Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Steelmaking Process*
By Kaxumi MORI **
(246) Review
Transactions ISIJ, Vol. 28, 1988 (247)
bubble size depends only on the gas-flow rate and noz- pends on the liquid physical properties is attained.
zle diameter. The empirical correlation obtained by It is clear that at low gas-flow rates the bubble size is
Davidson and Amick by air-water and air-oil sys- determined by a force balance under the static condi-
tems6) is tion while at higher gas-flow rates a balance between
the upward acceleration of the liquid surrounding the
dB = 0.54(Vgdo.5)o.289
.....................(2)
bubble and the buoyancy force becomes important in
where, Vg; the gas-flow rate (cm3fs). determining the bubble size.
Equations (1) and (2) were obtained for bubble Sano and Mori10) determined the size of bubbles
formation from wetted nozzle in water and organic in the bubble swarms in mercury from measuring the
liquids. In this case do is equated to the inner nozzle bubble frequency by using an electroresistivity probe
diameter. The applicability of these equations to technique. From examining the effect of liquid physi-
non-wetted nozzles in various liquid metals was inves- cal properties on the size of bubbles in the swarms,
tigated in the author's laboratory.7-9> the following general equation is obtained.
The metals used were mercury, molten silver, and
molten iron. Bubbling was performed through a noz-
zle, the tip of which was of the upward faced type.
vs - 0.091
(c)O.5
0.44
s ..................(4)
The bubble volume was determined from the volu-
where, dvs : volume-surface mean diameter of bub-
metric gas-flow rate and the frequency of bubble for- bles (cm)
mation.
VS: superficial gas velocity (cm/ s).
It was found that the non-wettability of the nozzle The size of bubbles in molten iron estimated from
in liquid metals has decisive influence on the bubble
Eq. (4) is compared with that estimated by Fruehanil)
size. When the outer nozzle diameter was adopted for Q; BOP.
as do in Eqs. (1) and (2), close agreement between
the experimental results for liquid metals and calcula- 2. BubblingJetting Phenomenain Gas Injectioninto Liquid
tions was obtained. This finding has a significant
This section is concerned with the behavior of gas
meaning in practice such as gas bubbling through a
jet near the orifice. A series of studies has been made
porous plug in molten steel. in the author's laboratory on the behavior of injected
Based on the experimental results, an empirical
equation to predict the size of bubbles in a wide gas jets at a submerged orifice in liquid over a wide
range of gas-flow rates extending to the region of sonic
range of gas-flow rates is given by
velocity.l2-16) To know the influence of gas and liq-
uid properties on jet behavior, three models of ni-
dB
={()
6Q
Plgn2+0.0242(V2dn)0.867
g }1/6
.........(3) trogen-mercury, nitrogen-water and helium-water
systems have been employed. The jet behavior ex-
Equation (3) becomes Eq. (1) at low gas-flow rates, clusively in the initial jet formation zone was observed
while it agrees with Eq. (2) at higher gas-flow rates. by using a visual technique.
Sizes of bubbles formed in various liquids are cal- The formation of a bubble at a submerged orifice
culated by using Eq. (3) and compared in Fig. 1. in liquid can be represented by a rectangular wave
Here, the calculation is made for the nozzle of 0.6 the height and width of which are equal to the maxi-
cm in diameter. As seen from Fig. 1, at low gas-flow mum bubble base diameter (dM) and the bubble for-
rates the bubble size decreases in the order of de- mation time, respectively. Figure 2 is an example of
creasing a/pi, that is, molten iron, water, molten silver rectangular waves showing successive formation of
and mercury. As the gas-flow rate increases, the dif- bubbles for nitrogen injection into mercury. Here,
ference in the value of bubble size becomes smaller two phenomena are distinguished. One is the forma-
and the region where the bubble size no longer de- tion of bubbles of various base diameters. This is
called " bubbling ". Another phenomenon is that an
Fig, 1. Comparison of size of bubbles formed in various Fig. 2. Bubble base diameter (dM) vs. time at a high gas-
liquids.° flow rate.13j
Review
(248) Transactions ISIJ, Vol. 28, 1988
apparent coincidence between the base diameter (4) As seen from Fig. 4, the critical value of M' at
and orifice diameter (do) occurs over various ranges which the transition from bubbling to jetting begins
of time. In these time ranges the injected gas leaves to occur is very close to the subsonic-sonic boundary
the orifice as a continuous jet of gas. This phenome- almost independently of physical properties of gas and
non is called " jetting ". Figure 2 exhibits the alter- liquid. The critical value of M' for nitrogen injection
nation of jetting period with bubbling period. The into mercury was found to be the same for the three
phenomenon of bubbling observed here is considered orifice diameters as shown in Fig. 12 of Ref. 13).
to be very similar to that of " back-attack " observed The most important finding is that jetting occurs
by Ishibashi et a1.17) in the sonic flow region. It is presumed that the gas
In Fig. 3, the time fractions for both bubbling and jet whose flow velocity is equal to the sonic velocity
jetting are plotted against the gas-flow rate for nitro- at the exit of the orifice becomes a supersonic jet be-
gen gas injection into mercury. As shown clearly in yond the exit.14) To the best of our knowledge, these
the figure, jetting is initiated at a critical gas-flow rate were the first to show.
