PROCEEDINGS : CAMME - 1996

ONML JAMSHEDPUR, PP. 01-08

Dynamic refining control model for

LD converter
B. V. SUBBA RAO and N. B. BALLAL*
Technology Department, Mishra Dhatu Nigam Limited, Hyderabad - 500 058
* Department of Metallurgical Engineering, Indian Institute of Technology,
Mumbai-400 076

ABSTRACT
A mathematical model for the dynamic refining control of the LD Converter is
presented. Equations representing the reaction kinetics are formulated using
the oxygen demand function. The amount of oxygen accumulated in LD
Converter, which is calculated from the oxygen balance by using exhaust gas
data, corresponds to the amount of FeO, Fe203 and Mn0 in the slag during
blowing. With the control of the refining reaction with oxygen balance carbon,
phosphorus and manganese contents of steel and the total iron content of slag
at blow end were controlled at preferable levels and their fluctuations were
remarkably reduced. The equations were coupled with appropriate thermal
balance equations. The resulting set of non-linear first order equations have
been numerically solved to predict the reaction path in the converter as a
function of the operating conditions like oxygen blowing rate, lance height,
iron ore, limestone additions rates etc. With continuous on line data feeding
from the converter facilities like sublance systems and waste gas analysers the
entire converter operation could be automated. The model works on a simple
micro computer and could be easily incorporated on to the on-line control
circuit of the Converter.

INTRODUCTION
The extensive use of automatic-control techniques for steel making
processes has expanded considerably. in recent years. This has beeii—rriade
possible because of improved expertise in the data acquisition from the
complex high temperature processes and consequently, there has been an
added stimulus to develop more accurate mathematical models to simulate
process stages. At present there are two basic types of models that are applied
to oxygen steel making processes, known as 'Static' and 'Dynamic'
mathematical models simulating process behaviour.
Static models comprise a set of heat and mass balance equation that enable
the prediction of material demand from known initial conditions of charge
1
 B. V. SUBBA RAO and N. B. BALLAL

composition, weight and temperature to give the required end-point

specification of steel carbon content, temperature and weight. In contrast
Dynamic models['-31 comprise differential equations which describe the rate of
change of bath and slag composition throughout the blowing stage and are
designed to take into account variations in blowing conditions and sequence of
additions as continuous sensors feed back data to the control system throughout
the process. Thus, it is within the capability of the Dynamic model to prescribe
the appropriate corrective action to ensure that the trajectory of the process
path achieves the required end-point target for steel temperature and carbon
content in particular,

The reactions taking place in the steel making process can be divided into
sequential steps as mass transfer of reactants to the phase boundary, interfacial
chemical reaction, and mass transfer of reaction products away from the phase
boundary. At the temperatures prevailing during steel making, the chemical
reaction step is much faster than the other step described above and the refining
reactions are effectively limited by mass transfer. In this paper an attempt has
been made to develop a Dynamic model which takes into consideration the
changes in bath temperature on reaction kinetics and predicts the composition
of slag, metal and gas as a function of time as the blowing progresses. In a true
control system the output/predictions of the model can be compared to that of
the actual composition of the metal in the bath with the help of sensors and
corrective actions taken on-line.

THEORY

According to the multicomponent mixed transport control theory'41, an

instantaneous Thermodynamic equilibrium is assumed to exist at various
metal-gas-slag interfaces and the rates of reaction are governed by transport of
species present to or away from the interface.

Refining Reactions

During LD steel making, the oxygen required for the refining reactions is
supplied in the gaseous form and both metal and slag are initialy oxidised
1/2 02 (gas) <=> [0]
Fe + [0] <=> (FeO)

The actual distribution of Oxygen between slag and metal is easily deter-
mined, since it is a function of a number of variables, including lance height
and oxygen flow rate. The principal refining reaction is, the removal of
carbon
 B. V. SUBBA RAO and N. B. BALLAL

[C] + [0] <=> CO(gas)

[C] + (Fe0) <=> CO(gas) + Fe

The basic thermodynamic data for the above reactions is well established,
and the equilibrium carbon and oxygen contents may be readily calculated for
all the temperatures and pressures encountered during steel making's].

Decarburisation Mechanism

Decarburisation is assumed to take place principally in the slag-metal-gas

emulsion, which occupies a large fraction of the converter volume during a
major part of the blow. Oxygen is initially delivered to the bath via the jet
impingement zone. This oxygen enriched material is then swept away from the
interaction zone by the combined effect of rising CO bubbles, natural convec-
tion and the absorption of energy from the oxygen jet, thus exposing a further
layer of metal to the oxygen jet, and the pattern is repeated. After the formation
of the considerable amount of oxidized metal over the bath, the jet helps in
dispersing metal droplets in the slag phase. Refining of these metal droplets
takes place as they travel in the gas rich slag-metal emulsion and join back the
bulk metal. Oxygen is transferred to the droplets by unsteady state diffusion
and a high degree of supersaturation with respect to the [C] + [0] —> CO
reaction can be achieved.

