ISIJ International, Vol. 50 (2010), No. 7, pp.

10231031

Dripping Liquid Metal Flow in the Lower Part of a Blast Furnace

Hongjong JIN,1) Sangmin CHOI,1) Jun-ichiro YAGI2) and Jinkyung CHUNG3)
1) Department of Mechanical Engineering, KAIST, 2) Tohoku University, Sendai, Miyagi 980-8577 Japan. Gwangyang, 545-711 Korea. Daejeon, 305-701 Korea. E-mail: smchoi@kaist.ac.kr 3) Technical Research Laboratories, POSCO, Pohang,

(Received on October 30, 2009; accepted on February 2, 2010 )

Numerical simulation of blast furnace phenomena has signicantly contributed to the better understanding of iron making process. Recent interest on minimizing fuel consumption and reducing environmental problems have also benetted from the development of comprehensive simulation models based on physical principles. One of the under-developing elds, however, is related with the internal phenomena in the lower part of the blast furnace under the cohesive zone, where the liquid phase of metal and slag ows downward over the bed of solid coke particles. Hot ow of sluggish liquid phase is further complicated by the chemical reactions including the transfer of silica into the silicon in the hot metal. Silica enters the furnace as a constituent of coke ash and ferrous gangue, and exits as either molten silica in slag or dissolved Si in the hot metal. Silica reduction is an endothermic reaction, which would alter the heat transfer in the lower furnace, thus affecting the hot metal temperature. Effective ow in the dripping zone is important for stable operation of the blast furnace with high productivity of iron. Study of the liquid ow behavior and secondary reactions in a packed bed allows to investigate the effect of various operational changes in the dripping zone. In this research, a systematic numerical approach for the liquid ow is presented where the ow behavior is solved along with heat transfer associated with physic-chemical reactions among representative components. KEY WORDS: blast furnace; liquid ow; cohesive zone; coke silica reduction; numerical simulation.

1.

Introduction

A blast furnace process is modeled as a continuously operating reactor, where a stream of solid particles (coke and iron ore) and a stream of blast gas interact through heat transfer, uid ow and chemical reactions. The furnace can be classied as a counter-current moving bed reactor where the bed of the solid particles of iron ore and coke moves downward while the blast gas ows upward. At the top of a blast furnace, layers of coke and iron ore are charged alternately, as illustrated in Fig. 1. The iron ore arrives at the cohesive zone after being heated and partially reduced, while the hot blast gas which continuously interacts with neighboring solids moves upward from the tuyere to the top of the furnace. Temperature of the cohesive zone is usually over 1 300 K and solid iron ore starts to melt, thereby forming the hot metal and slag, which drip down through a coke-packed bed. In the lower part of a blast furnace, coke particles are piled up in the center of the furnace while being combusted with limited oxidizer available in the blast. Several reactions such as melting of ash, carbon dissolution, coke silica reduction and silicon transfer reaction occur while the molten iron ows downward. Coke silica reduction and transfer of silicon from silica to the hot metal are important secondary processes occurring in the lower furnace. Silica enters the furnace as a constituent of coke ash and ferrous gangue, and exits as either molten silica in slag or dissolved
1023

Si in the hot metal. Silica reduction is an endothermic reaction, which may alter the heat transfer in the lower furnace, thus affecting the hot metal temperature. Therefore, the predicted temperature in the lower part is higher than measured one in an actual furnace, if the reactions by silica are ignored. This affects distribution of chemical species and ow rate of molten species in the lower furnace. Many studies for the chemical reactions and its associated heat transfer have been conducted for modeling the in-

Fig. 1. Schematic diagram of species in the lower part.