for each orifice diameter. With increasing gas-flow From Fig. 4, it is seen that the value of M' required
rate jetting fraction increases and bubbling fraction to achieve a given jetting time fraction increases with
decreases. However, in the explored range of gas- decreasing gas to liquid density ratio. This finding
flow rates perfect jetting did not occur. Thus, it has could be ascribed to the fact that the stability of the
been cleared that a gradual change from bubbling to gas jet influenced by bath movement increases with
jetting takes place in a transitional gas-flow range. decreasing the liquid density and with increasing the
Similar bubbling jetting phenomena were observed injecting gas density.
for nitrogen-water and helium-water systems. Concerning bubbling behavior, further detailed ob-
In Fig. 4, the time fraction for bubbling jetting is servations were made. " No linking " time, that is,
plotted against the nominal Mach number M' for gas the time interval between the severance of the first
injection into mercury and water. Here, M' is de- bubble from the gas supply and the subsequent link-
fined as the ratio of nominal gas-flow velocity to the age of the severed bubble to the forming second bub-
sonic velocity. The dotted line and broken line in ble was measured.15) On the basis of results of this
the figure indicate the subsonic-sonic boundary for study combined with those of the above-described ob-
nitrogen injection and that for helium injection, re- servations on the bubbling jetting transition, appar-
spectively. The values of M' at the boundary are ently different views on the gas jet characteristics
calculated theoretically on the basis of fluid dynam- proposed by various investigatorsl$_21)could be cor-
1cs.13) related with each other.l5)
Physical behavior of injected gas in the initial jet
formation zone has important connections with serious
process problems such as refractory erosion due to jet
action and clogging of tuyeres by frozen metal.
Sakaguchi et al.22>made an important investigation
on the erosion of a double-tuyere for injection of oxy-
gen into molten steel. From their data giving the
injection condition under which the tuyere refractory
is little eroded, Mori et al.12"3)have estimated the criti-
cal Mach number for erosion of the tuyere refractory
used for injecting oxygen into steel bath. In Fig. 4
the value of this critical Mach number is shown by a
dot-dash line. It is seen that the estimated value
almost coincides with the critical velocity for the bub-
Fig. 3. Time fractions for bubbling and jetting plotted
bling-jetting transition in the He-H20 and N2-Hg
against gas-flow rate (N2 Hg system).13) models. Thus, it has been cleared that the bubbling-
jetting transition is of considerable significance in the
practical problem of erosion of the tuyere refractory.
Studies on the mechanism of the penetration of liq-
uid back into the tuyere during gas injection are very
few. Engh et al.,23) from model studies of gas injec-
tion in water and glycol, obtained a correlation equa-
tion for the formation velocity of droplets entering into
the nozzle. However, a criterion for predicting liquid
penetration leading to nozzle blocking was not given.
Marukawa et a1.24)found, from studies of gas injection
in a 250 t ladle, that gas injection at sonic velocity is
necessary to prevent nozzles from clogging. Very
recently, Sharma25) has done experiments of injecting
Fig. 4. Time fractions for bubbling and jetting for three inert gases into molten steel. He concluded that the
models plotted against nominal Mach number.1s~ controlling factor for metal penetration in the tuyere
Transactions ISIJ, Vol. 28, 1988 (249)
is the jet Froude number, and not the gas velocity tive relation between the mixing time tm and the
(Mach number). According to him, injecting gases stirring power per unit mass of melt &M.
at sonic or supersonic velocity may not be sufficient to Sano and Mori29~ noticed that the relations
prevent metal penetration. between tm and ~M obtained from experiments on
different bath sizes scatter largely. They made a
3. Penetration of Solid Particle into Liquid Metal theoretical analysis on circulating flow in a bath by
In the process of submerged powder injection, setting up a steady-state energy balance equation.3o)
whether the particle passes through the gas-liquid They showed that tm is dependent not only on the
interface or becomes trapped inside bubbles is one of ~Mbut also on sizes of the bath and the plume (bub-
the most fundamental problems. Engh et al.26~dis- ble dispersed zone). Their theoretical correlation is
cussed this problem by using a theoretical model shown by a solid line in Fig. 6, in which the experi-
based on the equation of motion of a sphere moving mental results obtained using water models and 6, 60 t
in liquid. However, their model has not been verified ladles are also shown. The mixing time tm is in in-
experimentally. verse proportion to sM(D2fHo)-2 roughly to the one-
Ozawa and Mori27~have made a study of the pene- third power. Here, D is the bath diameter, and Ho
tration of a spherical body dropped onto a stagnant is the bath depth. Close agreement between the
mercury bath. The penetration behavior was re- calculation and the experiments is seen in the figure.
corded by a high speed cinecamera. In Fig. 5, the Asai et al.31~made a dimensional analysis of mixing
critical entry velocity calculated from the critical time on the basis of Navier-Stokes equation and ob-
dropping distance by using the free fall equation is tained correlating equations corresponding to the flow
plotted against the diameter of the spheres of various regimes. If the mixing length is equal or propor-
materials. This experimental result is well explained tional to the characteristic length of the bath, their
by the calculation using a theoretical model for the equations for laminar and turbulent flow regimes are
critical condition for the penetration of the sphere into similar to the equation obtained by Sano and Mori.29~
the liquid. It is found that the virtual mass of the A mention is made here of the definition for the
sphere and the cavity formation have important mean- stirring power of injected gas specially because of dis-
ings in the mechanism of penetration of particles into agreements among investigators. Several investiga-
liquid. tors overestimated the stirring power by assuming that
the work done by a bubble during its rise is the sum
4. Mixing Characteristicsin a Gas-stirred Molten Metal of work done by the buoyancy force and the volu-
Bath metric work. This sum is exactly twice the actual
Fluid flow phenomena and mixing characteristics work done by a bubble during its rise.
in steelmaking processes have decisive influences on
the steelmaking reaction rate. Many experimental III. Kinetics of Gas-Metal Reactions
and theoretical studies have been made on circulating Kinetics of reactions between gas and molten iron
flow and mixing time in the molten metal bath. have been studied extensively during last three dec-
Nakanishi et al.28~have first introduced the concept ades. Characteristic features in the mechanisms of
of mixing time to evaluate the mixing characteristics decarburization of molten iron-carbon alloys, desorp-
in the steelmaking processes and obtained a quantita- tion of CO and absorption-desorption of nitrogen in
molten iron, and absorption of oxygen into molten
iron are fairly well understood. In the present chapter
three subjects in which the author has had a great
interest will be presented.
