpendicularly polarized components behave differently. In fact, the reflectance of the parallelly polar-

Combination of (1), (2) with Snells law (4) yields the


reflectances as functions of the incident angle qi. Plotting gives (for nR 2.5 and nI 0.5) the result in Fig.
4, which is similar to that of [18].
As Ettwig [18] points out, obviously, the measurement angle for the pyrometer should be kept as close
as possible to the minimum point of (1) in order to
minimize errors due to background radiation. Another useful recommendation, given by [25], is that
care should be taken to maintain the same location of

240

(4)

Heat transfer in steelmaking ladle refractories


Table 1. Reviewed experimental works relating to ladle thermal state
Corresponding
author

Samways [17]
Hlinka [20]

Measurements
thermocouples
hot face

Ameling [8]

lining

Kitamura [10]
Saunders [16]

lining
lining

Rieche [14]

lining

Minion [11]

lining

Perkins [12]

lining

Hoppmann [9]

lining

Christensen [23]

shell

Petegnief [13]

lining

Rutqvist [15]
Grip [19]

lining

Ettwig [18]

pyrometer

Features
steel temp.

special
techniques

supporting
theor. model

remark

yes
phys. mod.,
scale-up
calc. of tap temp.,
from meas.

hot face

lining
thermal state
thermal
steady state
yes

ladle,
tundish

phys. mod.,
tracer exp.

ladle,
tundish
hot face
hot face

ladle

general
hot surface

the measurement point on the hot face of each ladle,


in order to minimize errors due to differences in surface conditions.

Summary of the experimental studies


The reviewed experimentally focused works relating
to thermal state of ladle systems have been summarized in Table 1. In the classification of the investigations, experimental setup (type of measurements)
and special features have been used as criteria. As
special features are considered both novel experimental methods and supporting theory, such as tracer experiments or a flow field model for the molten steel.
Further characterization of the work, such as its principal aim or an emphasized sub-topic, is included in
the remark entry.
From Table 1, it is observed that a large part of
the research is focused very much on applications and
results specific to one particular plant. Basic research
is only represented through [18] and [20]. This is
understandable, keeping in mind the high cost of obtaining extensive measurement data from which more
general conclusions can be derived. Producing results
for one specific plant is, however, fairly straightforward. A downside with this approach is the need for
renewed research effort in case of a major overhaul of

fluid-flow
study

radiation loss
measurement
coupled heat
mass transfer
monitor ladle
tare weight
measurement
focusing

aid design,
preheat eval.

eval. lid vs.


preheating

thermal state,
scheduling
lining
thermal state

1-D thermal,
flow field

optics

aid decisionmaking
steel temp.
model
minimize
meas. errors

the equipment, e.g., the ladle fleet or tundishes. On


the other hand, since good temperature control makes
steel production more cost-effective, it is often not desired to generalize the knowledge to the benefit of
competitors.
Lining temperature measurement having become a
well-established technique, research into fluid dynamics and thermal stratification of the molten steel has
gained popularity in recent years. The problems of
this area are related to the aggressive environment
and the fact that the measurement easily disturbs the
flow field. Furthermore, in some cases thermal stratification has been observed to be beneficial for steel
temperature control, giving a flatter temperature trajectory at the tundish exit valve. Possible areas still
open to research might be the change of thermal properties of lining materials and ladle scheduling. The
former topic would require integration of basic research of lining material chemistry with applied work
in the plant. For the latter, it has been known for some
time that optimal ladle scheduling is characterized by
regularity and good process flow. Frequently, though,
ladle scheduling is overridden by other factors in the
plant logistics and temperature control suffers due to
bottlenecks, e.g., in crane capacity. Consequently, improving process integration improves temperature
control.

241

Fredman

Theoretical modeling of steelmaking


ladle systems
Overview
When continuous casting processes were introduced,
estimation of temperature drop between tapping and
pouring (casting) by traditional methods, i.e., on the
basis of long-term foundryman experience, became
much more difficult. This motivated research to identify factors behind heat losses from ladles and tundishes and later to formulate theoretical models of these
systems for simulation purposes. Several theoretical
investigations of steelmaking ladle thermal state have
been done in the last five decades. Early contributions
include simulation of heat losses and refractory temperature profiles on an analog computer [26, 27]. Later
[2833], the heat conduction through the refractory
was calculated numerically. Some workers [28, 30, 34
36] used thermocouple measurements from the refractory and molten steel to validate theoretical results
and tune their models to consistency with empirical
data. Since all mathematical models for thermal conduction and convection contain a multitude of physical parameters, some of which are temperature dependent, model performance strongly depends on
choosing correct values. Tuning model output to
match thermocouple samples of refractory temperature profile is a multidimensional optimization problem. Frequently, this problem is ill-posed, and a
unique solution may not always exist.

Reviewed investigations
Paschkis [26] and Paschkis & Hlinka [27] investigated
temperature drop of steel in ladles between tapping
and pouring with the simulation model The Heat
and Mass Flow Analyzer implemented on an analog
computer. Two ladle sizes, of capacities roughly
equivalent to 70 kg and 7300 kg with thicknesses of
the firebrick lining of 2.5 or 7.5 cm and 15 or 25 cm,
respectively, were investigated. In the first study [26],
the metal in the ladle was assumed completely stagnant and only conduction of heat occurred in the bulk.
In the second [27], the molten steel was perfectly
mixed. It is argued that these two constitute limiting
cases for the temperature drop of the steel as the former gives an underestimate and the latter an overestimate of the cooling of the steel. For the completely
stagnant model case, part of the steel solidifies on the
hot face of the ladle, releasing its heat of fusion to the
refractory. This would imply that if ladle sculling
takes place, it is favorable to wait longer before casting, a conclusion that would be deemed absurd in any
steel shop. Consequently, the limit model results have

242

to be judged with reservation. It was also found that


proper preheating is essential and that for a ladle to
be well preheated not only the hot face temperature
but also the temperature profile in the refractory must
reach a high enough level. The thicker the lining, the
more can be gained by preheating as the storage of
heat increases. The same argument applies to lid use;
higher preheat and a thicker lid enable more energy to
be stored. There is a critical lining thickness, though,
beyond which no improvement in thermal economy
can be expected. On basis of the temperature drop for
both limiting cases, the prospects of estimating the
real temperature drop are discussed and it is concluded that this must be done on the floor based on
observations of the, admittedly, vague concept degree of agitation of the molten steel.
Alberny & Leclercq [37] developed a mathematical
model for the temperature evolution of steel in the
tundish. Insufficient thermal control at casting causes
problems when the superheat (difference between the
temperature of the molten metal and solidification
temperature) drops too low, as clogging of the teeming nozzles and sculling in the casting mold are then
likely. Furthermore, at low superheats the shape of the
teeming stream becomes incoherent, leading to increase in re-oxidation from the atmosphere and quality problems in the final product. Too high a superheat will cause break-out of liquid steel from the solidified shell after the strand has passed through the
mold in the casting machine. The solidification speed
was found to be proportional to the negative superheat. Thus, as the superheat increases, the solidification is retarded, increasing the risk of a break-out.
Another negative effect of high superheat is the resulting inferior structure of the solidified product, with
typical formation of internal cracks and axial porosity
giving poor product quality. The events in the tundish
being highly dependent on previous stages, such as
ladle initial thermal and physical state, holding time,
ladle metallurgy etc., the ladle was included in the
model. The model estimates the variation in steel temperature, through energy balance equations, at the
exit valves of both the ladle and tundish. Heat losses
through the overhead slag and refractory walls were
calculated by a finite difference method. For the slag
emissivity a value of 0.8 was used and thermal diffusivity of the refractory was approximated as constant.
From the appropriate balance equation an average
temperature of the steel could be obtained, considering a time dependence of the heat flux from the melt
to the hot face proportional to the inverse square root
of time.
The same time dependence for the heat flux is ex-

Heat transfer in steelmaking ladle refractories

hibited by a semi-infinite solid with prescribed surface temperature equal to one and uniform initial temperature equal to zero:
T(x,t)

exp (h2) dh ,

(5)

x/(2at

where the thermal diffusivity, a k/rcp, was introduced. The integral in (5) is customarily termed the
complementary error function at the position indicated
by the lower integration limit. Taking the spatial derivative to calculate the heat flux across the hot face,
yields

q(x0,t) k

T(x,t)
x

x x0

p at

exp

x20
.
4at
(6)

Assuming that the entire heat loss of the melt occurs


through conduction to the refractory and convection
from the slag surface, described by a constant term
Qconv, a simple energy balance for the steel has the
form
Cp

dTsteel (t)
q(x0,t)S0Qconv 0 ,
dt

(7)

where S0 denotes the area of contact between steel


and refractory. Integration leads to the form, (disregarding the exponential factor in (6))
Tsteel(t) Tsteel(0)AtBt .

