Theoretical Aspects of Scrap Dissolution in Oxygen Steelmaking Converters

G. Sethia, A. K. Shuklaa, P. C. Dasb, P. Chandrab and B. Deoa

a
Department of Materials and Metallurgical Engineering,
b
Department of Mathematics
Indian Institute of Technology Kanpur, 208016
Tel: +91-512-2597256
Fax: +91-512-2597505
Email: {guneet.sethi@alumni., akshuk@, pcdas@, peeyush@, bdeo@} iitk.ac.in

Keywords: BOF, scrap dissolution, analytical, Fourier, finite difference

INTRODUCTION

The scrap dissolution rate in BOF is known to affect the bath temperature, slag formation, foaming of slag and post
combustion in the first 8-10 minutes of blow. Kinetics of scrap dissolution essentially belongs to the class of moving
boundary problems with phase change (Stefan problems). Although literature is replete with solutions of Stefan
problems, a complication in the case of scrap dissolution in BOF is that the bath (liquid metal) temperature and
composition change with time. Several models of scrap dissolution in BOF have been published in literature1-10 and
in each model simplifying assumptions have been made.

The mechanism of scrap dissolution comprises of simultaneous heat transfer and mass transfer (of carbon) in the
melt and inside the scrap. The process of dissolution can be divided into following three parts:
1. “Solidification” of liquid metal on the parent scrap (chill effect)
2. “Fast melting” of the solidified shell
3. “Normal melting” of the parent scrap
The rate of solidification or dissolution may be defined as the product of scrap-melt interface area and velocity of
the interface. The velocity of the scrap-melt interface at any moment depends upon (Figure 1) the thermal
conductivity of the scrap, heat transfer coefficient of the melt, temperature difference between interface and the bulk,
temperature gradient inside the scrap and at the interface and latent heat of melting. The interface temperature and
composition are related to the equilibrium iron-carbon phase diagram.

(a) (b)

Figure 1: Temperature (a) and carbon concentration (b) profiles in scrap and metal

Among all the investigations reported so far the key items of discussion have been the method estimation of
temperature gradient inside the scrap, composition (carbon content of the solidified skin) of scrap and the
composition of the liquid melt at the interface during melting. The theoretical basis of some the important
investigations are critically examined before proceeding to develop an analytical model.

AISTech 2004 Proceedings - Volume II 915

 Hartog et al1 developed an interesting numerical procedure based on solution of simultaneous, unsteady state, heat
transfer and mass (carbon) transfer equations at the bath-scrap interface. They considered moving boundary layer
approach to estimate heat transfer coefficient and Chilton-Colburn analogy to estimate mass transfer coefficient. The
scrap-melt interface temperature and carbon composition was deduced from Fe-C phase diagram. The rates of
energy generation due to chemical reactions were calculated from off-gas analysis and bath sampling. The heat
losses from BOF were calculated by posteriori calculations in real heats. Solutions were obtained for different sizes
of scrap and carbon content. The evolution of temperature of the bath and complete dissolution of the scrap
depended upon the scrap size distribution. In the presence of higher proportion of light scrap, the temperature of the
bath was predicted to be almost near to the liquidus temperature. Dissolution of the colder heavy scrap had to wait
until the light scrap dissolved. The diffusion of carbon inside the solid scrap was not considered. Szekely2,3
considered the diffusion of carbon from melt into the solid scrap and assumed that melting process was controlled
by mass transfer of carbon from the melt. Specht and Jeschar4 assumed bath temperature and composition to be
constant. The diffusion of the carbon inside the solid scrap was considered by using Ficks’ law. Solidification of
metal on scrap (chilling effect) was not considered. The temperature profile at the interface was calculated by
considering the solid as a semi-infinite body. The heat of fusion included an additional term to account for the
energy required to raise the temperature of melting mass from interface temperature to the bulk temperature. The
heat and mass transfer coefficients were estimated by Sherwood and Nusselt numbers for spherical, cylindrical and
laminar-slab shaped solids under forced convection conditions. The time for complete melting was proportional to
the initial size of the scrap. Gaye and Wanin5 considered heat transfer coefficient as a function of input energy to the
system and thus incorporated the effect of blowing conditions on heat and mass transfer coefficients. They also
studied (similar to Hartog et al1) the dissolution behavior for different thickness and scrap mixes. The permissible
limits of scrap size, to ensure complete melting by the time of in-blow sublance measurement, were recommended.
As expected, bottom stirring improved the melting rate significantly. Preheating of the heavy scrap did not improve
its dissolution rate even if blowing time was increased. Yoruchu and Rolls6 employed finite difference procedure
and considered the diffusion of carbon inside the scrap. Diffusion Coefficient depended upon temperature (given by
Arrhenius relationship). They assumed that carbon composition at the scrap surface was same as that of bath and the
temperature as equal to the melting point of the bath.