Fig . 7. Supersaturation limits for nucleation of CO b ubbles Fig. 8. Comparison of the rate constant of decarburization
in Fe-C---O at 1600°C.33) between liquid and solid iron.
-
Transactions ISIJ, Vol. 28, 1988 (251)
trol is expressed by liquid iron and nitrogen is very instructive in the his-
tory of basic research for steelmaking reactions.
Go/Go = ko~kc The rather earlier work was that done by Pehlke
where, CC,C0: concentration of C and O (mol/cm3) and Elliott40~who measured the rate of the reaction
kc, ko: mass transfer coefficient of C and O of nitrogen with liquid iron by using a modified
Sieverts apparatus. In this study, the rates of ab-
(cm/s).
The results are explained fairly well by the model of sorption and desorption were found to be equal and
liquid-side mass transfer. controlled by a first order reaction. Taking also into
The technique similar to that used in the above consideration the experimental results that the rates
work was applied by Suzuki et a1.39~to study the were greatly influenced by inductive stirring in the
effect of Co, Ni and Cr on the rate of CO removal metal phase, Pehlke and Elliott40~concluded that both
from liquid iron. As presented in Fig. 10, the mass the rates of absorption and desorption were controlled
transfer coefficient of oxygen (k0) is little affected by by mass transfer in the melt.
addition of Co or Ni, while k0 for the Fe-Cr alloy In all kinetic studies made in Japan on the reac-
decreases markedly with increase of Cr content. The tions of nitrogen absorption and desorption, tech-
results could be interpreted reasonably in terms of niques of gas blowing onto the iron melt were em-
thermodynamic interaction parameters in the alloy ployed. It was shown that the absorption of nitrogen
systems. in liquid iron of low oxygen content was controlled by
the mass transfer of nitrogen in the liquid.41~
2. Kinetics of ReactionsbetweenLiquid Iron and Nitrogen A serious problem arose from the studies on the
Kinetics of nitrogen absorption and desorption in desorption of nitrogen into pure argon. It was shown
liquid iron is an attractive problem from both aca- first by Banya et a1.42~that the desorption of nitrogen
demic and practical points of view. In the sixties and obeyed the second-order kinetic law. Other inves-
seventies numerous studies were made on this subject. tigators,43,44~including Choh et al.45~who studied un-
A review of the kinetic studies of reactions between der reduced atmosphere, all obtained the same result
on the kinetics of nitrogen desorption. The rate can
be described by the equation:
_ d[%N] _k, A [o N 2 $
Fig. 10. Effect of X (X: Go, Ni, Cr) on the mass transfer Fig. 11. Effect of oxygen on the apparent rate constant of
coefficient of oxygen at 1580°C.3s) nitrogen desorption from liquid iron at 1600°C.4s)
(252) Transactions ISIJ, Vol. 28, 1988
reaction. This is the first result showing that not all which was faced upward. The immersion depth of
steel refining reactions is controlled by mass trans- the nozzle was 3.3-.4.7 cm. The gas-flow rate was
fer. 70 117 Ncm3/min. The oxygen and sulfur concen-
Comparison of the mechanisms of nitrogen absorp- trations in the melt were varied widely. During each
tion and desorption in liquid iron stated above pre- experiment the frequency of bubble formation was
sents the apparent discrepancy existing between the measured to calculate the volume of a bubble formed
two processes. This kind of discrepancy was called as at the nozzle.
" asymmetry of absorption and desorption " in the Figure 12 shows a typical example of change in
paper by King et al.46~ It appeared that this " asym- nitrogen concentration with time. The rate data ex-
metry " is inconsistent with the microscopic reversi- pressed as the nitrogen absorption efficiency were
bility of chemical reaction. compared with those calculated by using a mixed
Sano et al.47~proposed a mixed control model for control model of liquid phase mass transfer and chem-
the kinetics of gas-metal reactions in which the over- ical reaction at the bubble-metal interface.50~ In the
all reaction rate is controlled both by surface reaction model calculation, nitrogen absorption is considered
and by mass transfer in the metal neglecting gaseous to take place during both bubble formation at the
mass transfer. This model gave a reasonable expla- nozzle tip and bubble ascent. The chemical reaction
nation of the type of the rate equations. Chatterjee rate constant k~ is determined by the trial and error
and Bradshaw48~developed equations relating the over- method in comparing calculation with experiment and
all mass transfer coefficient with liquid and gas phase presented in Fig. 13 by closed circles. It is seen that
coefficients and gave a diagram showing the relative k~ decreases with increasing [%O]+[%S]/2.