(8)

This equation, with empirically determined coefficients A, B, has frequently been used in the industry
for statistical modeling of temperature drop of steel
in the ladle cycle, and is occasionally referred to as the
Yngve Sundberg equation.
Alberny & Leclercq [37] introduced the difference
between the mean steel temperature and that of the
pouring stream exiting the ladle in order to describe
stratification. Consideration of the convection currents
in the ladle in combination with tests gave an estimate
of the rate of decrease of the stratification temperature
difference. A number of simulations were carried out
with the model to set guidelines for good temperature
control. Here, the importance of preheating and suppressing stratification was illustrated.
Chone & Teyssier [38] outline a mathematical
model for ladle heat loss estimation developed at IRSID, Maizieres-les-Metz in France. The main objective
of the model is to aid evaluation of lining design alternatives and changes to practice, such as adding insulation layers to the refractory or the effect of prolonged holding times between consecutive heats. The
core of the model is the energy balance equation for

the liquid steel in the ladle. The different stages of the


ladle cycle, such as preheating, waiting prior to tap,
tapping, loaded waiting, casting and cooling can all
be described by altering terms and boundary conditions in the energy balance. Thermal impact of ladle
lid, covering substance and slag can also be simulated. The model simulates the temperature profiles in
the lining, lid and slag/surface layer, heat loss from
the tapping jet, metal/slag surface or through the lining and the metal temperature as functions of time.
Mathematically, the model is straightforward, the geometry is one-dimensional and the metal is assumed
to be perfectly mixed. Arising differential equations
were solved numerically. Heat losses due to more
complex phenomena, such as use of lid, covering
agents and argon stirring are simulated by altering
the waiting time so as to adjust the heat loss. Several
other factors are considered; convection boundary
layers, tapping jet shape, slag and metal surface properties as well as lining, slag and steel initial temperatures. To evaluate the model, temperatures at preheat,
end of tap and after 79 min of waiting were determined for a 7.5 t ladle. Good agreement was obtained
during waiting and casting, but discrepancies were
observed for the preheating and tapping stages. Possible reasons were incoherent tapping stream shape
and melt splashing. Model output was also compared
with measurements from a 60 t and a 240 t ladle. For
the former, agreement was satisfactory but for the latter the results were somewhat inconsistent. In this
case, the cause was thought to be melt stratification,
inevitable with the ladle capacity and residence times
involved. Finally, the authors remark that accurate
knowledge of lining thermophysical properties was
lacking, especially for the heat conductivity. It was
found that the best surface heat loss protection was
provided by a 10 cm thick slag layer (for casting times
shorter than 1 h). Conduction is most significant for
cold ladles and the heat flow is inversely proportional
to the square root of time, see eq. (6). Insulation materials in the lining were judged interesting if holding
times are long in comparison with empty times and if
the lining is thin or has high thermal conductivity. It
was noted that, for good performance, the insulation
must be installed within the dynamic part of the
lining.
Omotani et al. [39] adopted a similar function for
the heat conduction to the refractory as in [37], but
with a convection boundary condition at the hot face.
A model for steel temperature was formulated
through the heat balance equation, which was solved
numerically. Customary idealizations, such as uniform preheat state, no stratification and instant tap-

243

Fredman

ping were introduced. The ladle lining was thick


enough to keep the thermal field of the outermost
layers at quasi steady state, which was confirmed by
calculations. Heat loss from the slag surface was assumed to be directly proportional to the temperature
difference between slag and metal, when the slag
cover was thicker than 12 cm and consisted of carriedover furnace slag. The heat balance can be expressed
in terms of the first-order ordinary differential equation
dTsteel(t)
P(t)Tsteel(t) Q(t) ,
dt

(9)

with general functions of time P(t), Q(t). This equation


can be integrated after multiplication by an integrating factor of the form exp (tP(s)ds), yielding
t

(10)

Fig. 7. Steel temperature drop as a function of time for different ladle


opening configurations. Adapted from data in [31].

A number of examples were worked out with the heat


balance model. Typically, in a billet or slab continuous
casting operation, 5560% of the total heat lost is
stored in the ladle wall refractory, 1520% in the ladle
bottom and 2530% is lost through the slag. Smaller
ladles are generally more sensitive to changes in tap
temperature, preheating practice and type of refractory brick. It was demonstrated how the loss through
the slag layer increases slightly before commencing to
a very slow decline. Increasing tap temperature was
shown to be advantageous in emergencies with risk
of excessive sculling or nozzle clogging, but not as a
routine temperature control measure.
Pfeifer et al. [31] developed a thermal model for the
overhead (above the melt surface) part of the ladle.
In particular, the effects of changes to slag thickness,
stirring practice, ladle opening and lid geometry were
investigated. The mathematical model includes partial solidification of the slag, transient heat conduction
through the wall and lid as well as different practices
of gas stirring and two alternative designs of the ladle
upper section. Simulations for different ladle sizes
and geometries were carried out in order to compare
their energy economy. The temperature dependence
of the thermophysical properties was considered for
the wall but not for the lid and slag. The boundary
conditions on the (isothermal and grey) hot surfaces
are functions of the radiation from the slag surface.
After calculation of the view factors for the surfaces
and surface temperatures using the heat conduction
model with known surface heat flux, the surface heat
flux was updated using the radiation model by solving a system of linear equations. The conduction

equations were solved numerically by the Euler difference scheme. Two ladle designs with no stirring
and different slag thickness were compared, one had
the conventional cylindrical shape and the other had
a conical upper section, making the ladle narrower at
the lid than in the middle. The design alternatives did
not differ much in performance whereas a thicker slag
layer could reduce temperature drop by more than
10C for an 80 t ladle. On the other hand, when stirring with argon or nitrogen, use of a lid may be impossible, making the difference in thermal performance more significant. It was found that the largest
reduction of temperature drop is obtained with conical ladle head in combination with a lid, when possible. The steel temperature drop as a function of time
for the conventional and conical ladle heads are depicted in Fig. 7, from which a final difference of over
10C in favor of the conical form can be observed. In
comparison with a conventional ladle, the conical concept reduced heat loss from the steel by 16%. Especially with prolonged gas stirring and ladles of
smaller capacity, the conical ladle head was found
beneficial for thermal control.
In [32] Pfeifer et al. combined the model in [31]
with thermal models for the tapping stream, alloying
additions and transient conduction through the sidewall and bottom of the ladle in order to estimate casting stream temperature. The thermal model of the
tapping stream consists of an energy balance equation
for an infinitesimal length element of the stream, together with simple expressions for the average mass
flow and velocity of the steel exiting the ladle. Irregu-

Tsteel(t)

244

Q(h) exp ( P (s) ds) dh


.
t
exp ( P (s) ds)

Heat transfer in steelmaking ladle refractories

larity in the stream shape (deviation from circular


cross section) was considered through a correction
factor. When the ladle was filled, calculations were
done using the model in [31] as well as numerically
solved equations for transient heat conduction below
the slag surface. Alloying was considered by a simple
empirical expression. The thermal properties of the
lining materials were functions of temperature in the
explicit finite difference equations. The heat balance
was solved for the steel temperature by Eulers
method at tapping, holding and casting, considering
relevant loss terms for each stage. When the ladle was
empty, the thermal status was obtained using a procedure similar to that of [31]. Performance of the
model was tested for a 100 t ladle during a measurement campaign. In order to minimize temperature
drop, a low tapping height and short duration were
important as well as maintaining circular cross section
of the stream. A typical temperature drop in favorable
conditions was around 5C. Model temperature profiles in the lining agreed reasonably with measurements. Steel temperature measurements immediately
after tapping agreed with an average deviation of 7C
with model output over a campaign of 20 heats. A
better indicator of model accuracy is the overall temperature drop for a heat compared with measurements. The measurements averaged over 38 heats
gave a value of 70C as model predictions spread between 20C and 100C, the larger deviations corresponding to heats with greater amounts of alloying
additions. Finally, the authors outline a tuning procedure for the model on basis of measurements before
and after vacuum treatment subsequent to tapping.
Model steel temperature is here corrected by taking
over the measurement value. The average deviation
of model predictions from measurements at this stage
was 5.9C.
In [40] Pfeifer et al. report a model for preheating.
The work supports their previous work in [31, 32].
The mathematical model consists of an energy balance
between net energy per time released from combustion and energy flow to the lining sidewall, bottom,
the ladle lid and losses. The combustion gases were
assumed to be well mixed and uniform in temperature, all surfaces inside the ladle were isothermal and
combustion was complete, i.e., no combustible compounds remained in the exhaust gases. The balance
was solved numerically for the combustion gas temperature after having obtained the gas volume flow
from combustion chemistry calculations. Possible recuperation (preheating of the input air to the burner)
was also considered. Heat exchange between the combustion gases and inner surfaces of the ladle was de-

scribed by radiation and convection heat transfer.