Asai and Muchi7 presented a detailed model for scrap dissolution in BOF. Temperature and carbon concentration of
the melt were considered as important variables and their influence on progress of the melting of the scrap for
different operating conditions was investigated. They assumed the temperature and concentration profiles only
inside the liquid at the interface. The composition of the solidified shell was assumed to be same as that of the scrap;
this is however not true (see discussions later in the paper). Temperature inside the scrap at any moment was
considered to be uniform and the velocity of the moving interface was considered to be constant; these
approximations are however valid only for the case of thin scraps. Gupta8,9 considered a uniform temperature and
carbon composition inside the scrap but a steep temperature and composition gradient at the interface for a semi-
infinite body (similar to Asai and Muchi7). The composition of the solidified layer was assumed to follow the
changes in bulk liquid metal composition as a function of time. The interfacial velocity was calculated at each time
step but was found to be nearly constant during the major period of scrap dissolution (except towards the end of
dissolution). The effect of various parameters on the peak value of solid (scrap + solidified shell) to liquid ratio, for
a given scrap to hot metal ratio and for different preheat temperatures and heat generation factors, were studied. The
amount of build up of solidified shell was predicted to be up to 33% of the initial scrap mass, which is rather high.

Zhang and Oeters10,11 considered the dissolution of sponge iron and scrap in high carbon melts. For the different
stages of solidification, three simultaneous equations involving heat transfer, carbon mass transfer and carbon-
temperature relationship (as given by liquidus line) were solved for the three unknowns, namely, carbon and
temperature at the interface and the velocity of the moving interface. During frozen shell development, interface
carbon content in liquid and solid phases was related to the Fe-C phase diagram. The thermal boundary layer was
assumed to form as soon as the scrap came into contact with the liquid whereas the development of concentration
boundary layer took some time due to the slow process of mass transfer. It was shown that heat transfer was nearly
forty times faster than mass transfer during the initial period of shell formation. Since the initial solidification
velocity was high due to the steep temperature gradient, the initial frozen layer had the same composition as that of
the melt (decided by Fe-C phase diagram). As the velocity of the shell formation reduced the concentration
boundary layer was established at the interface. The time beyond which contribution of thermal gradient inside the
scrap could be neglected was calculated.

916 AISTech 2004 Proceedings - Volume II

According to some of the earlier investigations12, the contribution of frozen layer to overall melting process can be
ignored. For example, Goldfarb and Sherstov showed that the total time for solidification and fast melting is less
than 2 minutes in the case of a sphere of 20 cm diameter iron rod, for a heat transfer coefficient of 24000 J/m2s.
According to Phelke also, the freezing causes delay of only 30 seconds for melting of bar of 3.7 cm in diameter.
Fatukami et al reported that a 15 cm thick flat plate scrap melts between 16 to 20 minutes. Hills concluded that the
melting rate was very sensitive to the temperature-carbon-time profile and the most important aspect of the blowing
path in BOF is to develop a difference as large as possible, at any temperature, between the carbon content of the
bath and liquidus of the Fe-C system. Thus, by creating optimum conditions inside the BOF, it may be possible to
melt even very thick ingots (71 cm thick) in as little time as 23 minutes; Hills assumed the mass transfer coefficient
of the order of 2.5 x 10-4 m/s and heat transfer coefficient of the order of 20000 J/m2 s.