importance of liquid phase mass transfer coefficient in 2. Rate of Nitrogen Desorption from Molten Iron by Ar
determining the overall coefficient. The study by Injection51
Chatterjee et al. suggested possibility of solving the The experimental apparatus and method used for
problem of " asymmetry of absorption and desorp- the study of nitrogen desorption were almost similar
tion ". to those of nitrogen absorption. The injected Ar gas-
Mori49~obtained the overall rate equation for gas flow rate was 74N96 Ncm3/min. During each experi-
absorption and desorption in liquid metal expressing ment, an argon-nitrogen gas mixture whose nitrogen
that the overall rate is controlled by the three pro- concentration was controlled to be in equilibrium with
cesses of gas and liquid phase mass transfers and in-
terfacial chemical reaction and showed a diagram
similar to that by Chatterjee and Bradshaw.48~ Re-
ferring to this diagram, one could explain very rea-
sonably the apparent discrepancy in the kinetics
between gas absorption and desorption.
It is to be noted in Fig. 11 that the data of k' ob-
tained by various investigators scatter in low oxygen
concentration range much more than those in higher
oxygen concentration range. This problem was in-
terpreted by Choh et a1.45~who pointed out that the
rate of nitrogen desorption is controlled by liquid
phase mass transfer as well as by interfacial reaction
kinetics.
the nitrogen concentration in the melt was blown onto onto the melt.53)
the melt, so that the reaction through the free surface
could be neglected. From nitrogen concentration vs.
time relations, the efficiency of nitrogen desorption of
injected Ar (f) is calculated. Figure 14 shows typical
examples of the relation between f and the nondimen-
sional nitrogen concentration B (=[%N]/[%N]e),
where [%N]e is nitrogen concentration in molten iron
in equilibrium with nitrogen pressure of 1 atm. A
mixed control model is developed in which mass trans-
fers in both liquid and gas phases and chemical reac-
tion at the bubble-metal interface are considered to
be rate controlling. The chemical reaction rate con-
stant is determined from the comparison of the effi-
ciency of nitrogen desorption between model calcula- Fig. 16. Ratio of the effective surface area of the melt to the
tion and experiment and plotted in Fig. 13 by open cross sectional area of the crucible.53)
circles. As can be seen from Fig. 13, ke determined
from the study of nitrogen desorption agrees well with
that of nitrogen absorption. Very interestingly, ke of the effective free surface area (AS).
thus determined is in good agreement with that ob- On the basis of this preliminary result, Takahashi
tained by Byrne and Belton52)from an experiment of et al.53)made a systematic study for the enhancement
the isotope exchange reaction. Considering the fea- of the rate of nitrogen desorption from molten iron by
ture of the plotting embodied in Fig. 13, one could Ar injection together with blowing onto the melt. It
conclude that both nitrogen absorption and desorp- has been cleared from the comparison between cal-
tion in molten iron are controlled by mass transfer culation and experiment that the effective free surface
and chemical reaction and that k~ obtained rather area of the melt is about 2.1 times as large as the
indirectly from the comparison between experiment cross sectional area of the crucible (Ae) (Fig. 16).
and calculation is the real chemical reaction rate con- Furthermore, Takahashi et al.54)have observed that
stant. the rate of nitrogen desorption using A1203 crucible
3. Rate of Nitrogen Desorption from Molten Iron by is lower than that using MgO crucible. Deoxidation
Argon Injection Together with Blowing onto the Melt53) with Al also lowered the rate. In these cases, inter-
In earlier stage of the progress of the study of nitro- facial resistance was found to exist at the free surface
gen desorption presented in the previous section, of the melt, retarding the rate of nitrogen desorption
Takahashi et a1.53)got a preliminary result shown in through the free surface.
Fig. 15. The open circles represent the data for ni-
trogen desorption by Ar injection together with blow- Iv. Kinetics of Slag-Metal Reactions
ing onto the melt. The closed circles represent the
experimental result with Ar blowing onto the melt.43) 1. Slag-Molten Metal ReactionswithoutFormation of CO
The dotted line shows the change of nitrogen concen- Slag-metal reactions at high temperatures, partic-
tration calculated for the desorption through the bub- ularly those without formation of CO, are generally
ble interface. It is clear that the rate of nitrogen assumed to be controlled by the rates of transport of
desorption with Ar injection is enhanced markedly by reactants and products in the slag and metal phases.
Ar blowing onto the melt. Possibly, a high degree of Brimacombe and Richardson,55) in a study on the
turbulence is produced from the appearance of bub- mass transfer in the reaction of Tl(in Pb)-Pb2+ (in
bles at the surface of the melt, leading to the increase fused salt), have calculated from the estimation of the
(254) Transactions ISIJ, Vol. 28, 1988
dt Vm
where, A : the interfacial area
Vm: the metal volume
ks1: the apparent mass transfer coefficient of
silicon.
In Fig. 17 ks1 is plotted against the rotational speed
of the stirrer, R. ks1increases just in proportional to Fig. 18. Relation between log [%Si] and time with various
the square root of R (rpm). This strongly supports [%Sl]0,59)
the view that the reaction is of mass transport con-
trolling. an oxidation reaction of a strong deoxidizer dissolved
Mori et a1.,58~in the course of kinetic study of reac- in liquid metal. Therefore, the above study may be
tion (13), have observed a tendency for the apparent generally related to kinetic mechanisms of reactions
mass-transfer coefficient in the metal phase (ks1) to in- between molten slag and molten metal containing
crease with lowering of Si concentration in the metal. solute elements which interact strongly with oxygen.