2010 ISIJ

ISIJ International, Vol. 50 (2010), No. 7

Fig. 2. Diagram of the computational domain and the modeling concept for the lower part in a blast furnace.

ternal state of a blast furnace including the lower part. Reactions such as coke silica reduction and silicon transfer reaction are studied experimentally and numerically. Transformation reaction of SiO2 in coke are claried kinetically1) and mathematical models of silicon transfer through gaseous SiO in the lower part were developed.2,3) In addition, multi-dimensional transient mathematical model based on multi-uid and kinetic theories considers not only the iron ore reduction but also silica reduction and silicon transfer reaction in the lower part.4) However, many previous researchers do not consider in their model for the lower furnace the variation of cohesive zone shape which is controlled by the operating conditions, even though they have been analyzed and many parameters have been identied. The objective of this study is to simulate dripping liquid ow through a packed bed of particles where gas ows in the counter-current ow direction while the gas, solid and liquid phases are interacting through a set of chemical reactions. Selected cases cover the operating conditions of a blast furnace focusing on the lower part of a blast furnace. Computational domain covers the region below the cohesive zone which was separately simulated for the selected cases. This study is an extension from the preceded study reported by the authors which simulated the isothermal ow in the lower part.5) Investigation of non-isothermal ow provided signicantly meaningful information, and established a fundamental base for the simultaneous simulation of ow, heat transfer and reactions. 2. Mathematical Modeling

Table 1. Phase and species considered in the model.

2.1. Governing Equations Modeling system for the lower part of a blast furnace is shown in Fig. 2. The calculation domain is limited to the lower part below the cohesive zone, which consists of two parts: the deadman and the dripping zone. The deadman has a conical shape of a quasi-stationary packed bed of coke. The dripping zone is between the lower boundary of the cohesive zone and the deadman, where the gas and the liquid ow counter-currently in the slowly moving bed of coke particles. The raceway is assumed to be an annular ring cavity, in the present axi-symmetric 2-dimensional simulation. The materials in the computational domain consist of three phases; packed solid particles of coke, the gas phase and the liquid phase of hot metal and slag. In the liquid phase, only the hot metal is considered for simplication. 2010 ISIJ
1024

Three phases of liquid, solid and gas are described mathematically with governing equations for continuum ow of the mass and momentum. The unsteady-state terms in the governing equations are neglected for analyzing the ow phenomena in the steady state operation. Two-dimensional axial symmetry is assumed for computational efciency. Mathematical modeling is composed of constructing system equations in the conservation form based on the assumption that the gas, liquid and solid are continuum. Momentum equations for liquid and gas ow consist of convection, diffusion, pressure gradient and source terms for interactions with the other phases. The interactions among three phases are correlated from the experimental and numerical analysis. Non-isothermal liquid ow is based on the preceded study reported by the authors.5) The governing equations for each phase are described below. Mass conservation of the gas phase: r Gg g .................................(1) Momentum conservation of the gas phase: r r r r ( g g ug ug ) ( g g ug ) g Pg Fgs r Fgl ...(2)

The momentum equation of the gas phase consists of convection, diffusion, pressure gradient and two source terms for interactions with the solid and the liquid phase. Table 1 summarizes the correlations for the interactions. The effect of the solid phase on the gas ow is determined by using the Erguns relation.6) Effect of the liquid on the gas phase is calculated from the contact area, drag coefcient, the diameter of liquid droplet2) and the velocity difference between the gas and the liquid.7,8) Mass conservation of the liquid phase: r Gl l ..................................(3)

ISIJ International, Vol. 50 (2010), No. 7

Momentum conservation of the liquid phase: r r rr r ( l l ul ul ) l Pl Fl s Fl g l l g .........(4) The equations for the liquid phase are similar to those for the gas phase except for the gravity force in the momentum equation. Momentum conservation of the solid phase is described on the basis of potential ow assumption for the solid phase ow. The gradient of stream function is proportional to the mass velocity. Detailed correlations for interaction and contact area are described elsewhere.5) Mass conservation of the solid phase: r Gs s .................................(5) Momentum conservation of the solid phase is given as a potential function: r Gs ...................................(6) The cohesive zone is determined in the computation by the specied solid temperature region. For this purpose, solid temperature distribution is transferred from the iron ore reduction model developed by Yang,9) which considers gaseous reduction of iron ores and the other chemical reactions in the blast furnace together with uid ow, heat and mass transfer. It is assumed that iron ore starts to melt at the upper boundary of the cohesive zone and the melting completes at the lower boundary of the cohesive zone. No solid iron ore exists below the cohesive zone. Below the cohesive zone, uniform size is assumed for the coke particles in the whole packed bed, which is treated as a porous media. The void fraction of the coke bed is estimated from the sampling data of coke particles in the lower part of a blast furnace. The shape of the deadman located in the center of the lower part is expressed by a simple quadratic polynomial equation. Liquid hold-up is dened as the volume fraction of the liquid phase which is held on to the surface of the solid. The total hold-up of the liquid phase ht consists of static hold-up e l(s) accumulating on the surface of solid particles and the dynamic hold-up e l owing downwards.10) The summation of the volume fractions of gas, solid and liquid becomes unity as shown in Eq. (7).