Model calculations were done to compare performance of different fuels, such as blast furnace exhaust
gas, coke oven gas and natural gas. The influence of
combustion parameters like temperature and oxygen
content of combustion air and preheating time was
also investigated. For instance, high fuel supply and
short preheating time resulted in larger temperature
gradients in the lining as compared to preheating
using lower fuel supply and longer times. For a constant input of fuel, the steel temperature drop as a
function of preheating time exhibited a minimum.
This was because losses from ladle and lid shells increase with time when the overall temperature level
of the lining goes up. Combustion with excess of air
gave rise to a larger steel temperature drop than stoichiometric air supply, while combustion with pure
oxygen gave the lowest temperature drop. Using a recuperator could also be recommended. To validate the
complete model, consisting of the preheating model
coupled with heat conduction through the lining,
simulations were compared with thermocouple measurements from a ladle during drying of a fresh lining
assembly. Three thermocouples were mounted in the
bottom of the ladle, one in the safety and two in the
working lining. For the measurement point in the
working lining, the model estimates were consistent
with the results, but for the others an offset was observed. The simulated temperatures during drying
were consistently lower than the measured, which
might be due to a systematic measurement error, misjudgement of thermocouple positions or incorrect
values of thermal properties in the model.
Morrow & Russell [30] describe a thermal model
for ladle refractories and steel temperature. The basis
of the model is a finite difference solution of the onedimensional heat conduction equation at the bottom,
sidewall intermediate level and slag line of the refractory. Variation of thermal conductivity and specific
heat with temperature was considered in the calculations by fitting curves to data on these properties
and using the obtained equations in the finite difference procedure. To verify the model, a degassing ladle
was fitted with thermocouples in the refractory and
temperatures were monitored during preheat and
during the second and third heats on the lining. The
thermocouples were mounted 64 mm from the hot
face into the working lining and at each material interface for three height positions. For the steel temperature, considered uniform throughout the melt, the
model vs. measurement exhibited good agreement
and the largest discrepancy occurred at end of teeming when the mass of steel in the ladle was small. The

245

Fredman

lining temperatures, on the other hand, showed only


partial agreement. At the bottom the consistency was
fairly good during preheat, but deteriorated for a
period at start of the second heat. At the slag line,
consistency during the first heat was poor but improved on the third heat, whereas in the sidewall the
situation resembled that of the bottom. A possible
error source in the measurement procedure could be
heat transfer resistances at the material interfaces arising from imperfect contact, such as gaps, mortar residue etc. Morrow & Russell [30] used their model to
investigate whether insulation layers of different configurations could bring down shell temperatures of an
ingot teeming ladle (lined with 70% alumina brick
and a direct-bonded basic slag line) and compared the
results with those of a similar evaluation using a
steady state model. The steady state calculations
yielded higher temperatures throughout and particularly the slag line area was much cooler in the calculations with the dynamic model. Thus, when studying
different design alternatives for reducing safety lining
and shell temperatures, conclusions drawn from
steady state calculations remain valid if the dynamics
is considered. An exception is that steady state calculation indicates an advantage of placing the insulation
between the shell and safety linings over placement
between working and safety linings, due to lower heat
loss. This advantage is enhanced if a higher quality
safety lining material is used, permitting operation
over multiple campaigns on the working lining. The
dynamic model showed that the insulating layer was
more effective in reducing steel heat loss when placed
between the working and safety linings, which is in
agreement with [38]. However, costly replacement of
the insulation is then necessary whenever the working lining is renewed. As further applications of the
model, studies were done of the effects on steel temperature of lining material and wear, preheat time and
use of refractory lid versus insulating slag cover. The
greatest impact of preheating was observed during
the first hour. In a continuous casting operation, use
of insulating slag cover was found to decrease steel
heat loss especially during initial heats. However,
when the ladle was cycled, a refractory lid was more
advantageous.
Hlinka et al. [29] report a thermal model for ladle
linings and steel temperature drop, developed at
Bethlehem Steel. A previously developed computer
code for solution of the dynamic heat conduction
equation for a one-dimensional geometry was taken
as starting point. The ladle geometry is reduced to one
dimension by replacement with a slab, by considering
the ratio of the mass of steel to the contacting surface

246

area on the bottom and sidewall. The slab concept involves multiple layers, consisting of all refractory material layers, the ladle shell and the liquid steel. On
teeming, the ladle is drained layer by layer from the
bottom. This is modeled by reducing the thermal
diffusivity of the slab according to a scheme developed from experiments. If desired, stratification in
the liquid steel can be considered for by altering the
conductivity of the corresponding liquid steel layer
of the slab. Material properties were chosen according
to the various events in the ladle cycle. For instance,
tapping was simulated by introducing real coefficient
values in place of the infinite thermal conductivity
and zero specific heat used for the empty ladle. On
surfaces exposed to the ambient, the heat transfer coefficient included both radiation and convection, the
emissivity was 0.8 throughout and a view factor of
unity was used for the ladle shell and cover/slag surface. Integration of elemental view factors for the inside of the empty ladle yielded the average 0.5. In
each heat, the model is updated with the liquidus and
solidus temperatures, from which sculling is estimated. For the slag, the heat conductivity was manipulated in order to simulate heat transport across
the slag layer. No heat of solidification is liberated to
the system on freezing of the upper slag layer. Refractory lining materials are handled using their commercial trade names and relevant temperature dependent
properties are imported from a data base incorporated
into the model. The same applies to ladle geometry,
referenced through plant names. Cover use can also
be modeled and the initial temperature profile of the
cover can be saved for future use. A number of events
in the ladle cycle can be simulated after initialization
of the temperature field, including preheat/reconditioning, tapping with possible additions (considered
through their entalphies), stirring, teeming and slag
removal. Stirring is considered by increasing the internodal thermal diffusivity within the steel as a function of stirring time and intensity. The model was
used to investigate when quasi steady state is reached
after introduction of a cold ladle into the cycle and
what preheat level this state corresponds to. The outcome was that three cycles were sufficient to establish
steady conditions in most of the lining and that the
same state would be reached with 6 h of preheat with
a 980C gas flame. In addition, the benefits of slag versus ladle cover use were studied and company policy
was changed to use of both ladle covers and surface
insulation to improve temperature control.
Tomazin et al. [25] investigated alternative ladle refractories and cycle practice. In two plants of LTV
Steel in Aliquippa, PA and East Chicago, IN, ingot

Heat transfer in steelmaking ladle refractories

casting processes were revamped to continuous casting. As a result, the conventional firebrick ladle
linings could not withstand the higher temperatures,
more aggressive slags, ladle metallurgy and prolonged holding times. Steel cleanliness and desulphurization also suffered when firebrick was in use.
Thus, it was decided to phase out the firebrick linings
in favor of 70% Al2O3 linings with MgO slag lines for
both plants. Considering the higher thermal conductivity and specific heat as compared with firebrick, it
was necessary to study the thermal behavior of the
ladle and its impact on temperature control. Preheating facilities were installed at both works. One
ladle at Aliquippa was monitored during preheating
and cycle with 30 thermocouples, yielding data for the
development of a preheating practice and tuning of a
mathematical model. The mathematical model was a
numerical solution of the heat conduction through the
ladle wall based on an explicit finite difference procedure with temperature dependent refractory thermal properties. Steel temperature was assumed uniform throughout the melt and axial heat transfer in
the refractory was disregarded. Radial temperature
profiles were calculated at a number of heights in the
wall in order to consider the steel surface position at
each point of the ladle cycle. The bottom and slag
layer temperature variations were assumed to be
purely axial. Above the steel surface, view factors for
all surfaces were calculated for the radiative heat
transfer. The model was also used to evaluate measurement variables for observation of the heat content.
Apart from hot and cold face temperatures, the waiting time between heats was important for determining required corrections at tapping. Covers were
judged most important during casting and empty
waiting time, when the radiative heat loss is most significant. In some simulations worn ladles, when moderately preheated from cold, actually reduced heat
loss due to smaller heat storage. However, a worn
ladle in cycle (at thermal quasi steady state) had
larger heat loss than a freshly lined one.
Hoppmann et al. [41] present a process model for
calculation of the converter tapping temperature
when considering estimated heat losses in the ladle
and tundish as well as initial thermal state of the vessels and thermal efficiency of the previous heat.
Model input can be divided into two categories; the
precharge and the postcharge data. As precharge variables were considered the waiting and preheat times
before tapping, lid use, sculling extent and duration
of the previous heat as well as the age of the lining
(from which the remaining thickness was estimated).
Postcharge input consisted of the planned stirring

time and alloy weight, planned transfer and waiting


times and the casting time. In the output are indicated
the estimated temperatures at tapping and start of
casting, starting and ending times for tapping and
casting and the predicted stirring time. In addition,
the temperature profile across the lining and temperature drop in the ladle can be obtained at any time.
The heat loss calculations were based on experimental
data together with energy and mass balance computations. The temperature trajectory during casting was
estimated on the basis of initial cooling rate, duration
of cast, tundish preheat and number of heats cast. The
slag thickness in the ladle was estimated from data on
the oxidation and mean transferred weight of converter slag, heat loss through the slag was dependent
on thickness only. After charging the converter, the
tapping temperature was calculated iteratively with
the desired superheat at end of casting as an end criterion. Finally, a comparison between the estimated
tapping temperature and real process data was done,
showing an average difference of 5C with a standard
deviation of 8.3C.
In [35], Hoppmann et al. describe an on-line process model for the steel temperature during secondary
metallurgy integrated with estimation of the casting
temperature. The system consists of mass and energy
balance equations for the ladle, ladle furnace and vacuum degassing process. A numerical solution of the
heat conduction equation is incorporated into the
model. The state of the slag and heat loss due to alloying were considered using cooling coefficients and
deoxidation factors. At casting, the net heat flux out
of the homogeneous molten steel was expressed in the
form
q(t)

a exp (b t),
i

(11)

where the ai, bi were determined by fitting the heat


flux (11) to numerically calculated values from the online model. Eq. (11) was also used for extrapolation,
i.e., heat fluxes were calculated for times outside the
time range of the coefficients ai, bi. Expressing the net
heat flux as above leads to an energy balance equation
for the molten steel in the form
dTsteel

dt

c exp (b t) ,
i

(12)

which can be readily integrated to yield an expression


for the average temperature of the molten steel in the
ladle during casting

247

Fredman

b [exp (b t)1] .
ci

Tsteel (t) Tsteel (0)