In the actual BOF process even the heat transfer coefficient varies with time due to changes in fluid flow conditions
imposed by the impinging oxygen jet as well as scrap geometry due to presence of a mixture of heavy scrap and
light scrap of various sizes. In fact it is not possible to simulate the conditions of scrap dissolution in BOF under
laboratory conditions. In the present work theoretical aspects of scrap dissolution are discussed and, for the first time,
an analytical model is developed so as to verify the results of finite difference models or numerical models which
have been normally employed so far.

MODEL DESCRIPTION

Analytical Model
The basic assumptions made in the analytical model are:
1. No mass transfer control during the solidification and fast melting stage.
2. Convective heat transfer in the melt.
∆L
3. v= is constant in the small time step, ∆t . This velocity would be positive in the initial solidification
∆t
and would be negative in the subsequent stages.
dWm dW sc dWsc
4. =− − σ 1 s where = ρAv . The amount of carbon consumed by the CO reaction is
dt dt dt
also included in the weight balance.
5. Since specific heat of scrap would be varying with temperature we would find out its average value using
the interface temperature and center line temperature of scrap. The temperature dependence is as follows:
C psc (T ) = 17.49 + 24.769 × 10 −3 T (J/molK)

Attention will be given to the ratio of solid metal to liquid metal in the early stages of the blow in converter as a
function of hot metal temperature, rate of heat generation due to oxidation reactions, scrap to hot metal ratio in the
charge, as well as the actions taken during the blow (change of lance height and amount and timing of coolant
additions). ). Effect of heat transfer coefficient on kinetics of scrap dissolution and the solid to liquid ratio will be
examined.

Governing Equations
From figure 1, following equations are formulated:

1. Heat Flux Balance at the interface x=0:

∂Tsc ( x, t )
hA(Tm − T ′) = ρA(− ∆H Fe )v − λA (1)
∂x x =0

Here λ = ρC pscα where C psc is the average specific heat for the scrap calculated using one end temperature as the
interface temperature (T’) and the other as the center line temperature is Tsc.

The term ∆ΗFe will have contributions depending on the stage of process:

AISTech 2004 Proceedings - Volume II 917

 • Solidification: The term ∆ΗFe incorporates heat of fusion as well as the heat required to raise the
temperature of liquid metal from interface temperature (T′) to the temperature of melt (Tm).
∆HFe = ∆h+ cp (Tm-T ) (2)
• Fast Melting: The term ∆ΗFe incorporates heat of fusion as well as the heat required to raise the
temperature of scrap layer (Τav) to the temperature of melt(Tm).

∆H Fe = ∆h + C p (Tm − Tav ) (3)

∆L
Tsc ( x, t )∂x
Τav is calculated as: Tav = 0
(4)
∆L

• Normal Melting: Similar to fast melting the term ∆ΗFe incorporates heat of fusion as well as the heat
required to raise the temperature of scrap layer (Τav) to the temperature of melt(Tm).

∆H Fe = ∆h + C psc1 (T ′ − Tav ) + C psc 2 (Tm − T ′) (5)

C psc1 is the average specific heat for the scrap at temperature Τav. C psc 2 is the average specific heat for the
scrap calculated using one end temperature as the interface temperature (T’) and the other as the melt
temperature Tm. Τav is again calculated as in (4)

To solve the heat balance heat equation (1) we need to know the temperature distribution of the scrap, Tsc (x, t). For
one dimensional heat flow, the Fourier Equation states
∂ 2Tsc ( x, t ) ∂Tsc ( x, t )
α = (6)
∂x 2 ∂t
Solving equation (6) (available in Appendix I) we obtain the following solution:

∞
nπx
Tsc ( x, t ) = T ′ + An sin exp(−λ2αt ) (7)
n =1 2L

The limitation of this solution is the fixed boundary temperature and constant dimension which in our case are
changing with the time. Hence we devise a step wise solution procedure in which the whole run time is divided into
several time steps ( ∆t ). The initial form of solution for any time t=m ∆t and tprev= (m-1) ∆t as seen from (7) and
by calculating An, is:

∞
(1 − cos nπ ) i nπx
Tsc ( x, m∆t ) = T ′ + (Tsc − T ′) sin exp(−λ2α ( m∆t )) (8)
n =1
n π 2L
2
where T’ and L are values used from previous time step, (m-1) t.

∂Tsc ( x, t ) ∞
nπ
Hence using (7) or (8) we get = An . exp(−λ2αt ) (9)
∂x x =0 n =1 2L

where An can be evaluated from Appendix I.

918 AISTech 2004 Proceedings - Volume II

Once Tsc ( x, t ) is known, then Tav, v can be obtained from (7). Thus

2L ∞
An nπ∆L
Tav = (1 − cos ). exp(−λ2αt ) (10)
∆L n =1 n π 2 L

1 ∂T ( x, t )
v= h(Tm − T ′) + λ sc (11)
ρ (−∆H Fe ) ∂x x =0

Thus we can solve for the melting velocity at each time step from (11).

2. Mass transfer control in liquid metal:

• Solidification: C ′ =C m (12)
• Fast Melting: No mass transfer control in liquid metal

C’=f(Wsc) (13)

The different portions of skin have different carbon concentration and that is equal to carbon
concentration in melt during the skin formation stage. So we can take Cm values with respect to
thickness in solidification stage and regress the data for polynomial regression. Due to lesser accuracy
of the regressed constants, this caused jump of the carbon concentrations at the interface and melt at
the start of Normal melting stage. We ignore such a jump as it doesn’t cause an occurrence of error in
the calculations.
• Normal Melting: No Mass transfer control in liquid metal

C’=Cm (14)
Mass transfer can be considered to be rate controlling if the following equation is followed:

k (Cm − C ′) = v(C ′ − Csc ) (15)

3. Phase Diagram Equation: The following equation is used for solidus line

T ′ = 1809 − 90C ′ (16)

4. Change in heat content of melt:

d (WmTm ) dWm
• Solidification: Cp = ∆H coσ 1s − hA(Tm − T ′) + C pTm which on simplification
dt dt

dTm
yields C pWm = ∆H coσ 1 s − hA(Tm − T ′) (17)
dt

d (WmTm ) dWsc
• Fast and Normal Melting: Cp = ∆H coσ 1 s − hA(Tm − T ′) − C p Tav which
dt dt
on simplification yields:

AISTech 2004 Proceedings - Volume II 919

 dTm dWsc
C pWm = ∆H coσ 1 s − hA(Tm − T ′) + C p ((Tm − Tav ) + Tmσ 1 s ) (18)
dt dt
5. Change in carbon content of melt:

d (Wm C m ) dWm
• Solidification: = −σ 1 s − βC m which on simplification yields
dt dt

dC m dWm
Wm = −σ 1 s − C m ( β + 1) (19)
dt dt

d (Wm C m ) dWsc
• Fast Melting: = −σ 1 s − C s On simplifying we get:
dt dt

dC m dWsc
Wm = −σ 1 s + ((C m − C s ) + C mσ 1 s ) (20)
dt dt
where Cs=C’ at that instant of time.

• Normal Melting: Cs=Csc in (20) (21)

The scheme of calculation sequence is explained in Figure 2.

(m) t (m) t (m) t

v v (11)
v
(11) (17) (11) (13) (18)

T’ Tm Cm C’ T’ Tm

(19) (16) (18)

(16) (20) (16)
(m-1) t
Cm T’ C’ Cm
C’ Tm
(12) (m-1) t (18) (m-1) t (21)

(a) (b) (c)

Figure 2: Calculation sequence in the simulation used for solving the solidification stage (a) fast melting stage (b)
and normal melting stage (c)

dWsc
The solidification stage will continue till ≥ 0 corresponding to the maximum thickness of the solidified layer.
dt
The fast melting stage will continue till Wsc returns to its initial value when the experiment was started,
corresponding to removal of the solidified layer. The normal melting stage will continue till whole of the scrap piece
melts.