r (GlC plTl

k l Tl )
N react

hls Als (Tl

Ts )

hgl Agl (Tg

Tl )

R (
n n 1

H n )

.........................................(10) Energy conservation of solid phase: r (GsC psTs k s Ts )

N react

hgs Ags (Tg

Ts )

hls Als (Tl

Ts )

R (
n n 1

H n )

.........................................(11) Mass fraction of components of each phase is considered for the species conservation equation. Convection as well as generation/extinction of each component is reected, while the effect of diffusion is neglected. Mass fraction of inert species in each phase is obtained by subtracting those of the other species from unity because total mass fraction is unity. Species conservation in the gas, liquid and solid phases is described respectively as Eq. (12). N react r Si , j ,n .....................(12) (Gi i , j )

n 1

g
ht

s l

ht

1 ..............................(7)

l ( s ) ...............................(8)

Detailed relations for the static and dynamic liquid holdup are described elsewhere.5) Conduction, convection, inter-phase heat transfer and reaction heat are considered for the energy conservation. Source term of each conservation equation is closely combined through the interactions of heat and mass transfer between phases. Energy conservation of gas phase: r (GgC pgTg k g Tg )
N react

2.2. Reactions Many species are involved in chemical reactions of the lower part, but only 10 species are considered in this modeling as seen in Table 1. Coke as packed particles ows towards the raceway in the dripping zone while coke packed bed is kept qausi-stationary in the deadman. There are two kinds of SiO2 participating in the reduction process of the lower part. The silica is fed from the top of the furnace in the form of ash in coke and gangue of iron ore. Ratio of silica in coke is about 45%. Silica in coke is studied for simulating reductions in the lower part, while iron ore as well as gangue is assumed to complete the melting process in the cohesive zone. The gangue in iron ore is known to melt before 1 480 K.10) Other species in ash or gangue are assumed to be inert for simplication except for silica in ash. Fe exists as liquid metal, because melting of Fe is assumed to complete in the cohesive zone. SiO2 in liquid phase originates in the gangue of iron ore. Si is absorbed in liquid metal through silica transfer reaction while C is absorbed in liquid metal by carbon dissolution reaction. SiO is an intermediate product which is generated from coke silica reduction. Mass fraction of CO and SiO increases due to the reduction in the dripping zone. Silicon transfer to the molten metal is one of the key phenomena in the lower part of blast furnace, and ve reactions are considered as listed in Table 2. Gasmetal reactions via
Table 2. Reactions considered in the model.

hgs Ags (Tg

Ts )

hgl Agl (Tg

Tl )

R (
n n 1

H n )

...........................................(9) Energy conservation of liquid phase:

1025

2010 ISIJ

ISIJ International, Vol. 50 (2010), No. 7

Table 3. Generalized equation form used in the solution process. Table 4. Operating condition of Kwangyang No. 1 blast furnace.