Heterogeneous nucleation of carbon monoxide occurs when the droplet

comes into contact with rising CO bubbles. Alternatively homogeneous
nucleation of CO is possible, if the oxygen supersaturation is sufficiently high.
Homogeneous nucleation may take place explosively thus creating a finer
dispersion of the metal. In order to reduce the carbon level significantly, each
droplet must pass through the complete cycle several times.

Silicon Removal

In accordance with thermodynamic predictions, the removal of silicon is

usually completed relatively early in the blow. The reactions may be
represented by the following equations :

[Si] + 2[0] <=> (Si02)

[Si] + (Fe0) .t=> (Si02) + 2Fe

According to the above reactions all the oxygen blown in the bath reacts
with silicon during the initial period of blowing. In the model it is assumed that
 A

B. V. SUBBA RAO and N. B. BALLAL

silicon is dropped till the bath silicon reaches a equilibrium level or till that
time when the slag layer of specified thickness is formed. After the formation
of slag layer the droplets generated due to the impingment of the jet disperses
into the slag and the droplet-slag reactions are started. For the simplicity of the
model, it is assumed that the droplets of equal sizes are formed.

The droplet-slag interface is considered to be the prime location for the

oxidation reactions taking place, i.e. oxidation of Silicon, Manganese, Carbon
and Phosphorus. It is assumed that the lime dissolution takes place in the slag
according to the concentration of Silica and the pre-set basicity (2.5 to 3.0) of
the slag is maintained.

For the purpose of the calculation of the amount of each element oxidised
it is assumed that in a particular time step whatever is injected reacts with Fe
forming Fe0 initially and the Fe0 is distributed equally among the droplets
ejected. This Fe0 in the droplets/slag in turn reacts with the elements C, Si, Mn
and P depending on their respective thermal equilibrium with Fe0 and rest of
the elements. All these reactions are assumed to be taking place during the
travel of the droplet in the slag-gas-metal emulsion.

MODEL FORMULATION

According to the the multicomponent mixed transport control theory an

instantaneous thermodynamic equilibrium is assumed to exist at various metal-
slag-gas interfaces and the rates of reactions are governed by the transport of
the species present to or away from the interface.

The concentration of species present is obtained from equilibrium at the

interface for the reactions :

2 (Fe0) + [Si] 2 Fe + (Si02)

(Fe0) + [Mn] <=> Fe + (Mn0)

(Fe0) + [C] <4. Fe + CO(gas)

5(Fe0) + 2[P] <=, 5Fe + (Pp)

For each of the above reactions at the slag-metal interface it is possible to

write the relevant equilibrium constants which relate the interfacial
concentrations of components, since the reactions are assumed to take place
quickly, at the high temperatures of steel making. Equilibrium constants for the
above mentioned reactions can be written as

-••■
••alis
—`6011L
 B. V. SUBBA RAO and N. B. BALLAL

ySi02 . X' Si02

K. ... 1
yFe0215,.X'Fe02.X'Si
yMnO. Mn0
Km. ... 2
yFe0151..X'Fe02.X‘Mn
PCO
Kc ... 3
' yFe0.fc.X‘Fe0.X'C

1P205 . X' P205

Kp ... 4
yFe05.fp2.X`FeV.X2p2

Heterogeneous reactions such as the above can be divided into sequential

steps, mass transfer of reactants to the phase boundary, interfacial chemical
reaction and the mass transfer of reaction products away from the phase
boundary. As mentioned earlier the chemical reaction step is much faster than
the other involving diffusion, and that the refining reactions are effectively rate
controlled/limited by the mass transfer.
Knowing the rate of mass-transfer of each of the reactants (FeO, C, Si, Mn
and P in the model), it is possible to calculate the bulk-metal composition at
a particular time during blowing.
The mass transfer coefficient of the individual elements is calculated using
the Rang-Marshall correlation [6].
Nu = 2.0 + 0.6(Re)la (Sc)1/3
The change in the bulk composition over a time interval 'dt' is given by the
conservation equation. Applying the conservation equation for the four
elements considered in the model.
- .
— (nSi) = (X`nsi- X's,) 5
dt Vm Ai
d - Dmn . Ai
- (nMn)– Mum.- X11,0 ... 6
dt Vm
d Dc .
— C) = (Xmc- X'c) ... 7
dt C Vm
d - Dp . A'
- (°p) (Xm- Xp) ... 8
dt Vm
5
 B. V. SUBBA RAO and N. B. BALLAL