(13)

In addition to estimation of the casting temperature,


the model was used for optimization of ladle furnace
operation. Tuning the on-line model was done off-line
and the outputs were compared with actual process
measurements. At casting, an accuracy of 4.7C could
be achieved, an excellent result in view of the measurement accuracy of approximately 5C.
Chapellier [42] describes temperature control at the
Sollac plant of Florange, France. The converter tapping temperature is calculated on basis of charging
parameters, aim temperature for oxygen blowing, slag
oxidation level etc. Because the plant has a ladle furnace facility, tapping temperature management is separated from temperature control in the ladle and at
casting. The tapping temperature is kept as low as
possible for a given steelgrade and ladle furnace operation is determined by the desired casting temperature trajectory. The casting operator gives the projected casting start time as input to a computer program which calculates a series of aim temperatures
for different points in time for the ladle furnace. The
code uses simple empirical linear expressions dependent on liquidus temperature, maximum superheat in
the ladle, foreseen casting duration and time elapsed
between the last temperature measurement in the furnace and start of casting. Cooling of the steel during
holding is estimated using a constant cooling rate dependent on ladle treatment time and length of the previous empty period. Raising the heat by 1C is equivalent to an energy input of approximately 100 kWh.
The practice used is designed to give a tundish superheat of 26C for a fresh ladle and 23C for a cycled
one. At a production rate of 32 heats a day, 6 ladles
were in cycle. It was estimated that holding a
coverless ladle for one hour costs 0.6 kWh/ton of steel
more than for a covered one and that insulation
linings save 2.0 kWh/ton. Ideal ladle cycle lengths in
the plant were 2.33.0 h and the actual lengths 3.75.0
h. Temperature control in the plant is reported to be
accurate to within the measurement accuracy, typically smaller than 4C.
Koo et al. [43] investigated stratification in the steel
during holding and the impact of Ar gas bubbling.
The heat loss from the steel to the lining was estimated by numerical solution of the heat conduction
equation under the assumptions of perfect mixing and
uniform temperature of the molten steel. Stratification
was modeled by setting up the necessary equations of
change; the continuity equation and the equations of
motion and energy, all with the radius r and the

248

height z (from the bottom) as independent variables.


Hence, all equations were axisymmetric. Turbulence
was considered with a ke model featuring equations
of turbulent kinetic energy and dissipation rate of turbulent energy. In the equation of motion, buoyancy
due to temperature gradients as well as Ar injection
were included. The boundary conditions were zero
heat flux at the steel-slag boundary, no slip and transient heat conduction (from the mentioned numerical
solution) at the hot face. The solution of this system
was carried out using the PHOENICS program package for computational fluid dynamics. For verification
of the heat conduction model, thermocouples were
fitted into the lining of a trial ladle. The measurements, however, appear to be lagged with respect to
the simulations, suggesting that the time constant of
the thermocouples has been disregarded. Simulations
were done to evaluate effects of holding duration, refractory wear and stratification during holding as well
as restratification after gas stirring. Each heat was
compared with a standard case, with respect to steel
residence time and waiting time. Corrections to the
standard case were handled by temperature compensating factors, introduced before tapping or during
ladle metallurgy. The most interesting results, however, were the stratification simulations, where a temperature difference between top and bottom of the
ladle amounting to 24C was obtained for a holding
time of 20 min. Bubbling the melt with Ar reduced
the maximal temperature difference to 3C, indicating
the advantages of mixing prior to casting. In Fig. 8,
the temperature and velocity fields as well as the distributions of kinetic energy are shown for a stratified
melt and an Ar-bubbled one. Supporting the work by
a heat balance for the steel would have given a nice
illustration of temperature trajectory in the tundish
and impact of gas bubbling on it.
Gaston et al. [44] present a model for predicting
steel temperature and thermal state of the tundish
during continuous casting. The theory and methods
are applicable also to the ladle, which is a similar system. The model consists of the usual heat balance
equations for the steel in the tundish while it is being
filled and while casting is in progress, including heat
losses due to radiation and conduction heat transfer.
Steel level variation in the tundish was assumed to be
linear with time during the filling stage. Heat losses
to the walls were calculated under the idealization
that the wall be a semi-infinite solid. Solution of the
coupled steel heat balance and heat conduction model
was carried out numerically. In order to test the sensitivity of the model to stratification, simulations were
compared with available experimental data, confirm-

Heat transfer in steelmaking ladle refractories


Fig. 8. Results of Koo et al. [43] on the influence of
mixing with Ar gas on stratification in the ladle.
Courtesy of the Iron & Steel Society.

ing the model results. Furthermore, to describe the


temperature of steel entering the tundish, a heat balance for the steel in the ladle was formulated. In this
case, thermal state of the ladle refractory was regarded stationary and the radiative heat losses were
ignored assuming that the top surface of the molten
steel was insulated. Experimental results from the

Fig. 9. Estimated steel temperature trajectories and measurements of


Gaston et al. [44]. Courtesy of the Institute of Materials.

plant General Savio, Aceria LD-SOMISA in Rosario,


Argentina were used for model validation. The steel
temperature was measured in the ladle before casting
and in the tundish with an interval of 10 min during
casting of two consecutive heats. The tundish was
cold before the first heat. Agreement between simulations and measurements in the tundish was within
10C, with overestimation in the beginning and
underestimation at the end of casting, cf. Fig. 9. Thermal stratification in the ladle was not considered: If
this had been done, the model would have produced
lower temperature estimates in the beginning, when
the colder bottom layers of steel are teemed from the
ladle.
Shklyar et al. [45] present a thermal model concept
for a steelmaking ladle. Dynamic heat conduction
through the ladle lining, heat transfer above the slag
surface (or in an empty ladle) and heat transfer
through the slag layer are considered in the model.
Other features include temperature dependent thermal properties, variation in the steel level, thermal effects of metallurgical additives, lid use and efficiency,
refractory wear and preheating. The ladle cycle is divided into stages: tapping, ladle metallurgy, teeming,
empty waiting time and ladle preheating. The lining
is divided into regions of constant surface temperature, where radial (sidewall) or axial (bottom) tem-

249

Fredman

perature profiles are computed. Moreover, perfect


thermal contact is assumed at the boundaries between
the lining layers and at the hot face. For the slag layer,
a conduction-convection equation with a constant
heat transfer coefficient is included. At the slag line,
the difference between average slag temperature and
the hot face temperature is used as driving force for
heat transfer. Preheating is also modeled, fuel energy
content and input as well as moisture ratio of the lining, combustion parameters and exhaust temperature
are considered in the heat balance. A few example
simulations are presented in the form of graphs depicting calculated hot face temperature as a function
of time for situations with lid use/no lid and gap between lid and ladle/absence of gap. In addition, the
calculated and set point fuel flowrate together with
calculated and set point hot face lining temperature
are given for simulated preheating of a 370 t ladle
lined with magnesia-spinel clay. No experimental
data, in support of the model, is presented.
Mucciardi & Grandillo [46] used a general-purpose
software application called FASTP (Facility for the
Analysis of Systems in Transport Phenomena) for
thermal analysis of a steelmaking ladle in cycle. This
software package has been developed at the Metals
Processing Centre of McGill University for simulation
of dynamic heat and mass transfer processes. The
code is based on an explicit finite difference method.
Simulation was carried out for two alternative refractory configurations, one of which had a layer of insulating tiles between the ladle shell and safety lining.
In order to consider the cooling of the upper hot face
during casting, the ladle was sectioned into 5 segments each with a capacity of 16 t, each requiring 20
min to empty. Hence, the uppermost segment had an
average contact time with the steel of 50 min and the
lowermost segment 130 min. As might be expected,
the temperature profile in the lining was clearly
higher for the insulated lining, where the boundary
between the safety and insulating linings reached a
quasi steady state temperature of almost 800C. When
the energy content of the lining was examined at different stages of the ladle cycle, significant drops were
observed during preparation and waiting. Equivalent
temperature drops for the steel arising from lining
storage and losses to the surroundings were calculated for the 2 lining types. The conclusion was that
the insulation had impact on only the losses to the
surroundings and none on the storage of heat in the
refractory.
Saha et al. [33] developed a mathematical model for
scheduling preheating of fresh dolomite refractory
linings. The authors recommend basic refractories,