SiO is the key path to the silicon transfer, and reaction rate equations and associated rate constants have been proposed.11) The current simulation utilizes rate equations and constants, and reaction rate constants as summarized in Tables A and B of Appendix. SiO is mainly generated from coke ash in high temperature region, and contribution of molten slag is considered to be minor. Modern operation of pulverized coal injection (PCI) has identied additional path of SiO generation from ash from the pulverized coal,12) which is not, however, considered in the current simulation and is left for future work. 3. Simulation Methods

3.1. Solver The generalized equations for conservation of mass, energy and chemical species are rearranged and listed up in Table 3. The mathematical model was then solved by using COMSOL Multiphysics 3.4 which can solve partial differential equations. The system equations are arranged so that a general partial differential equations solver can handle each of the equations. The computational domain was divided into 9 362 triangular cells. After solving the continuity and momentum equations of the solid phase, computation was continued to solve the momentum equations for the gas and the liquid ow and then, the computation of conservation equations on energy and chemical species were simultaneously repeated until convergence. For convergence criteria, relative precision of 1 10 3 and damping factor of 1 10 6 were used. 3.2. Computational Domain, Inlet and Boundary Conditions The boundary types and input values for the computational domain of momentum balance are illustrated in Fig. 3(a). The inlet conditions of the three phases were based on the practical operation data of the blast furnace No. 1 operated by POSCO in Kwangyang. The inlet velocity of the solid particles in the cohesive zone was calculated from the hot metal production and coke rate in the operating conditions of the furnace. Table 4 lists the operational conditions of the furnace. The inlet velocity of the gas phase is calculated considering the circumferential area of tuyere due to the axial symmetry condition. The lower boundary of the computational domain is considered to be the bottom of the blast furnace, which is assumed to be the top of the liquid pool as seen in Fig. 2. 2010 ISIJ
1026

Boundary condition of simulation for temperature distribution is listed in Fig. 3(b). Left side of the geometry is xed as axial symmetry. Inlet condition of solid and liquid in the cohesive zone is set by temperature 1 623 K which is from the denition of the cohesive zone. Liquid phase goes out from the bottom line and the solid particles combust completely at the upper boundary of the raceway. Inux of gas is from the outer boundary of the raceway by 2 300 K which is calculated as adiabatic ame temperature. No heat ux is allowed through the bottom for gas and solid phase. For chemical species, initial mass fraction is set at the boundary of the geometry as seen in Fig. 3(c). Mass fraction of liquid metal is 74% of liquid phase in the cohesive zone, SiO2 in gangue is 9% and other components of gangue are 17% in the liquid phase. Carbon of coke is 88% and solid silica is 4% of the solid phase in cohesive zone. Other components of ash are assumed to be inert. The liquid phase silica starts from zero and increases in the dripping zone by melting reaction of the solid silica. Mass fraction of gaseous components is obtained from the complete combustion of coke. Iterative calculation of gaseous CO and SiO starts to react as initial value 0.411 and zero in the raceway while hydrogen and nitrogen are assumed as inert gas. Gaseous CO increases a little by silica reaction and SiO is generated in the dripping zone as an intermediate product. 3.3. Physical Properties The physical properties of air, molten iron and coke particles are listed in Table 5. The properties of gas and coke particles were obtained from a handbook.13) The surface tension of molten iron and contact angle between coke and molten iron were taken from the literature.6) Heat capacity, inter-phase enthalpy and conductivity are dependent on temperature while other properties are assumed to be independent of temperature distribution. Heat capacity of a species is a function of temperature and additivity rule is applied to calculate the property of a mixture. Heat capacity of species used in this study is tabulated in Table C of Appendix. Three enthalpy transfer coefcients are tabulated in Table D of Appendix, which have been used in heat balance equations as convective heat transfer coefcient between gas and solid. The modied RanzMarshall equation is used, which depends on conductivity of gas and solid particle diameter.12,13) Reynolds and Prandtl number inuence heat transfer as seen in the relation. Resistance in both the solid and the liquid phases is considered for the liquidsolid convective heat transfer coefcient. Resistance in liquid phase is a function of particle diameter, liquid conductivity, Reynolds number and Prandtl number of liquid. Resistance in the solid phase depends on the solid conduc-

ISIJ International, Vol. 50 (2010), No. 7

Fig. 3. Boundary conditions for (a) momentum, (b) energy and (c) chemical species.