920 AISTech 2004 Proceedings - Volume II

Numerical model
We would like to compare the results of the numerical finite difference method (FDM) only in the normal melting
stage. Hence we would be given the temperature distribution and other such variable values after the solidification
and prior to melting are taken from the analytical solution. The details of the FDM procedure are provided
elsewhere8. More details on the formulation are also available9,15.

RESULTS AND DISCUSSION

Mass Transfer vs. Heat Transfer

Mass Transfer can be considered to be rate controlling if the following equation is followed:

k (Cm − C ′) = v(C ′ − Csc ) (15)

We note that at the end of the melting stage the kξ and v are in comparison and hence we will be able to find out the
value of Cm (which would be near to the value of Csc) such that mass transfer is rate controlling. Thus it is noted that
during the initial period of normal melting heat transfer is rate controlling while in the later end stages when the
carbon concentration has decreased to a large extent then mass transfer is rate controlling. By the analyzing the
sensitivity of the interface temperature with the mass transfer co-efficient we are able to arrive at and appropriate
value of mass transfer co-efficient such that our model works in conjunction with what is expected (total melting
time observed in the actual process), when we assume the process to be heat transfer controlled. We were able to
find out value of such a mass transfer co-efficient that after attaining that value the mass transfer becomes
asymptotic to heat transfer. The value amounts to .005 m/s as seen in Figure 3. It can be seen that the condition of
mass transfer coefficient of 0.005 m/s corresponds to a particular fluid flow situation in BOF. In actual practice the
mass transfer coefficient will change with time, depending upon the operating conditions of bath mixing oxygen
flow, lance height, scrap mixture of different sizes, etc.). Our aim here is to demonstrate the application of analytical
method and compare it with the numerical solution for the same set of initial and boundary conditions.

1850

1800 Mass Transfer

Interface Temperature (T'), K

1750 Heat Transfer

1700

1650
0.0002 m/s
1600
0.0005 m/s
1550 0.0010 m/s

0.0050 m/s
1500

1450
1650 1700 1750 1800 1850 1900 1950 2000 2050
Melt Temperature (Tm), K
Figure 3: Variable mass transfer and heat transfer

ξ
Value from literature

AISTech 2004 Proceedings - Volume II 921

 Fourier Series Solution vs. Finite Difference Solution

Figure 4 compares the variation of the each of the parameters in the Fourier solution (which is the analytical solution)
and the finite difference solution (which is the numerical solution).
Figure 4(a) showing the length of the scrap, clearly depicts the occurrence of stepped profile in finite difference
method. This was expected since FDM is a numerical procedure. Both analytical and FDM have similar average
profiles although incomplete melting is observed in the analytical method. This occurs due to non-equilibrium
nature of the present analytical solution. This is one of the drawbacks of the analytical methods in the sense that the
scrap gets melted at a sharp temperature whereas in the actual case there would be a melting range as would be seen
in the results of the finite difference method. Thus in the analytical solution complete scrap would melt only if the
temperature of the scrap is equal to the melting point. Since length and weight are proportional Figure 4(h) shows a
similar trend.

Figure 4(b) shows that at the end of the cycle the melt velocity, v reached is 0.00035m/s. The melt velocity doesn’t
remain constant but instead decreases, although the decrements are gradual. Thus we see that the interface
movement velocity is not constant as assumed by Asai and Muchi7.