To investigate the mechanism of this phenomenon a Pluschkell et a1.60~investigated kinetics of oxidation of
series of experiments has been done by Hirasawa aluminum in steel melts by iron and manganese oxides
et a1.,59~where initial concentration of Si ranged from in slag during Ar-injection stirring. They analyzed
0.01 to 0.14 %. Typical results are shown in Fig. the rate data on the basis of the assumption that a re-
18. It is indicated that ks1 is almost constant at action zone was formed within the metal-side bound-
relatively higher initial Si concentration ([%Si]0), ary film. Though the paper of Pluschkell et al. lacks
while at lower [%Si]0, ks1increases with decreasing detailed examinations on the reaction zone, the paper
[%Si]0. As the slag-metal reaction proceeds, k should be regarded as a notable report that proposed
tends to increase with decrease of Si concentration. a highly interesting problems on slag-metal reactions.
The above experimental result could be explained
by employing a new concept that the zone of reaction 2. Slag-Molten Iron Reactionswith Formation of CO
between Si and 0 is located within the metal-side The reactions treated below are concerned with
boundary layer. The location of the zone is sup- slag-molten iron reactions under reducing conditions.
posedly controlled by mass-transfer fluxes of oxygen One finds that even apparently simple reactions are
from the slag-metal interface to the reaction zone and complicated by the occurrence of side reactions and
silicon from the bulk metal to the zone. formation of CO. Extensive studies have been made
The slag-metal reaction (13) can be regarded as on the kinetics of desulfurization of carbon-saturated
Transactions ISIJ, Vol. 28, 1988 (255)
iron and reduction of iron oxide in slag by carbon in ence of a gas film at the slag-metal interface. It is
iron. proposed that the overall reaction of Eq. (15) takes
A very noteworthy paper was published in 1956 by place as the sum of the following reactions:
Ramachandran et a1.61) in which the rate of sulfur
transfer between carbon-saturated iron and oxide slag FeO (in slag) + CO (g) = Fe + C02 (g)
was measured with concurrent measurement of CO (16)
evolution. The most salient result is that, while sulfur and
transfer from the metal to the slag phase is occurring, C02 (g)+C = 2C0 (g) (17)
the silicon content of the metal initially decreases to
that below the equilibrium value and iron is initially At iron-oxide contents below about 2.5 % the rate-
transferred to the slag but again tends to decrease. controlling step is the chemical reaction represented
Ramachandran et al. interpreted their results in terms by Eq. (17), while at higher iron-oxide contents the
of an electrochemical mechanism. Hemptinne et a1.62) rate-controlling step is that of Eq. (16). Sommerville
gave an explanation to the result of Ramachandran et al. showed reasonable agreement between the value
et a1.61) by postulating the potentials and electrical of the rate constant of the iron-oxide reduction at
currents flowing across the slag-metal interface. Oh- lower iron-oxide contents and that of decarburization
tani63) and other authors have proposed explanations of molten iron obtained by Samnand Belton35)referred
of kinetics of slag-metal reactions using electrochem- to in Chap. III of the present paper. It appears that
ical theories. However, as Kawai64) pointed out, the the agreement mentioned above has provided a strong
electrochemical theory does not produce a fruitful ex- evidence in support of the reaction mechanism pro-
planation because of lacking in the knowledge on posed by Sommervi.lle et al.
polarization curves of single reactions.
Recently, Robertson et al.65) has developed a gen- 3. Rate of Mass Transfer betweenMolten Slag and Metal
eral kinetic model for multicomponent slag-metal and underGas InjectionStirring
slag-metal-gas reactions, assuming that the rates are In recent years, the kinetics of reactions between
controlled by multicomponent transport processes in slag and molten iron with gas injection in steel refining
the metal and slag phases. The model has given a processes have become of increasing interest to the
reasonable simulation of the sulfur transfer studied by metallurgists. Previously studies were made by Rich-
Ramachandran et a1.61) ardson and co-workers on elucidating the fundamen-
Mori et al.66) have made a study on the rate of tal relation between the slag-metal reaction rate and
desulfurization of liquid iron and iron-carbon alloys gas injection stirring conditions with using model re-
by CaF2-based slags. During desulfurization of iron action systems at low temperatures, i.e., aqueous solu-
the concurrent transfer of iron, similar to the phe- tion-amalgam system68'70'71) and molten salt-molten
nomenon found by Ramachandran et al., was ob- lead system.69-71)From the practical viewpoint, mod-
served. The results could be explained by a modified el studies have been made frequently with using or-
mass transfer model in which CO gas evolution was ganic solvent-water system.72) This kind of model
assumed to be controlled by a chemical reaction and studies may produce questionable results for the ap-
the transfers of other components were controlled by plication to actual slag-metal systems because of the
mass transfer. low interfacial tension in organic solvent-water sys-
The above studies by Robertson et a1.65) and by tems.