Table 5. Physical properties of the three phases.

effective conductivity is a function of porosity which depends on particle diameter and solid temperature. 4. Results and Discussion

tivity, heat capacity of the solid, density and velocity difference between the liquid and the solid. Convective heat transfer coefcient between the gas and the liquid is a function of Reynolds number and velocity of the liquid phase.14) Thermal conductivity of each phase is obtained from the relations listed in Table E of Appendix. Conductivity of of gas species is calculated using Wilkes method15) which is a polynomial equation. The constant of each term is found in the standard engineering handbooks.1619) Thermal conductivity of the gas phase is the summation of each conductivity multiplied by corresponding mass fraction. In the case of liquid it is a linear function of temperature while constant is assumed for slag. Thermal conductivity of the solid phase is calculated using the effective conductivity.20) The
1027

Chemical species and temperature distribution of gas, liquid and solid are calculated for the reference case of the isothermal ow which was reported previously.5) Radial proles and mass ow rates through the deadman and the dripping zone are used for evaluating the results. Radial prole at the height of 7.5 m is selected to see the change of the liquid ow. This location is chosen simply to represent one cross-section of the furnace. The results can be explained by comparing to the results of the isothermal ow because the ow eld does affect the chemical reactions and its associated heat transfer. Mass ow rate of impurities such as carbon and silicon can be calculated, and the total mass of each species at the bottom is discussed. Preliminary results are calculated with the reference operating condition, where the proportion of coke charged in the central column is 80%. Particle diameter is 0.0477 m and 0.03 m in the dripping zone and in the deadman, respectively. Viscosity is 0.339 Pa s which is corresponding to the correlated values at 1 873 K. The temperature and chemical species are visualized by contour diagram. Chemical species are expressed by mass fraction in each phase.

2010 ISIJ

ISIJ International, Vol. 50 (2010), No. 7

Fig. 4. Temperature distribution of solid and liquid; (a) gas, (b) solid and (c) liquid [K].

Fig. 5. Temperature prole at the height of 7.5 m from the bottom.

Temperature Distribution and Chemical Composition Temperature distribution of the solid phase shows similar pattern to that of the gas phase over the computational domain under the cohesive zone as seen in Fig. 4. Temperature proles of gas, solid and liquid at the height of 7.5 m from the bottom are shown in Fig. 5. Temperature at the central axis region is higher (i.e., T increases as r/R approaches 0), while gas temperature is higher than that of solid phase in the whole domain. Temperature of the liquid phase increases more sharply around the central axis region than in the outer wall region. Gas ow rate is higher in the central region because more coke is charged in the central axis region to ensure high permeability of gas. Temperature of the solid phase increases from the cohesive zone and reaches its maximum around the axis in the deadman as moving down. On the other hand, temperature of the gas phase decreases from the upper boundary of the raceway as going up. Thermal energy of the gas phase is transferred to the solid and the liquid phases. Isothermal contour lines of the solid, liquid and gas phases appear to follow the shape of the cohesive zone in the dripping zone. Contour of liquid temperature in the deadman shows that the temperature is dependent on the liquid ow. Mass fraction of gaseous species CO and SiO are shown in Fig. 6 using iso-concentration contour lines. CO is formed through coke silica reduction and SiO is increased as the balance of coke silica reduction and Si forming reaction. However, reaction rate of the silicon forming reaction 2010 ISIJ
1028

4.1.

Fig. 6. Mass fraction of gas species; (a) CO and (b) SiO.

is less than that of the coke silica reduction. Concentrations of both CO and SiO increase in the dripping zone, due to the coke silica reduction. There is no change of mass fraction in the gas phase, because coke silica reduction does not occur in the deadman. Mass fraction of SiO is zero in the raceway, because it was assumed that no mass fraction remains in the combustion gas of raceway. It increases up to 3% in the gas phase at around the central axis of the dripping zone. Mass fraction of CO was 41.1% in the raceway from the composition of the gas phase. The species CO increases also by 2% while owing upward. Mass fraction of the liquid phase is illustrated in Fig. 7 as a contour diagram. Liquid species such as carbon, silicon and liquid silica undergo reactions of carbon dissolution, melting and silicon transfer reaction. The mass fractions of carbon and silicon increase while that of Fe decreases. Fraction of liquid silica is 9% at the cohesive zone and decreases as owing downward because all species in iron ore are assumed as liquid phase in the cohesive zone. The decrease of liquid silica is about 0.3% in the deadman. Carbon absorbed in metal increases up to 4% at around the central axis in the deadman, while silica increases to about 1.5%.