Figure 4(c) and 4(d) demonstrate the variation of interface temperature (T’) and melt temperature (Tm). The
temperature rise is slightly lower in FDM, which implies that more heat is extracted by the scrap in the FDM. This
could possibly be attributed to the existence of a large mushy range. To justify such a large melting range we say
that the heat flow is much easier in FDM compared to the Fourier solution where there is heat concentration. Thus
this forced rising of the temperature caused the melting rate to be higher and the heat flow as such through
conduction into the material is low which leads to formation of a steep profile. No such artificial rise was used in the
case of analytical solution and hence the profile was consistent with the expected (a parabolic profile) which clearly
depicts the power of an analytical solution and weakness of numerical solution.

Figure 4(e) and 4(f) demonstrate the variation of interface carbon concentration (C’) and melt carbon concentration
(Cm). The melt carbon concentration is slightly higher in the finite difference solution which implies that the chances
of FDM to be driven by mass transfer control are less.

Figure 4(g) shows the variation of centre line temperature of scrap. A steep rise is observed in the case of FDM as
opposed to a steady rise in the case of analytical solution. This could be explained on the basis of mass transfer
occurring in the system. The FDM solution doesn’t scope in possibility of mass transfer. In the end the process will
become mass transfer controlled rather than heat transfer controlled. Thus the temperature rise should not be as steep
as observed in the FDM solution. Hence the analytical solution depicts the correct nature of the process. Also, the
heat extraction is by the scrap is more in FDM (Figure 4(c)) and yet the temperature rise of the scrap is smaller. This
is because of the existence of such a large mushy range. Another thing to note is that the temperature at all points at
all times is higher in analytical solution than in the case of numerical solution.

0.25 0.005
Analytical (b) Analytical
0.004
0.2 Finite Difference
melt velocity (v), m/s
length of scrap (l), m

0.003
0.15
0.002
0.1
0.001

0.05
0
(a) 0 200 400 600 800 1000
0 -0.001
0 200 400 600 800 1000
time, s
time, s

922 AISTech 2004 Proceedings - Volume II

 Figure 4(i) and 4(j) show the temperature profiles in the analytical solution and the finite difference solution. The
profile is steep in FDM in comparison with Fourier solution which has a near parabolic profile.

1800 2200
Temperature at interface (T'), K

Melt Temperature (Tm), K

2000
1600
1800

1400 1600

1400
1200 Analytical
Analytical
1200
(c) Finite Difference (d) Finite Difference

1000 1000
0 200 400 600 800 1000 0 200 400 600 800 1000
time, s time, s

5 5
Interface Carbon Concentration
Carbon concentration in melt

Analytical
Analytical
4 4
Finite Difference
Finite Difference
(Cm), wt.%

3 3
(C'), wt.%

2 2

1 1
(e) (f)
0 0
0 200 400 600 800 1000 0 200 400 600 800 1000
time, s time, s

2000 10000
Center line temperature of scrap

Analytical Analytical
Weight of scrap (Wsc), kg

1600 Finite Difference 8000

Finite Difference

1200 6000
(Tsc), K

800 4000

400 2000

(g) (h)
0 0
0 200 400 600 800 1000 0 200 400 600 800 1000
time, s time, s

AISTech 2004 Proceedings - Volume II 923

 Finite Difference Analytical Solution
1800 1800
(i)
(j)

Scrap temperature, K
Scrap temperature, K

1400 1400
increasing
increasing time
time
1000 1000

600 600

200 200
0 0.05 0.1 0.15 0.2 0 0.1 0.2 0.3 0.4 0.5
x, m x, m
time = 0s time = 92.8s time = 338.04s time = 0s time = 92.8s time = 338.04s
time = 538.04s time = 638.04s time = 838.04s time = 538.04s time = 638.04 time = 838.04

Figure 4: Variation of different parameters with time in finite difference solution and analytical solution

Effect of Heat Transfer Co-efficient

As heat transfer into the scrap will increase the melting rate will increase. The higher heat transfer into the scrap will
cause a drop in the temperature of melt as seen in the case of 10 times ‘h’ in figure 5(a). This drop was also observed
in the experimental data by Specht and Jeschar4. The drop of melt temperature negates the effect of higher melting
rate thus restoring the solution towards an equilibrium solution. Overall, the incomplete melting decreases (Figure
5(b)). Also the solidified shell thickness and time decreases which was expected due to steeper rise in the overall
temperature of the scrap.