250

such as dolomite, magnesite and chrome-magnesite


over conventional fire clay/high groghigh alumina refractories. Ladle linings in modern steelmaking must
withstand high tapping temperatures, aggressive slags
and injectants, extensive stirring by gas injection, thermal cycling, vacuum conditions and prolonged holding times. Basic bricks, having better thermodynamic
stability than aluminosilicate materials, are thus more
suitable for ladle linings. However, basic bricks require
different handling and preheating. Being hygroscopic,
these have to be handled dry throughout and the safety
lining must be heated to remove all moisture before the
working lining is assembled. Prolonged preheating is
not advisable, since this will cause bond degradation in
the working lining. Hence, the hot face temperature
must be increased to 9501100C during the first 2 h of
preheating and kept at this level for no longer than 12
h. Shell temperature should be kept below 250C as
compared to 350C for high alumina refractories. To
meet these specifications, a dynamic heat flow model
was formulated and solved using finite differences.
The temperature evolution of the ladle lining can be
simulated for various fuels, fuel rates and refractory
configurations. A number of simplifications were
made, first of all the geometry was reduced to one dimension and the tar dolo ramming mass used to bond
the dolomite bricks and the safety lining was considered an integral part of the dolomite brick lining. In
addition, specific heat and thermal conductivity were
regarded as constant within each lining layer. At the
hot face, the radiation emissivity was taken dependent
of both combustion gas composition and the temperature of the gas. A scheme for computing the radiation
heat flux as a function of temperature is outlined. Finally, a number of example simulations were studied
with the purpose of investigating suitable fuels and
fuel rates for a specified preheating task both for fresh
and circulated ladles. The advantage of preheating the
combustion air was also noted here. In a comparison of
measured lining temperatures with simulated ones,
agreement was satisfactory.
Austin et al. [28] outline a thermal model for steelmaking ladles developed at the BHP steelworks in
Australia. The model was used in an extensive parametric study of the various stages in the ladle cycle.
Mathematically, the model is based on a numerical
solution of the transient two-dimensional heat conduction equation. An energy balance for the molten
steel was included in order to simulate the temperature drop. Stage data, ladle geometry, refractory properties and results can all be accessed through a user
interface. Refractory configurations can be easily manipulated and temperature dependence of the thermal

Heat transfer in steelmaking ladle refractories

Fig. 10. Temperature profiles in the ladle lining for different preheating
and empty times. By Austin et al. [28]. Courtesy of the Iron & Steel
Society.

properties can be considered. The arising non-linear


system of equations is solved using the alternating direction implicit method with adaptive time stepping.
A few examples featuring temperature profiles in the
lining are presented. Tuning of the model was done
by adjusting convection coefficients, emissivities etc.
in stages by comparing simulations with measurements. The procedure is illustrated by results for the
preheat and first four cycles of a 275 t test ladle monitored with thermocouples in the lining. With the tuned model, thermal effects of preheating, slag thickness, holding time, lid use, empty time, refractory
type and wear were investigated.
In the parametric study, a well-cycled ladle (defined
arbitrarily as the state occurring after 10 heats) was
kept as reference. During preheating, the ratio of momentary to final heat content was considered a function of time only and quasi steady state for a freshly
lined ladle was established after four to six heats, depending on preheating time. The most important factor for steel chilling during the ladle cycle was working lining heat content, influenced mainly by the
length of waiting time between consecutive heats.
Long empty periods were compensated by preheating, however, for short periods preheating turned
out to be counterproductive. Consider an empty time
of 30 min, after which the hot face temperature has
fallen significantly but the temperature profile in the
working lining still is high. Applying preheat will
now only serve to increase the hot face temperature
to the flame temperature, after which the temperature
gradient between the flame and lining becomes small,
resulting in low heat flux. Since the heat content of
the working and safety linings then remains virtually

unaffected, heat is lost to the surroundings through


the steel shell. This was confirmed by the model calculation results for different preheating times after an
idle period, illustrated in Fig. 10. Overhead heat loss
from the molten steel was found to be mainly due to
radiation with no slag layer and conduction through
the slag when one was present. Conduction losses
were one order of magnitude smaller than radiative
losses. Increasing slag layer thickness did not reduce
the heat loss significantly, neither did lid use during
periods with the ladle filled, if a slag layer was present. Lid use was found significant mainly when the
hot face was exposed, like during casting and empty
periods. When the ladle was filled, much larger impact on steel heat loss was observed for the lining heat
content in comparison with lid use. In the BHP plant,
high alumina linings were used predominantly. To
compare this lining configuration with zirconia brick
refractories, the transient steel temperature was simulated for a well-cycled ladle in both cases. As expected, the simulations predicted larger heat loss for
high alumina than for zirconia although the initial
chill during tapping was slightly larger for zirconia
than for high alumina. Finally, the effect of lining
wear on heat loss was studied with the model. It was
found that a ladle which is only slightly worn delivers
metal at almost the same temperature as a new ladle,
whereas for a heavily worn one the delivery temperature is substantially lower, suggesting nonlinearity in
the impact of wear on heat loss.
Gaston et al. [47] estimated the temperature drop of
molten steel during tapping into the ladle. The work
is a part of a larger thermal model for the ladle cycle,
implemented as a computer code package. Heat losses
both due to exposure of the tapping stream to the ambient air and due to radiation from the steel head in
the ladle and to conduction into the ladle working lining were studied. In order to model the heat loss from
the tapping stream, stationary mass and momentum
balance equations were set up together with the differential energy balance for an infinitesimal height
element of the stream. Combination of these and integration of the energy balance yielded an estimate of
the temperature drop of the metal stream for a given
stream height. In agreement with experiments, the example calculation using realistic plant data shows a
temperature drop of about 0.5C for a constant tapping height of 2 m. For the steel in the ladle, the heat
balance equation for a linearly time dependent steel
level is integrated after substitution of loss terms arising from conduction to the refractory working lining
and radiation from the steel surface. Radiation heat
loss is assumed constant and a function of the steel

251

Fredman

tapping and ambient air temperatures. Losses due to


conduction are approximated inversely proportional
to the square root of time. The approximations for the
heat losses are valid for short tapping/filling times
and small overall temperature drops for the steel. The
temperature drop of the steel exhibits the same form
of time dependence as the Yngve Sundberg equation
(8) with the prefactor of the linear term describing
radiation from the melt surface of constant area and
the prefactor of the square root term describing conduction to the working lining. Finally, an example calculation for a 2 min tapping is carried out for two
different ladle preheat levels (800C and 1200C).
After an elapsed time of 6 min the total temperature
drops of the molten steel in the ladle are 33C and
25C, respectively. An advantage of the outlined ap-

Fig. 11. Ladle thermal tracking model display of Zoryk & Reid [48].
Courtesy of the Iron & Steel Society.

Fig. 12. Display for the steel temperature flight path model of Zoryk &
Reid [48]. Courtesy of the Iron & Steel Society.

252

proach is that it is not necessary to know the temperature profile in the ladle refractory.
Zoryk & Reid [48] describe an integrated system
for on-line estimation of liquid steel temperature in
the ladle and tundish, developed at the Scunthorpe
Works of British Steel plc. The system consists of two
separate models, one for uninterrupted calculation of
refractory lining temperatures for all operational
ladles and one for calculation and visualization of
liquid steel temperature in the ladle and tundish (the
temperature flight path). The former is called The
Ladle Thermal Tracking Model and the latter The Flight
Path Model. An example display of the Ladle Thermal
Tracking Model is depicted in Fig. 11 and an example
of the Flight Path Model display in Fig. 12. Both
models are based on finite difference methods to solve
the one-dimensional heat conduction equation and
liquid steel energy balance, respectively. Temperature
profiles in the lining are computed at two positions,
at the center of the ladle bottom and halfway up the
ladle wall. Boundary conditions for the lining temperature profile model during periods of cooling, drying, preheating and casting were studied by thermocouple monitoring of 3 different depths within the lining at the bottom and in the sidewall. The performance of the simulations with regard to matching measured thermocouple temperatures is demonstrated,
showing an average difference between measured
and calculated temperatures of approximately 25C.
In the calculation of the steel temperature trajectory,
the heat balance includes conduction to refractories,
convection and radiation from lining and slag surface,
tapping and teeming losses as well as chilling due to
alloying. When using the ladle arc furnace, the model
is fed with data on power inputs, heating times and
induction stirring.
The ladle thermal tracking model calculates the
thermal states of all ladles in use, considering the current status of the ladle, i.e., cooling, preheating, drying or in cycle. For the two profile positions, refractory types and thicknesses are kept in a data base and
refractory wear is modeled with an average wear rate
obtained from the current campaign lining life and
inspections. Model output is available in a variety of
ways, as history of ladle use in the plant, as temperature profiles for a specific ladle and as a table ranking
the operational ladles with respect to thermal state.
Additional features are prediction of thermal state for
a number of ladles based on planned process routes,
access to data bases containing ladle fleet parameters
and initialization of thermal state.
The flight path calculations start right after tapping, an initial steel temperature trajectory prediction