ISIJ International, Vol. 50 (2010), No. 7

Fig. 7. Mass fraction of solid species; (a) C and (b) Si in liquid metal.

4.2.

Effect of Secondary Reactions in the Lower Part on the Temperature Distribution Coke silica reduction is an endothermic reaction which occurs in the lower part of a blast furnace. Temperature distribution of liquid and solid phases is illustrated in Fig. 8, which considers the endothermic reactions. Temperature decrease of the gas and the solid phases is about 20 K after those reactions. The temperature change of the liquid phase due to the endothermic reactions is small in the peripheral region as seen in Fig. 8. The coke silica reduction occurs in the dripping zone as seen in Fig. 6 and it occurs substantially in the central region. Mass fraction increase of gaseous species is high around the central axis while carbon absorbed in liquid metal increases near the central axis. Therefore, the temperature decrease of liquid phase is higher in the central region than in the peripheral region. Figure 7 shows mass fractions of solid phase species C and Si in the liquid metal. The result can be discussed with operating conditions and its effect on the liquid metal composition. Silicon in the liquid metal comes through reduction of gaseous SiO with carbon dissolved in the liquid metal. Therefore, silicon concentration in the liquid metal depends on the contact time between SiO and liquid metal. Lowering the cohesive zone position would decrease the contact time between SiO and liquid metal, and would decrease the concentration of silicon in the liquid metal. Additionally, carbon in the liquid metal decreases due to the consumption for reduction of SiO from the silica in coke ash. Figure 8 shows the difference of the predicted temperature, with and without reactions. This temperature difference shows that the predicted temperature distribution by iron ore reduction model can be higher than the measured value in the operating furnace. The iron ore reduction model of the author group does not consider liquid ow and reaction in the lower part of a blast furnace. The cohesive zone would also be affected by these reactions. Carbon dissolution and silicon transfer reaction are directly related with the position of the cohesive zone in the blast furnace.

Fig. 8. Temperature distribution with and without considering reactions in the lower part; (a) solid and (b) liquid temperature difference.

5.

Conclusion

Liquid metal ow in the lower part of a blast furnace is simulated by describing the dripping liquid over a packed bed of particles where gas ows in the counter-current direction. Axi-symmetric 2-dimensional computational domain covers the region under the cohesive zone, which was computed separately and used as input condition. The simulations have been performed for the case of Kwangyang No. 1 blast furnace. Liquid ow is represented by the conservation equations which accounts for the dynamic and static hold-up. Numerical modeling of the heat transfer and chemical reactions is incorporated into the previously reported iso-thermal ow simulation model. Melting of ash, carbon dissolution, coke silica reduction and slag silica reduction are considered in the dripping zone and in the deadman. Silica enters the furnace as a constituent of coke ash and ferrous gangue, and exits as either molten silica in slag or dissolved Si in the hot metal. Flow (momentum), energy (thermal energy) and material conservation equations were numerically solved simultaneously. The effect of reduction reactions in the lower part on the temperature decrease was evaluated. Additional information on the internal state of liquid, gas and solid phases in the lower part would enhance the understanding of the operating conditions and its related parameters. Acknowledgements The authors wish to express their appreciation to POSCO and BK21 (Brain Korea 21) for the support to this research. Nomenclature G: h:
1029