Recent laboratory experiments have shown that when an iron rod is dipped into carbon saturated molten iron at 1673
K then the frozen (chilled) shell formed on the iron rod is not in perfect contact with the parent iron rod; the SEM
shows a very thin but not totally uniform gap between the iron rod and the frozen shell. This phenomenon will
affect the heat transfer between the frozen shell and the parent iron rod in an undetermined way. For this reason we
have compared the results of analytical model with the FDM model only after the melting of the frozen layer, with
an assumed temperature profile in the scrap. Further experiments are in progress to characterize the formation of
frozen shell so as to see the variation of carbon content along the thickness of the shell.

2050 10000
h = 3630.0 W/m^2K
Temperature of melt (Tm), K

h = 2*3630.0 W/m^2K
Weight of scrap (W sc), kg

1950 8000 h = 4*3630.0 W/m^2K

h = 8*3630.0 W/m^2K
Increasing heat
transfer co-efficient h = 10*3630.0 W/m^2K
1850 6000

1750 h = 3630.0 W/m^2K 4000

h = 2*3630.0 W/m^2K
Increasing heat
h = 4*3630.0 W/m^2K transfer co-efficient
1650 2000
h = 8*3630.0 W/m^2K
h = 10*3630.0 W/m^2K

1550 0
0 200 400 600 800 1000 0 200 400 600 800 1000
time, s time, s

(a) (b)
Figure 5: Effect of heat transfer co-efficient

924 AISTech 2004 Proceedings - Volume II

 SUMMARY

A new model based on analytical solution of heat and mass transfer equations using a Fourier series based solution
of the temperature distribution in the scrap is described. The results are critically compared with the FDM solution.
The velocity of the solid-liquid interface movement is evaluated at each time step. The temperature profile predicted
by analytical solution is likely to be more accurate than the FDM solution because of the assumption of mushy zone
in the latter. The difference in temperature profile between analytical solution and FDM solution is clearly reflected
in the manner of rise of center line temperature. Further experimental work is, however, needed to study the thermal
consequences of lack of perfect thermal contact between the solidified (chilled) layer and the core solid metal
(scrap).

REFERENCES

1. H.W. Hartog, P.J. Kreyger and A.B. Snoeijer, C.R.M, No.37, December 1973 pp.13-21
2. J. Szekely, Y.K. Chuang and J.W. Hlinka, “The melting and dissolution of low carbon iron-carbon melts”,
Metallurgical Transactions, 3, 1972, pp. 2825-2833
3. Y.K. Chuang and J. Szekely, International Journal of Heat and Mass Transfer, 14, 1971, pp.1285-1294
4. E. Specht and R. Jeschar, Steel research, 64, 1993, pp. 28-34
5. H. Gaye, J. Wanin, P. Gugliermina and P Schittly, Proc. 68th Steelmaking Conference, Detroit, U.S.A, April 14-
17, 1985, pp.91-102
6. H. Yoruchu and R. Rolls, Iron and Steel International, February 1976, pp.35-39
7. S. Asai and I. Muchi, Transactions ISIJ, 11, 1971, pp.107-115
8. Brahma Deo, Gaurav Gupta and Manish Gupta, Proc. Asia Steel Int. Conf. Jamshedpur India, April 9-12, Vol.
02, 2003, pp. 2.d.1.1-2.d.1.8
9. G.K. Gupta, B.Tech Report, Department of Materials and Metallurgical Engineering, I I T Kanpur, 1998
10. Liuyi Zhang and Franz Oeters, “Melting and mixing of alloying agents in steel melts”, Verlag Stahleisen Gmbh,
Dusseldorf, 1999, pp.87-91
11. Liuyi Zhang and Franz Oeters, Steel Research, 71, No. 5, 2000, pp. 141-144
12. R.D. Phelke, W.F. Porter, R.F. Urban and J.M. Gains, BOF Steelmaking, Vol. 02, pp. 219-234
13. E. Kreyszig, Advanced Engineering Mathematics, Wiley John & Sons, Incorporated 7th Edition, August 1992
14. J. Szekely, Process optimization, with applications in metallurgy and chemical engineering, Wiley, New York,
1973
15. J.P. Holman, Heat transfer, 9th ed., McGraw-Hill, New York, 2002