Mori et al.66) have revealed that the phenomena in These years, model experiments have been done in
sulfur transfer between slag and metal phases first the author's laboratory to investigate the role of gas
observed in the middle of fifties by Ramachandran injection stirring in slag-metal reactions at high tem-
et al. could be explained reasonably after three decades peratures with using a molten slag-Cu system where
by mass transfer models. physical properties are similar to those of slag-molten
Kinetics of the reduction of iron oxide from molten iron system.73) Concerning this slag-Cu reaction sys-
slaps by carbon in iron is interesting practically in tem mention has been made in Sec. IV. 1. The ex-
connection with reactions occurring in the hearth plored reaction is oxidation of Si by FeO (Eq. (13) )
zone in the blast furnace. In a paper most recently at 1250°C, taking place under the condition of rate-
published by Sommerville et a1.,677 it is pointed out controlling by Si transport in the metal phase. The
that in almost all previous studies the experimental slag-metal bath was stirred by Ar gas injected through
systems were rather complicated for obtaining clear- a nozzle located at the crucible bottom. The gas-
cut conclusions for the reaction mechanism. This flow rate (Vg), crucible diameter (dr) and metal depth
was mainly due to the presence of graphite. Hence, (hM) were varied.
in the work by Sommerville et a1.67) the reaction sys- Figure 19 shows typical experimental results for the
tem was selected so that only the simple reaction relation between the logarithm of the Si concentra-
tion and time. Here, h81is the slag depth and Vg is
FeO (in slag) + C = Fe + CO (g) ............(15)
expressed for the conditions of 1250°C and 1 atm
proceeded. The reduction rate was measured by the pressure. All the data are represented by linear lines
observation of the pressure change in the system. In predicted from Eq. (14). From the slope of the linear
attempting to explain the reaction mechanism, they line the apparent mass transfer coefficient of Si, ks1, is
devoted special attention to the possibility of the pres- calculated.
Review
(256) Transactions ISIJ, Vol. 28, 1988
Fig. 20. Relation between ks1 and gas-flow rate (d~= Fig. 21. Relation between ks1 and gas-flow rate (d~=
4 cm).73) 7.5 cm).")
seen from Fig. 22, ks1is proportional to h112below a cient of Si, ks1, and injection depth of metal phase,
correlated successfully by one theoretical equation. Fig . 25. Solidification of pig iron on cold scrap.7s)
It is found that the mass transfer data of practical
ladle desulfurization refining could be explained rea- assumed from the results of fundamental studies by
sonably by the present correlation equation.76) Nomura and Mori.7s~
The process of scrap melting can be broken into
V. Kinetics of Solid-Liquid Metal Interactions two stages. In the first stage, some molten iron
freezes on the scrap surface and then the frozen iron
1. Scrap Melting melts back until the original interface appears to the
The rate of scrap melting has important connec- molten bath. As suggested from Fig. 25, the time of
tions with end point temperature control in LD steel- solidification of iron on cold scrap is rather short.
making furnace and power consumption in electric In the second stage, melting of scrap proceeds first
furnace steelmaking processes. This problem has with carburization at the scrap surface and then with
aroused considerable attention early in western coun- subjecting of the scrap to the bath of temperature
tries.77) Recently, in Japan scrap melting rate is dis- higher than the true melting point of the scrap. Mori
cussed in developing new LD processes utilizing a and Nomura, assuming that the time for the first stage
large amount of scrap. is negligible, calculated the relation between scrap
From academic point of view, scrap melting pro- melting ratio and time (Fig. 26). It is indicated that
vides an interesting phenomenon in which simul- the model calculation gives a general agreement with
taneous heat and mass transfers occur. Mori and practical data in the case of simple form of scrap.78~
Nomura78~on the basis of the concept of transfer phe- The feature of the phenomena implied in Fig. 25
nomena, developed a mathematical model for scrap have an important meaning in the kinetics of melting
melting in the steelmaking process. Steady state dissolution of deoxidizers, i.e., A1,for steel. This field
equations are postulated for heat and mass transfer in has been extensively examined by Guthrie.80~
the boundary layer between solid and liquid. Heat
conduction in the solid is expressed by an unsteady 2. Dissolution of Solid Iron in a Carbon-saturatedLiquid
state equation. The values of the thickness of the Iron with Evolutionof CO
boundary layer and the heat transfer coefficient are The kinetics of solution of solids into liquid accom-
(258) Transactions ISIJ, Vol. 28, 1988
panied with evolution of gas is an important and cylindrical specimen was stationary in one series of
interesting problem and has been studied first by experiments; in another series it was rotated.
Mori and Sakuraya.81) The experiments consisted of It has been well established that the rate-control-
measuring the rate of dissolution of iron cylinders con- ling step in the isothermal dissolution of solid iron into
taining various amounts of oxygen immersed in the liquid iron-carbon alloys is mass transfer in the liquid-
melt of carbon-saturated iron. During dissolution side boundary layer. During dissolution of specimens
CO was evolved from the reaction between carbon in containing oxygen, CO bubbles are considered to be
the melt and oxygen in the iron specimen. The produced in the boundary layer. Therefore, the ef-
fect of bubble formation on the mass transfer rate will
be reflected on the value of mass transfer coefficients.
In the case of stationary dissolution, oxygen in the
iron specimen, that is, the evolution of CO was found
to have a marked effect on the dissolution rate. In
Fig. 27, the mass transfer coefficient, k, is plotted
against the rate of CO evolution per unit area of the
specimen surface, 'co. The highest value of k at
1400°C are 73 times as large as the lowest value at
1200°C. At lower temperatures, k increases with Vco
to a power of 0.3. At higher temperatures, k is pro-
portional to VcoS. This phenomenon is shown to be
analogous to the mass transfer at a gas evolving elec-
trode.
Figure 28 is the data for rotational dissolution at
1 250°C. At lower rpm the mass transfer coefficient
is controlled mainly by the extent of CO evolution.
Fig. 26. Typical relations between scrap melting ratio and At higher rpm (rpm>250), the effect of rotation be-
time.79~ comes predominant. Similar relationship has been
observed in the reduction of rotated solid Cr203 by
liquid Fe-Cr-C alloys as mentioned in the next sec-
tion.