Mass ow rate (kg m 2 s 1) Total liquid hold up, convective heat 2010 ISIJ

ISIJ International, Vol. 50 (2010), No. 7

P: C: a: F: u: d: g: S: Re : Ga : Cp : We : Fr : T: V: E: k: n: R: DH : Greek

transfer coefcient (kW m 2 K 1) Pressure (Pa) Drag coefcient Contact area (m2) Interaction parameter (N m 3) Velocity (m s 1) Diameter (m) Acceleration of gravity (m s 2) General source term Reynolds number Gallilei number Capillary number Weber number Froude number Temperature (K) Volume ow rate (m3 s 1) Blast energy (kg m s 1) Conductivity (W m 1 K 1) Index for reactions Reaction rate (kmol m 3 s 1) Heat of reaction (kJ kmol 1) Volume fraction of gas or solid, dynamic liquid hold up Static liquid hold up Viscosity (Pa s) Density (kg m 3) Contact angle of liquid on the solid surface (rad) Stream function Surface tension (N m 1) Shape factor Mass fraction Fraction of heat absorbed by gas Fraction of heat absorbed by liquid Fraction of heat absorbed by solid Potential function Total Gas Liquid Solid Packed particle Gas Liquid Solid
REFERENCES

8) T. Akiyama, R. Takahashi and J. Yagi: Tetsu-to-Hagan, 76 (1990), 848. 9) K. Yang, S. Choi, J. Chung and J. Yagi: ISIJ Int., 50 (2010), 972. 10) T. Fukutake and V Rajakumar: Tetsu-to-Hagan, 66 (1980), 1937. . 11) P. R. Austin, H. Nogami and J. Yagi: ISIJ Int., 37 (1997), 748. 12) Y. Matsui, S. Mori and F. Noma: ISIJ Int., 43 (2003), No. 7, 997. 13) Robert H. Perry and Don W. Green: Perrys Chemical Engineers Handbook, 7th Ed., McGraw-Hill. New York, (1997), 20-8. 14) P. J. Mackey and N. A. Warner: Metall. Trans., 3 (1972), 1807. 15) C. R. Wilke: J. Chem. Phys., 18 (1950), 517. 16) D. R. Stull and H. Prophet: JANAF Thermo-chemical Tables, 2nd Ed. (1971). 17) John A. Dean: Langes Handbook of Chemistry, 15th Ed., McGrawHill, New York, (1999), 155. 18) David R. Lide: Handbook of Chemistry and Physics, 87th Ed., CRC Press, Florida, (1999), 265. 19) Robert C. Reid, John M. Prausnitz and Bruce E. Poling: The Properties of Gases and Liquids, 4th Ed., McGraw-Hill, New York, (1987), 407. 20) W. Shotte: AIChE J., 6 (1960), 63. 21) M. Hatano and K. Kurita: Tetsu-to-Hagan, 66 (1980), 1898. 22) E. R. G. Eckert and R. M. Drake: Heat and Mass Transfer, 2nd Ed., McGraw Hill, New York, (1959), 173. 23) C. R. Wilke: J. Chem. Phys., 18 (1950), 517. 24) W. Shotte: AIChE J., 6 (1960), 63.

e: e (s) : m: r: q: y: d: f: w: a: b: g: z: Subscripts t: g: l: s: p: Superscripts g: l: s:

Appendix A
Table A. Rate equations and constants.

Table B. Reaction constants.

1) C. Yamagata, Y. Kajiwara and S. Suyama: Tetsu-to-Hagan, 73 (1987), 637. 2) S. Taguchi, H. Kubo, N. Tsuchiya, K. Ichifuji and K. Okabe: Tetsuto-Hagan, 68 (1982), 2303. 3) B. Ozturk and R. J. Fruehan: Process Tech. Proc. 5th NISC AIME, 6 (1986), 959. 4) H. Nogami, M. Chu and J. Yagi: Comput. Chem. Eng., 29 (1005), 2438. 5) H. Jin and S. Choi: Ironmaking Steelmaking, 37 (2010), 89. 6) H. Kawabata, Z. Liu, F. Fujita and T. Usui: ISIJ Int., 45 (2005), 1466. 7) W. E. Ranz and W. R. Marshall: J. Chem. Eng. Prog., 48 (1952), 173.

2010 ISIJ

1030

ISIJ International, Vol. 50 (2010), No. 7

Table C. Heat capacity and heat of formation. Table E. Thermal conductivity of each phase.

Table D. Inter-phase enthalpy transfer.

1031

2010 ISIJ