Similarly rate controlling equation for the transfer of Fe0 to the interface
can be written as
d-Dpeo .A
— O). (Xspco- X'F.0) ... 9
dt OR Vs

The mole balance equation for the elements Si, C, Mn and P against FeO
consumed can be written as the following :
A'.Dt
(X5Fc0- X'Fc0) = + n - 111,4241'c cI1-2.5(ngn-n0
Vs

From the above equation

Vs
X1 [2(n '"-n + nm.lu - nm. ner+2.5(npin-npf))+XF.0
... 10
For the transfer (oxidation) of each element from metal phase to slag phase
the corresponding increase in Oxides in slag phase can be written as :

nSiO2f= nSi'n - nSi' + nSiO2n1 11

nMn0f= nMnm - nMnf + nMnOg" 12
nCaOf = 3(nPOr + nSiO2f) 13
np0f= nPlr' - nig + nPOin 14

The above 14 equations are solved simultaneously for XiSi, X'Mn, Xip,
Vpco, nCf, nSit, nMnr, npf, nSiO2f, nP0`, nMn0' and nCa0`. At each time step
the mass balance is carried out and the molar composition of slag and metal
phases are modified. The interfacial area which is critical in the above calcu-
lations, is calculated based on the number of droplets generated and dispersed
into the slag-metal-gas emulsion. The number of droplets generated is calcu-
lated from the amount of metal splashed with the momentum of the jet using
the Rang-Rammlers sampling correlationm.

RESULTS
The predicted compositional changes with respect to C and Si during the
course of blowing is shown in Fig. 1. The decarburisation rate curve which is
an indication of the blowing profile in the converter, is predicted using the
above described model in Fig. 2. The model predicted path is similar to the
normal converter practice[3], wherein the decarburisation rate increases to a
maximum in the initial stages, attains a plateau for some time and decreases

6
 B. V. SUBBA RAO and N. B. BALLAL

z
e 2

o " 1 I I I -I

0 100%

Blowing Time % —>

Fig. 1. Predicted compositional change of carbon and
silicon during the blowing practice.
800

700
A
600

4 500

;
1;0 .100 •

F-; 300

200
C
100

Blowing Time % —>

Fig. 2. Predicted decarburisation rate curve for the
particular blowing practice simulated in the
model.

gradually towards the end of the blow. But as could be seen from the results
of decarburisation the decarburisation rate predicted by the model is lower than
the usual practice observed in the industrial practice. This may be due to the
anamoly in the calculation of the weight of the metal splashed due to the
impinging jet and the calculation of the interfacial areas in the model. This
could be overcome by introducing a process parameter into the model which

7
 B. V. SUBBA RAO and N. B. BALLAL

could take care of the deficiency. This process parameter would be dependant
on the parameters like refractory profile in the converter, lance configuration
and the blowing practice. After further fine tuning of the present model with
the industrial data this model could be used on the control circuit of the.
converter.

NOMENCLATURE
D = Diffusion Coefficient f = Henrian Activity Coefficient
K = Equilibrium Constant A = Interfacial Area
X = Mole fraction dt = time step
n = No. of moles Nu = Nusselt Number
V = Volume Re = Reynolds Number
= Activity Coefficient Sc = Schmidt Number

SUPERSCRIPT
M = Metal phase in = initial
S = Slag Phase f = final
i = Slag-metal interface int = interface

REFERENCES
[1] M. Kitamura. et al., 'Dynamic Control of LD Converter at Kakogawa Works of
Kobe Steel Ltd.' Proceedings of the Third International Iron and Steel Making
Congress, Chicago, Illinois, Apr. 1978, pp. 89-94.
[2] Hei-Ichiro Iso et al., 'Dynamic Refining Control by analysis of exhaust gas from
LD Converter', Transactions IR!, Vol. 27, 1987, pp. 351-359.
[3] R. Weeks. 'Dynamic model of the BOS process, Mathematical process models in
Iron and Steel Making', The Metals Society, Amsterdam, Feb. 1973,.Amsterdam,
pp. 103-116.
[4] D.G.C. Robertson et al., 'Multicomponent mixed control theory for kinetics of
coupled slag/metal and slag/metal/gas reactions : application to desulphurisation of
molten iron'. Iron Making and Steel Making; Vol. 11. 1984, pp. 41-55.
[5] R.D. Walker and D. Anderson 'Reaction mechanism in Basic Oxygen Steel
Making', Iron and Steel, June 1972, pp 271-276.
[6] R. Byron Bird. et. al., 'Transport Phenomena', John Wiley & Sons, New York,
1960, pp. 647.
[7] S. C. Koria and K.W. Lange, A new approach to investigate drop size distribution
in Basic Oxygen Steel making, Metallurgical Transactions. B, Vol. 15B, 1984,
pp. 109.