NOMENCLATURE

Notation Description Typical values Units

A Area of solid/liquid interface 5.4 m2
Cp Specific heat of melt - J/kgK
C psc Average Specific heat of scrap 0.01 J/kgK
Cm Bath (Liquid Melt) carbon concentration 4.26 wt. % C
C′ Carbon concentration of S/L interface (liquid) - Wt. % C
h Heat transfer coefficient in liquid metal 3.63*1000 W/m2K
k Mass transfer coefficient .0002 m/s
- o
T′ Temperature of Solid/liquid interface K
o
Tm Temperature of liquid metal 1573 K
i Initial temperature of the scrap at the start of 303
Tsc experiment
o
K
o
Tsc Center line temperature of scrap - K
Wm Mass of melt 38880 kg
Wsc Mass of scrap .2*38880 kg
ρ Density of scrap 7200 kg/m3
λ Thermal conductivity of scrap - W/moK
∆HCO Enthalpy change in decarburization (-33000.0*4.2)/12.0 kJ/kg of

AISTech 2004 Proceedings - Volume II 925

 carbon
Enthalpy change in scrap melting and raising the 66.0*4.2
∆HFe kJ/kg Fe
temperature of liquid metal to interface temperature
∆h Enthalpy change in melting of scrap - kJ/kg Fe
α Thermal diffusivity of scrap .0000062
L Half of the total thickness of scrap M
σ1s Decarburization rate .13601*12.0 kgs of C/s
Distribution co-efficient of Carbon in Liquid and .2
-
Solid
Cs Carbon concentration of solidified scrap - wt. % C
t Time step 0.01 s

APPENDIX I

Solution of Heat equation:

∂ 2Tsc ( x, t ) ∂Tsc ( x, t )
The Fourier Equation states α = (A1)
∂x 2 ∂t
Taking θ = Tsc − T ′ (A2)
∂ θ ( x, t ) ∂θ ( x, t )
2
We solve equation α = by separation of variables method
∂x 2 ∂t
Assuming solution to be of the form (x,t)=G(x).H(t) (A3)
'
'
Putting (A3) in (A2) we get => G ( x ) + λ G ( x ) = 0 and H (t ) + αλ H (t ) = 0
2 2
(A4a & A4b)
The standard solution of (A4a) is of the form:
G(x) = Acos x + Bsin x (A5)
Now applying Boundary Conditions θ (0, t ) = 0 and θ ( 2 L, t ) = 0 to (A5) (A6a & A6b)
nπ
we get A=0 and λ= (A7)
2L
nπ x
=> Gn ( x) = Bn sin (A8)
2L
Similarly the standard solution of (A4b) is of the form
H n (t ) = Cn exp(−λ2αt ) (A9)
where is calculated from (A7). Combining both the solutions (A8) and (A9) we get
nπ x
θ n ( x, t ) = Gn ( x).H n (t ) = An sin exp(−λ2αt ) (A10)
2L
nπ x
2L
1
where An = θ ( x,0). sin .dx
L 0 2L
which gives the expression for the temperature distribution in a series form as given below:
∞ ∞
nπ x
θ ( x, t ) = θ n ( x, t ) = An sin exp(−λ2αt ) (A11)
n =1 n =1 2L
or
∞
nπx
Thus Tsc ( x, t ) = T ′ + An sin exp(−λ2αt ) (A12)
n =1 2L

This is the analytical solution of the Heat Equation. This solution would be later seen as a parabolic profile in the
scrap as expected.

926 AISTech 2004 Proceedings - Volume II