The present result provides a reasonable explana-
tion on the melting of prereduced iron particles in an
electric arc furnace in which the utilization efficiency
of electric power decreases with the metallization ratio
of the charged material.82~
3. Reductionof Cr203 by Carbon Dissolvedin Liquid Fe-
Cr Alloys
Knowledge on the rate controlling mechanism of
reduction of Cr203 by carbon is of great importance
in understanding events occurring in the AOD pro-
Fig. 27. Mass transfer coefficient res. CO evolution rate for cess of stainless steel refining and the smelting reduc-
stationary dissolution at various temperatures.81~ tion of chromite ore. The kinetics of Cr203 reduc-
tion has been studied by Fruehan,83~ Sevinc and
Elliott,84~and Suzuki and Mori.85~
Suzuki and Mori85~measured the rate of reduction
of rotated Cr203 cylinder by carbon dissolved in Fe-
10%Cr melt at 1 580°C.
Figure 29 shows the relation between dissolution
rate of the cylinder and rotational speed at various
carbon contents. At the initial carbon concentration
of 0.3 %, where evolution of CO bubbles from the
cylinder-melt interface did not occur, dissolution rate
increases gradually with increasing rotational speed.
At carbon contents above 0.3 %, CO bubbles were
observed to evolve. Here, the effect of rpm becomes
to a lesser extent with increasing carbon content.
For the dissolution of rotating cylinder into liquid
without gas evolution, the mass transfer coefficients,
Fig. 28. Mass transfer coefficient res. rpm for rotational dis- in terms of dimensionless correlations, are compared
solution of iron cylinders of various oxygen contents with those for various non-metallic and metallic sys-
at 1250°C.81~ tems (Fig. 30). From the figure, it is clear that ex-
Transactions Is", Vol. 28, 1988 (259)
cept for the two points in the lower range of Re, the Fig. 30. Dimensionless correlation for dissolution of rotat-
data for the Cr203 reduction are in reasonable agree- ing cylinder into liquid phase.85)
10) M. Sano and K. Mori : Trans. Iron Steel Inst. Jpn., 20 44) K. Narita, S. Koyama, T. Makino and M. Okamura:
(1980), 675. Tetsu-to-Hagane, 57 (1971), 2207.
11) R. J. Fruehan: Ironmaking Steelmaking,3 (1976), 33. 45) T. Choh, T. Moritani and M. Inouye: Tetsu-to-Hagane, 64
12) K. Mori, Y. Ozawa and M. Sano: Metall. Trans. B, lOB (1978), 701.
(1979), 679. 46) T. B. King, R. A. Karesev and P. Dastur: Heterogeneous
13) K. Mori, Y. Ozawa and M. Sano: Trans. Iron Steel Inst. Kinetics at Elevated Temperature, ed. by G. R. Belton and
Jpn., 22 (1982), 377. W. L. Worrel, Plenum Press, New York, (1970), 409.
14) Y. Ozawa and K. Mori: Trans. Iron Steel Inst. Jpn., 23 47) M. Sano, K. Mori, M. Matsushima and K. Suzuki: Tetsu-
(1983), 764. to-Hagane, 58 (1972), 254.
15) Y. Ozawa and K. Mori: Proc. 4th Tech. Conf. ISS, 48) A. Chatterjee and A. V. Bradshaw: Metall. Trans., 4
RIME, Chicago, (1984), 15. (1973), 1359.
16) Y. Ozawa and K. Mori: Trans. Iron Steel Inst. Jpn., 26 49) K. Mori: Iron and Steel Handbook (Tekko Binran), I:
(1986), 291. Fundamental Theory of Iron and Steel, ISIJ, ed., ISIJ,
17) M. Ishibashi, K. Shiraishi, S. Yamamoto and M. Shimada: Tokyo, (1981), 171.
Tetsu-to-Hagane, 61 (1975), 5111. 50) M. Sano, K. Kadoguchi and K. Mori: Trans. Iron Steel
18) A. E. Wraith and M. E. Chalkly: Advances in Extractive Inst. Jpn., 24 (1984), 825.
Metallurgy, ed. by M. J. Jones, IMM, London, (1977), 27. 51) M. Takahashi, H. Matsuda, M. Sano and K. Mori:
19) E. 0. Hoefele and J. K. Brimacombe: Metall. Trans. B, Trans. Iron Steel Inst. Jpn., 27 (1987), 626.
lOB (1979), 631. 52) M. Byrne and G. R. Belton: Metall. Trans. B, 14B (1983),
20) L. Farias and D.G.C. Robertson: Injection Phenomena 441.
in Extraction and Refining, compiled by A. E. Wraith, 53) M. Takahashi, H. Ookuma, M. Sano, K. Mori and M.
Univ, of Newcastle upon Tyne, (1982), E1-25. Hirasawa : Trans. Iron Steel Inst. Jpn., 27 (1987), 633.
21) M.J. McNallan and T. B. King: Metall. Trans. B, 13B 54) M. Takahashi, Han Ye-Tao, M. Sano, K. Mori and M.
(1982), 165. Hirasawa : Trans. Iron Steel Inst. Jpn., 27 (1987), 866.
22) S. Sakaguchi, M. Ishibashi, M. Yoshii, S. Yamamoto and 55) J. K. Brimacombe and F. D. Richardson: Trans. Inst. Min.