Heat transfer in steelmaking ladle refractories

is obtained on basis of the ladle thermal state, final


BOF temperature measurement, planned process
route and parameters (including casting speed evaluated from the section size to be cast and recommendations for the particular steel grade). Subsequently, as
data from actual events, e.g., argon stirring, degassing, reheating, alloying or even unexpected delays
become available to the flight path model, updates of
the predicted process routes are made and the steel
temperature flight path for the remainder of the cast
is revised. From this, the plant operators can take
necessary corrective action to ensure that the aim temperature for the steel at start of casting is met. These
corrections can amount to ladle arc furnace reheating,
additional/reduced argon stirring, scrap cooling or altering the process route. When casting is in progress,
ladle and tundish dip temperature measurement is
carried out routinely enabling revision of steel temperature flight path.
The performance of the on-line system was reviewed on basis of ability to predict measured ladle and
tundish dip temperatures. The standard measurement
error for the ladle dip was 5C and on average 86%
of all ladle dip temperatures were calculated to within
measurement error bounds by the flight path model.
For some reason the error for tundish dips was larger
(7C) and for the slab caster 92% of the dip temperatures were estimated by the model to within error
bounds. The bloom and billet casting routes of the
plant showed almost 25% lower figures, due to their
significantly longer residence times for the steel in the
tundish, resulting in stratification.
Sistilli & Erny [49] describe a ladle tracking system
implemented at the Steubenville, OH plant of Wheeling-Pittsburgh Steel Corp. After conversion to continuous casting in 1991, a need for improvement in
ladle status tracking became evident. More stringent
quality requirements and the dramatic increase in the
amount of secondary refinement performed in ladles
threatened to substantially increase the number of rejected heats and ladle refractory failures (burnthroughs). For each ladle, a database was created,
containing variables critical for refractory performance, such as event times, stirring and reheating data.
When a heat is tapped, a new record is added to the
data base and the furnace crew enters the ladle number along with alloy and desulphurization material
additions. The rest of the data are automatically
added to the database, and it is possible to view the
statistics of any ladle in the fleet. An analysis of the
influence of key operating variables on refractory performance was carried out. One particular lining specification was evaluated over a campaign of 63 ladle

cycles. The most important variables were identified


to be average steel contact time, number of reheats,
number of desulphurizations and average stirring
time, the last two exhibiting the strongest negative effect on ladle lining life. Finally, a heuristic method of
predicting ladle life is outlined, where the results of
the study are used to relate the current mode of operation (values of the key variables) to an average expected ladle life for the entire plant. Over a reference
time of three months, the ladle life forecasting accuracy was 80% 3 heats.
Olika & Bjorkman [36] used temperature measurements from the ladle wall and molten steel to evaluate
a commercial computer program package, TempSim,
for simulation of ladle thermal state and steel heat
loss. The main component of TempSim is an energy
balance for the molten steel. The program features calculation of lining temperatures during ladle maintenance and preheating as well as heat loss and steel temperature during alloying, heating and vacuum treatment. Temperature drop during tapping and deoxidation, effects of melting and solidification of slag and
variations in steel level in the ladle can also be studied
using TempSim. The ladle lining is divided into four
sections for which the material, thickness and height
are specified. One section consists of the ladle bottom,
while the remaining regions make up the sidewall.
Each section can be divided into a maximum of ten
subsections. The tapping stage can be divided into
three temporal subintervals, where 1/4, 1/2 and 1/4
of the molten steel leaves the converter, respectively.
A simple linear regression model, with separate coefficients for different temperature intervals, is used for
estimation of the temperature drop between in-converter temperature measurement and end of tapping.
A measurement campaign was implemented at
SSAB Tunnplt AB of Lule, Sweden in order to monitor lining temperature profile, hot face, steel and slag
temperatures. Slag thickness and chemical composition were also investigated. 2 thermocouples were
placed at the bottom and 12 in the sidewall of a ladle.
Hot face temperatures were measured, using a pyrometer, at the end of casting just after slag-off, before

Table 2. Boundary conditions for the ladle thermal model in [34]


Parameter

convection coefficient,
[h]W/(m2K)
view factor, ( )
gas temperature (C)

Ladle cooling

Ladle preheating

wall

bottom

wall and bottom

12.6

8.4

0.12

0.06

12.0
0.0
1900

253

Fredman

and after ladle cleaning and about 23 min prior to


tapping. Steel temperature in the ladle was measured
every 5 min throughout ladle metallurgy and immediately before casting. With TempSim, steel temperature
on arrival to the caster could be estimated with a standard deviation from measurements of 3C. The simulations underestimated temperature drop during ladle
metallurgy and overestimated it during transfer to
casting. This was explained by overestimation of initial thermal state and poor understanding of the radiative heat transfer. Other reasons were insufficient
knowledge of parameters, such as slag weight from
the converter, waiting time in the converter, tapping
time, extent of stratification and lining wear. Disagreement of measurements with simulated lining temperature profiles was thought due to sculling and the
resulting reduction in emissivity.
Barber et al. [34] investigated how an existing ladle
fleet could be modified to withstand the prolonged
steel residence times and higher superheats of the
new horizontal continuous caster put into operation
at the Avesta Panteg plant of Sheffield, UK. It was
decided that some form of additional insulation
would be necessary to reduce steel heat loss and shell
temperature. Insulation had been retrofitted previously at other plants, mainly to prevent shell temperatures from exceeding safety recommendations.
The insulation had to be sufficiently thin, since ladle
capacity otherwise would have decreased. In order to
study various refractory configurations, thermal
models for the ladle and tundish were developed.
These were based on finite difference solution of the
heat conduction equation with parameters tuned so
as to match the model output with thermocouple
measurements in the ladle lining (Table 2). 5 thermocouples were installed in the sidewall and 5 in the
bottom lining of the monitored ladle. A total of 24
ingot casts over a period of 5 days were logged with
the thermocouples.
Further calculations were done on heat loss
through slag and insulating powder covers, liquid
steel exposure on stirring and melting and the effect
of alloy additions. Initial trials were carried out with
a commercially available material in the form of
thin, flexible sheets with a quoted thermal conductivity of 0.035 W/mK, capable of withstanding temperatures up to 1025C. Measurements were done
on an ingot casting ladle insulated with 5 mm of the
suggested material, where the effective thermal conductivity of the material was found to be approximately twice that of the quoted value. After 65
heats the lining was inspected and a compression of
the insulation roughly equivalent to 20% could be

254

observed. In view of these results and further model


calculations, the ladles intended for operation with
the new horizontal continuous caster were equipped
with 5 mm of the insulation in the sidewall and 12.5
mm at the bottom. The final design went through a
third measurement campaign, where hot face temperatures for the insulation were observed to be
safely below the recommended limit. An operating
practice for the continuous casting was also established and ladle dip temperature measurements
were taken, where the total temperature drop now
turned out to be 46C. Moreover, it was decided that
the ladle be kept idle for 1.52 h in order to soak
the lining thermally before start of casting. An alternative practice is to make the cycle as consistent
and regular as possible without long waiting
periods. Finally, it was observed that insulation resulted in hotter linings, a potential threat to refractory integrity. However, working lining life for the
modified ladle fleet reached 3000 min of steel contact time, which was roughly the same as for the
old ingot casting ladles. The insulation lifetime in
the new fleet was three working linings.
Grip [50] outlines simple methods for calculation
of steel temperature drop in the ladle. Elementary
principles of conduction heat transfer and use of the
Yngve Sundberg equation (8) are discussed. The commercial code TempSim is briefly described and it is
demonstrated how the result from TempSim conforms
to equation (8) with suitable coefficients. Through a
series of measurement results, Grip illustrates the
quasi-stationary nature of lining temperature profiles
and the importance of thermal history (e.g., preheating) and regularity in the ladle cycle. The frequently overestimated chilling effect of gas stirring is
indicated in a simple manual calculation, showing a
much smaller temperature drop than the measured
one. Grip claims that this is a result of stratification
and that stirring is not an effective means of temperature adjustment. Finally, a temperature control system
based on simple regression expressions of the form
(8), developed using TempSim, is described.
Fredman & Saxen [51] developed an analytical
model for estimation of temperature profiles in ladle
refractories. The model is based on local solution, by
separation of variables, of the one-dimensional transient heat conduction equation within the lining
layers. Using the local solutions, an overall temperature profile can be formed after computation of the
material boundary temperatures. Model dynamics
are considered in the working lining only, the outer
layers of the refractory are at steady state. This approach of simplifying the dynamics and geometry

Heat transfer in steelmaking ladle refractories


Table 3. Reviewed theoretical works relating to ladle thermal state
Ref.

Model structure

Features

independent
variables, solution
method

steel temp.
included

tundish
included

Paschkis [26]

(,t)

no

Paschkis [27]

(,t)

type of
measurement

Alberny [37]

(r,t),FDE

yes, stagnant
melt
yes, stirred
melt
stirred

Chone [38]

(r,t)

stirred

yes, steel
temperature
no

Omotani [39]

(r,t)

stirred

no

Pfeifer [31]

(r,t),FDE

stirred

no

Pfeifer [32]

(r,t),FDE

no

lining

Pfeifer [40]

(r,t)

yes,also
tap stream
stirred

no

lining

Morrow [30]

(r,t),FDE

stirred

no

lining

Hlinka [29]

(r,t),FDE

no

Tomazin [25]

(r,t),FDE

variable
stirring
yes

lining

Hoppmann [41]

(,t)

yes

yes

Hoppmann [35]

(,t)

yes

no

Chapellier [42]

(,t)

yes

no

Koo [43]

(,t)

no

lining

Gaston [44]

(r,t)

yes,
stratified
yes

yes

steel
temperature

Shklyar [45]

(r,t)

yes

no

Mucciardi [46]

(r,t),FDE

no

no

Saha [33]

(r,t),FDE

no

no

lining

Austin [28]

(r,z,t),ADI

yes

no

lining

Gaston [47]

(r,t)

no
yes

lining
steel temp.
scheduling
data
lining, hot
face, steel temp.
lining

no

Zoryk [48]

(r,t),FDE

yes, also
tap stream
yes

Sistilli [49]

(,)

no

no

Olika [36]

(r,t)

yes

no

(r,t),FDE

yes

yes

Grip [50]

(r,t)

yes

no

Fredman [51]

(r,t)

no

no

Fredman [52]

(r,z,t),FEM

yes

no

Barber [34]

ensures rapid computation and maximum flexibility


in boundary conditions, beneficial for, e.g., on-line

steel
temperature

steel
temperature
steel
temperature

lining,
steel temp.
lining,
steel temp.
lining
steel temp.

purpose
of model

remark

estimate steel
temp. drop
estimate steel
temp. drop
improve
temp. control
aid design,
decisions
decision
support
design of
ladle head
improve
temp. control
simulate
preheating
lining
design
decision
support
design,
therm. practice
calc. converter
tap. temp.
calc. casting
temperature
calc. converter
tap. temp.
study eff.
of stirring
estim. strat.
in tundish
simulate
preheating
lining
design
simulate
preheating
improve
temp. control
est. tapping
temp. drop
on-line est. of
steel temp. in
ladle, tundish
analyze
scheduling
calc. casting
temperature
aid redesign
of ladle fleet
develop simple
temp. models
rapid est. of
th. state
aid design,
decisions

analog comp.
implementation
analog comp.
implementation
stratification
considered
whole cycle
considered
steel heat
balance
stirring,
slag th. inv.
mod. tuning
described
preh. strat.
investigated
variable
therm. props.
whole cycle
considered
variable
therm. props
part. statist.
model
part. analyt.
model
statistical
model
CFD-calc. of
stirring
steel heat
balance
steel heat
balance
used avail.
software
studied
recuperative preh.
used adaptive
time step
ladle load
linear in t
th. state of
ladle fleet
statistical
investigation
part. statist.
model
inv. various
heat losses
studied regularity,
preheating
rapid
computation
dynamics after
preheating

simulation. Apart from computational issues, such


as efficient solution of eigenvalue spectra, cooling of

255

Fredman

the exposed hot face is discussed. To illustrate the


feasibility of the approach, simulations were compared with data from a measurement campaign implemented at the facilities of a Finnish steel producer. A ladle with a two-layer refractory was fitted
with 6 thermocouples in the lining and the readings
logged over some 26 heats. The simulations were in
acceptable agreement with the measured temperatures.
Fredman et al. [52] present a two-dimensional
mathematical model for simulation of thermal state of
ladle linings and steel temperature during casting. For
the lining geometry, the two-dimensional dynamic
heat conduction equation is solved, using the Finite
Element Method (FEM), simultaneously with the energy balance equation for the perfectly mixed steel.
Radiation heat transfer above the slag surface is
coupled with both the thermal state of the lining and
the energy balance. Hence, radiation, convection as
well as conduction in the lining are considered at each
time step in the solution of the energy balance for the
molten steel. As the computations become rather
heavy, the model is suitable mainly for off-line simulation. Lining design, scheduling and decision support are feasible application areas for the model,
which also features a user-friendly interface for input
data specification. To validate the model, a measurement campaign was carried out at the facilities of a
Finnish steel producer. A 3-layer refractory lining was
fitted with 16 thermocouples at 2 heights in the sidewall and one position at the bottom. The campaign
was more extensive as compared to [51], and temperatures were recorded for a total of 82 heats. To test the
dynamics of the model and study the gradual increase
in heat content of the lining, simulations were done
over some ten initial heats for a fresh ladle. Apart
from minor technical problems with the thermocouples, due to overheating, the campaign was successful in demonstrating the feasibility and usefulness
of the model.

Summary of the theoretical models


The reviewed theoretically focused works relating to
thermal state of ladle systems have been summarized
in Table 3. In the classification of the investigations
the model structure and special features of the work
have been used as criteria. Model structure is described through the dimensionality (set of independent variables), the solution method used in the simulation and the output. Possible output might be one
or several of the lining thermal state, steel temperature in(to) the ladle and thermal state/steel tempera-

256

ture in the tundish. Assumptions pertaining to the degree of agitation in the steel are indicated for some
works. For the set of independent variables, (r, z, t)
denotes a two-dimensional, dynamic model where r
is the radial coordinate, z is the height coordinate and
t is time. When the height is omitted, the model is
essentially one-dimensional, although temperature
profiles may have been computed both at several positions on the ladle sidewall as well as on the bottom.
In cases where all of the variables have been omitted
it was not possible to establish the exact structure of
the model on basis of the text. The same applies to
the solution method used. Where indicated, the abbreviations FDE, FEM and ADI have been used for
finite difference equations, finite element method and
alternating direction implicit, respectively. These are
explained in the numerical mathematics literature, see
[53]. As features are considered supporting measurements (in some cases used for validation and tuning
of the model) as well as the purpose of the model and
special methods or details indicated under remark.
From Table 3, it is seen that the objectives of the
investigations have been quite varying. Nevertheless,
the basis of the works is in many cases a one-dimensional model for the temperature profile across the lining of the ladle and/or the tundish. In most cases, the
straightforward and simple finite difference scheme
was chosen for a method of solution. Only a few
models, [28] and [52], were 2-dimensional and used
special solution methods. Austin et al. [28] used the
ADI-method, based on finite differences, and Fredman et al. [52] used FEM. One work, [48], was primarily concerned with on-line simulation. Thus, in most
cases, except [48] and [51], the computation time was
probably not critical for the choice of model. It also
appears that the one-dimensional approach has become a commonly accepted method among the
workers in the field. Another reason for its popularity
is perhaps the increased complexity in boundary conditions for two-dimensional models, resulting in more
parameters open to assumption and more difficult
tuning of the model. A fact, however, is that none of
the reviewed one-dimensional studies address the
issue of axial (in the z-direction) heat transfer in the
lining quantitatively. On the other hand, for the molten steel, thermal uniformity is a reasonable assumption in light of the measurements and CFD (computational fluid dynamics) studies included in a number
of the reviewed works. As was noted in the summary
of the experimental investigations, the slight thermal
stratification occurring in the ladle evens out the temperature of the steel exiting the tundish. Hence, assuming the melt to be perfectly stirred causes the heat

Heat transfer in steelmaking ladle refractories

balance to give a lower steel temperature at start and


in the end of casting a heat. Making some suitable
approximation of the stratification might make it
possible to consider this phenomenon through the
heat balance, without expensive fluid dynamics computations. In a few works, the variation of the thermophysical properties with temperature was considered.
Apparently, the functions for thermal conductivity
and specific heat had been supplied by the refractory
vendor or manufacturer. A potential problem here is
the change taking place in the material over the lining
campaign. Consequently, the used functions may not
have been even close to the real temperature dependencies after the first five heats on the lining.
The performance of the steel temperature models
summarized in Table 3, when compared with measurement data in the investigations, was varying. Good
agreement, e.g., in [48], was obtained when sufficient
measurement data was available, making it possible
to tune the model through adjusting its parameters
so as to produce lining temperature profiles and steel
temperature trajectories agreeing with campaign
measurements. Clearly, this applies to the presented
lining thermal models as well.

Conclusions
A literature review dealing with topics related to
monitoring and modeling the heat loss from ladles in
steel production has been presented. A wide range of
works were covered, including issues from process
scheduling in continuous casting to optics and physical modeling. Scheduling and process control have a
direct impact on the heat loss from the steel-ladle system and can be combined with design of process
equipment to reduce heat loss. Studying these systems experimentally is fundamental not only for
understanding the involved mechanisms of heat and
mass transfer but also for formulation of theoretical
models describing the thermal state of the system. In
the part dealing with theoretical modeling, focus has
not been as strongly on results in the form of agreement with measurements as on model structure and
central ideas implemented in the model formulation.
The reason was that comparison of different models
describing different plants and validated against experimental data obtained under different circumstances was difficult. Looking back at the reviewed
works awakes some thoughts on modeling this type
of system. Design evaluation and on-line simulation
seem so require completely different model structures
in order to address key issues, such as model realism
and computation speed. Therefore, a more complex,

perhaps two-dimensional dynamic model is suitable


for design purposes and a simpler one-dimensional
model for on-line simulation. The simpler model does
not have to be merely a regression equation for the
drop in steel temperature as a function of time in a
particular plant, but it could be a combination of analytical and numerical results for the heat conduction
and energy balance in the system. In a situation where
both model types would be required, it might also be
possible to validate the on-line model against the design model if necessary. Naturally, this would be feasible only after thorough validation of the more complex design model.