H. Kajioka: Tetsu-to-Hagane, 63 (1977), 5534. Metall. Sect. C, 82 (1973), C83.
23) T. A. Engh, H. Tveit, H. Bertheussen, P. Stromsnes, K. 56) R. Prange, K. Heusler and K. Schwerdtfeger: Metall.
Venas and T. M. Svartas: Scand. J. Metall., 5 (1976), 21. Trans. B, 15B (1984), 281.
24) K. Marukawa, S. Anezaki, Y. Tozaki and N. Hiroki: 57) K. Schwerdtfeger and R. Prange: Proceedings of Second
Tetsu-to-Hagane, 65 (1979), 5154. International Symposium Metallurgical Slags and Fluxes,
25) S. K. Sharma: Iron Steelmaker, 14 (1987), No. 3, 73. ed. by H. A. Fine and D. R. Gaskell, RIME, New York,
26) T. A. Engh, H. Sandberg, A. Hultkvist and L. G. Norberg: (1984), 595.
Scand. J. Metall.,1 (1972), 108. 58) K. Mori, M. Hirasawa, M. Shinkai and A. Hatanaka:
27) Y. Ozawa and K. Mori: Trans. Iron Steel Inst. Jpn., 23 Tetsu-to-Hagane,71 (1985), 1110.
(1983), 769. 59) M. Hirasawa, M. Matsu-ura and K. Mori: Trans. Jpn.
28) K. Nakanishi, T. Fujii and J. Szekely: Ironmaking Steelmak- Inst. Met., 28 (1987), 507.
ing, 2 (1975), 193. 60) W. Pluschkell, B. Redenz and E. Schurmann : Arch. Eisen-
29) K. Mori and M. Sano : Tetsu-to-Hagane, 67 (1981), 672. huttenwes., 52 (1981), 85.
30) M. Sano and K. Mori : Trans. Iron Steel Inst. Jpn., 23 61) S. Ramachandran, T. B. King and N. J. Grant : J. Met.,
(1983), 169. 8 (1956), 1549.
31) S. Asai, T. Okamoto, J. C. He and I. Muchi : Tetsu-to- 62) X. Hemptinne, H. Eyring and T. Ree: Physical Chemistry
Hagane, 68 (1982), 426. of Process Metallurgy, RIME, New York, (1959), 65.
32) P. A. Distin, G. D. Hallett and F. D. Richardson: J. Iron 63) M. Ohtani: The 14th Nishiyama Memorial Seminar,
Steel Inst., 206 (1968), 821. ISIJ, Tokyo, (1971), 1.
33) N. H. El Kaddah and D.G.C. Robertson: J. Colloid and 64) Y. Kawai : Tetsu-to-Hagane, 70 (1984), 1640.
Interface Sci.,15 (1977), 349. 65) D.G.C. Robertson, B. Deo and S. Ohguchi: Ironmaking
34) H. Nomura and K. Mori : Trans. Iron Steel Inst. Jpn., 13 Steelmaking, 11 (1984), 41.
(1973), 265. 66) K. Mori, Y. Furuno and Y. Kawai: Process Technology
35) D. R. Samn and G. R. Belton: Metall. Trans. B, 7B (1976), Proceedings, 5th Int. Iron Steel Congress, 6 (1986), 705.
235. 67) I. D. Sommerville, P. Grieveson and J. Taylor: Ironmaking
36) R. J. Fruehan and L.J. Martonik: High Temp. Sci., 3 Steelmaking,7 (1980), 26.
(1971), 244. 68) W. F. Porter, F. D. Richardson and K. N. Subramanian:
37) A. W. Cramb and G. R. Belton: Metall. Trans. B, 12B Heat and Mass Transfer in Process Metallurgy, ed. by A. W.
(1981), 699. Hills, Inst. Min. Met., London, (1967), 79.
38) K. Suzuki and K. Mori : Trans. Iron Steel Inst. Jpn., 17 69) J. K. Brimacombe and F. D. Richardson: Trans. Inst. Min.
(1977), 136. Metall. Sect. C, 82 (1973), C63.
39) K. Suzuki, K. Nakabayashi and K. Mori: Tetsu-to-Hagane, 70) D.G.C. Robertson and B. B. Staples: Process Engineering
63 (1977), 1470. of Pyrometallurgy, ed, by M. J. Jones, Inst. Min. Met.,
40) R. D. Pehlke and J. F. Elliott: Trans. Metall. Soc. AIME, London, (1974), 51.
227 (1963), 844. 71) F. D. Richardson, D.G.C. Robertson and B. B. Staples:
41) M. Inouye and T. Choh: Trans. Iron Steel Inst. Jpn., 8 Proceedings Darken Conference on Physical Chemistry in
(1968), 134. Metallurgy, US Steel Corp. Research Lab., Monroeville,
42) S. Ban-ya, T. Shinohara, H. Tozaki and T. Fuwa: Pro- (1976), 25.
ceedings of International Conference on the Science and 72) S. Asai: The 100th-lOlst Nishiyama Memorial Seminar,
Technology of Iron and Steel, II, Suppl. Trans. Iron Steel ed, by ISIJ, Tokyo, (1984), 65.
Inst. Jpn., 11 (1971), 538. 73) M. Hirasawa, K. Mori, M. Sano, A. Hatanaka, Y. Shima-
43) K. Mori and K. Suzuki: Trans. Iron Steel Inst. Jpn., 10 tani and Y. Okazaki: Trans. Iron Steel Inst. Jpn., 27 (1987),
(1970), 232. 277.
Review
Transactions ISIJ, Vol. 28, 1988 (261)
Review