Acknowledgements
Permissions to reproduce previously published figures, granted by The Association of Iron and Steel
Engineers, The Iron & Steel Society and The Institute of Materials are gratefully acknowledged. The
author also wishes to express his gratitude for the
financial support received from Liikesivistysrahasto,
Finland.

References
1. Mish FC. Websters 9th New Collegiate Dictionary. Merriam-Webster Inc., Springfield, Mass. USA, 1988.
2. Fruehan R. Iron & Steelmaker 1996: July: 2533.
3. Kivela E, Sipola M. Teraksen lampotilahallinta senkkakasittelyjen ja jatkuvavalun aikana. Technical course print, in Finnish, Jun. 1994. Taydennyskoulutuskurssi Tulenkestavat Terasteollisuudessa, PohTo, Oulu Finland.
4. Sillanpaa M. Model for predicting the evolution of steel temperature in a continuous casting tundish on the basis of expert knowledge. Report 91-1, bo Akademi University, Heat
Engineering Laboratory, Biskopsgatan 8, FIN-20500 bo
Finland, 1991.
5. Cramb AW. Trends in caster operation, part 1. Iron & Steelmaker Sep. 1989: 45.
6. Cramb AW. Trends in caster operation, part 2. Iron & Steelmaker Oct. 1989: 56.
7. Cramb AW. Trends in caster operation, part 3. Iron & Steelmaker Nov. 1989: 4748.
8. Ameling D, Walden K, Wethkamp H. Stahl und Eisen 1981:
101 (15): 3539.
9. Hoppmann W, Pfeifer H, Fett FN, Fiege L. Stahl und Eisen
1987: 107 (1): 4954.
10. Kitamura M, Kawasaki S, Kawai S, Kawai K, Miyake K.
Transactions ISIJ 1983: 23: B165.
11. Minion RL, Leckie CF. Steel temperature control in the ladle
in a high productivity BOF shop. In: 5th International Iron
and Steel Congress. Proc the 69th Steelmaking Conf. Iron &
Steel Society 1986: 69: 335343.
12. Perkins A, Robertson T, Smith D. Fachberichte Huttenpraxis
Metallweiterverarbeitung 1986: 24 (8): 649656.
13. Petegnief J, Birat JP, Larrecq M, Bobrie M, Lemiere F, Fautrelle Y. Revue de Metallurgie CIT 1989: Jan: 4754.

257

Fredman
14. Rieche K, Kohn W, Wunnenberg K. Stahl und Eisen 1985:
105 (19): 4146.
15. Rutqvist S, Bergman D, Olika B. Scandinavian Journal of
Metallurgy 1990: 19: 146152.
16. Saunders LM. Preheating and controlled thermal cycling of
steel handling ladles. In: 66th Steelmaking Conf Proc. Iron &
Steel Society 1983: 66: 6975.
17. Samways NL, Dancy TE. Journal of Metals 1960: 331337.
18. Ettwig HH. Stahl und Eisen 1995: 115 (6): 6770.
19. Grip C-E. Measurement of ladle wall temperature to improve control of steel temperature in BOF plant. In: 77th
Steelmaking Conference Proceedings, Iron & Steel Society
1994: 77: March.
20. Hlinka JW, Miller TW. Temperature loss in liquid steel-refractory systems. Iron and Steel Engineer, Aug. 1970.
21. Widdowson R. Ironmaking and Steelmaking 1981: 8 (5): 195
200.
22. Cardouat JM. Revue de Metallurgie CIT 1986: 203210.
23. Christensen KW. Heat losses and scull formation in Fe-Si
and Si-metal ladles. In: Electric Furnace Conference Proc.
Iron & Steel Society 1988: 195199.
24. Hecht E. Optics, 4. Addison-Wesley, 2nd edition, 1987.
25. Tomazin CE, Upton EA, Wallis RA. The effect of ladle refractories and practices on steel temperature control. Iron &
Steelmaker 1986: Jun. 2834.
26. Paschkis V. Temperature drop in pouring ladles. AFS Transactions 1956: 64: 565576.
27. Paschkis V, Hlinka JW. Temperature drop in pouring ladles,
part two. AFS Transactions 1957: 65: 276281.
28. Austin PR, ORourke SL, He QL, Rex AJ. Thermal modelling
of steel ladles. In: 75th Steelmaking Conference Proceedings,
vol. 75, pp. 317323. Iron & Steel Society, 1992.
29. Hlinka JW, Cramb AW, Bright DH. A model for predicting
the thermal history of a ladle of steel. In: 68th Steelmaking
Conference Proc. Iron & Steel Society 1985: 68: 3546.
30. Morrow GD, Russell RO. Thermal modeling in melt shop
applications. Am Ceram Soc Bull 1985: 64 (7): 10071012.
31. Pfeifer H, Fett FN, Schafer H, Heinen K-H. Stahl und Eisen
1983: 103 (2526): 13211326.
32. Pfeifer H, Fett FN, Schafer H, Heinen K-H. Modell zur thermischen Simulation von Stahlgiesspfannen. Stahl und Eisen
1984: 104 (24): 12791287.
33. Saha JK, Ajmani SK, Chatterjee A. Ironmaking and Steelmaking 1991: 18 (6): 417422.
34. Barber B, Watson G, Bowden L. Ironmaking and Steelmaking 1994: 21: 150153.
35. Hoppmann W, Fett FN, Hsu G, Fiege L. Stahl und Eisen
1989: 109 (23): 11771186.

258

36. Olika B, Bjorkman B. Scandinavian Journal of Metallurgy


1993: 22: 213219.
37. Alberny R, Leclercq A. Heat losses from liquid steel in the
ladle and in the tundish of a continuous-casting installation.
In: Mathematical process models in iron and steelmaking,
Proceedings, Amsterdam, pp. 151156. The Metals Society,
1973.
38. Chone MJ, Teyssier F. Senkan lampohavioiden laskeminen.
Technical Report RE 137, IRSID, Maizieres-les-Metz, Oct.
1973. Translated from French at Rautaruukki Oy, Finland.
39. Omotani MA, Heaslip LJ, Maclean A. Ladle temperature
control during continuous casting. Iron & Steelmaker 1983:
Oct: 2935.
40. Pfeifer H, Fett FN, Schafer H, Heinen K-H. Stahl und Eisen
1985: 105 (14/15): 759764.
41. Hoppmann W, Fett FN, Fiege L, Bachmann R. Stahl und
Eisen 1988: 108 (2): 6166.
42. Chapellier P. Revue de Metallurgie CIT 1989: 687692.
43. Koo YS, Kang T, Lee IR, Shin YK, Gal HY. Thermal cycle
model of ladle for steel temperature control in melt shop
and its application. In: 72nd Steelmaking Conf Proc. Iron &
Steel Society, Apr 1989: vol. 72.
44. Gaston A, Laura R, Medina M. Ironmaking and Steelmaking
1991: 18 (5): 370373.
45. Shklyar FR, Malkin VM, Korshunov VA, Sovetkin VL. Mathematical model of thermal performance of steel pouring
ladle. Steel in the USSR Feb. 1991: 21: 6870.
46. Mucciardi F, Grandillo A. Optimizing ladle cycling strategies. In: 74th Steelmaking Conf Proc. Iron & Steel Society
1991: 74: 757760.
47. Gaston A, Laura R, Medina M. Thermal cycle of continuous
steel casting ladles: a mathematical model for predicting the
steel temperature during the tapping and the filling period.
Latin American Applied Research 1993: 195200.
48. Zoryk A, Reid PM. On-line liquid steel temperature control.
Iron & Steelmaker 1993: June: 2127.
49. Sistilli FR, Erny EL. Computerized steel ladle tracking.
Iron & Steelmaker 1993: Oct: 6366.
50. Grip C-E. Calculation of the ladle temperature. Technical
course print, Feb. 1995. Metallurgisten prosessien mallintamisen perusteet ja tyokalut, PohTO, Oulu Finland.
51. Fredman TP, Saxen H. Metallurgical and Materials Transactions B 1998: 29B: 651659.
52. Fredman TP, Torrkulla J, Saxen H. Metallurgical and Materials Transactions B 1999: 30B: 323329.
53. Vemuri V, Karplus WJ. Digital computer treatment of partial
differential equations. Prentice-Hall Series in Computational
Mathematics. Prentice-Hall, Englewood Cliffs, NJ, 1981.

You might also like