See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/233391599

Modeling and Validation of an Electric Arc Furnace: Part 1, Heat and Mass
Transfer

Article  in  ISIJ International · March 2012

DOI: 10.2355/isijinternational.52.402

CITATIONS READS

53 1,585

3 authors, including:

Vito Logar Dejan Dovzan

University of Ljubljana University of Ljubljana
53 PUBLICATIONS   573 CITATIONS    34 PUBLICATIONS   702 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Research of evolving systems View project

Intellignet autonomous mobile platform for rehabilitation View project

All content following this page was uploaded by Vito Logar on 21 May 2014.

The user has requested enhancement of the downloaded file.

 ISIJ International, Vol. 52 (2012), No. 3, pp. 402–412

Modeling and Validation of an Electric Arc Furnace: Part 1, Heat

and Mass Transfer

Vito LOGAR, Dejan DOVŽAN and Igor ŠKRJANC

Laboratory of modeling, simulation and control, Faculty of Electrical Engineering, University of Ljubljana, Tržaška 25, SI-1000
Ljubljana. E-mail: vito.logar@fe.uni-lj.si, dejan.dovzan@fe.uni-lj.si, igor.skrjanc@fe.uni-lj.si
(Received on September 2, 2011; accepted on October 11, 2011)

The following paper presents an approach to the mathematical modeling of heat and mass transfer pro-
cesses in a 3–phase, 80 MVA AC, electric arc furnace (EAF) and represents a continuation of our work on
modeling the electric and hydraulic EAF processes. This paper represents part 1 of the complete model
and addresses issues on modeling the mass, temperature and energy processes in the EAF, while part 2
of the paper focuses solely on the issues related to the thermo-chemical relations and reactions in the
EAF. As is generally known, the chemical, thermal and mass processes in an EAF are related to each
other and cannot be studied completely separately; therefore, the work presented in part 1 and part 2 is
related to each other accordingly and should be considered as a whole. The presented sub-models were
obtained in accordance with different mathematical and thermo-dynamic laws, with the parameters fitted
both experimentally, using the measured operational data of an EAF during different periods of the melt-
ing process, and theoretically, using the conclusions of different studies involved in EAF modeling. In con-
junction with the already presented electrical and hydraulic models of the EAF, the heat-, mass- and
energy-transfer models proposed in this work represent a complete EAF model, which can be further
used for the initial aims of our study, i.e., optimization of the energy consumption and development of the
operator-training simulator. The presented results show high levels of similarity with both the measured
operational data and the theoretical data available in different EAF studies, from which we can conclude
that the presented EAF model is developed in accordance with both fundamental laws of thermodynamics
and the practical aspects regarding EAF operation.

KEY WORDS: EAF; heat-transfer model; mass-transfer model; experimental validation.

necessary point as (PART 2).

1. Introduction When examining the literature, some papers that propose
The paper proposes an approach to the mathematical different approaches to the modeling of the mass transfer
modeling of heat and mass transfer processes in an 80 MVA and/or thermal processes in the EAF already exist, from
AC, electric arc furnace (EAF) and represents the continu- simplified and more complex general heat- and mass-transfer
ation of our work on modeling the electrical and hydraulic models2,3) to a more focused analysis of different heat sub-
EAF processes.1) The relations and sub-models studied in models.4,5) The idea of the model presented in this paper was
this paper are considered as part 1 of the complete EAF derived from the papers of Bekker2) and MacRosty;3) how-
model, with the thermo-chemical relations in the EAF being ever, the model proposed by Bekker is oversimplified for the
addressed in part 2. As is generally known, the chemical, needs of our study, while the model proposed by MacRosty
thermal and mass processes in an EAF are related to each either describes differently or does not address some issues
other and cannot be considered separately; therefore, the on the radiative and conductive heat transfer between the
work presented in part 1 and part 2 is related to each other steel, the slag and the gas zones and the CO post-
accordingly and should be considered as a whole. The main combustion, which are in our opinion also important for the
objective of the models proposed in this paper is to obtain overall accuracy of the model. Therefore, a more complex
accurate and reliable relations that describe different types model, taking into account additional relations and eliminat-
of heat and mass transfers between the EAF zones, i.e., the ing the deficiencies of the previously mentioned models is
solid and liquid steel and slag zones, the gas zone, the EAF’s proposed in this study. Besides the modeling of the funda-
roof and walls, etc. The thermal processes considered in this mental laws of mass and heat transfer, special attention is
paper include the heat transfer dissipated from the arcs, the devoted to the parameterization of the developed model
radiative heat exchange, the view factors and temperature using available initial, endpoint and online measurements of
changes, while part 2 represents the chemical reactions and the EAF’s operation. As some relations in the EAF process
the corresponding thermal relations. As for the mass-transfer are hard to be modeled mathematically, approximate equa-
model, this paper presents mass-transfer models as a conse- tions are proposed to fit the obtained measurements.
quence of steel and slag melting, while additional mass Like the electrical model developed in our previous work,
transfers that occur as a consequence of chemical reactions the model proposed here is based on the 80 MVA AC fur-
are described in part 2. A reference to part 2 is added at each nace installed in one of the ironworks in Slovenia. There-

© 2012 ISIJ 402

 ISIJ International, Vol. 52 (2012), No. 3

fore, for the needs of a successful model parameterization, as: Fe, Si, Mn, Cr, C and P in solid form,
the operational data of the EAF related to the heat- and • liquid scrap zone (lSc), containing the same elements
mass-transfer processes was obtained, including the initial as the solid scrap zone, but in liquid form,
and endpoint steel and slag masses, the temperatures, the • solid slag zone (sSl), containing charged slag-forming
power-on times, the steel yields, the temperatures of the compounds such as: CaO, MgO and Al2O3 in solid form,
cooling panels, the energy consumption per baskets and heats, • liquid slag zone (lSl), containing the slag-forming and
etc. Other measurements, such as the arc powers, which are oxidation-process compounds in liquid form such as: CaO,
needed to complement the proposed models, were already MgO, FeO, SiO2, MnO, Cr2O3, P2O5 and Al2O3 in liquid
obtained when modeling the electrical processes in the EAF. form,
• gas zone (gas), containing the gases either lanced into
the EAF or produced as a product of oxidation/reduction,
2. Modeling such as: N2, O2, CO, CO2 and CH4.
The following section presents the approach to the mod- Two more EAF zones were added for the needs of the
eling of heat- and mass-transfer processes for the particular conductive and radiative heat-transfer model:
EAF. The values of all the parameters used in the developed • roof zone,
model are listed in the Appendix section. • wall zone,
both primarily defined for computing the cooling powers
2.1. Modeling Assumptions and Simplifications and the EAF cooling losses. For the needs of the heat- and
As is generally known in modeling practice, the devel- mass-transfer models, each zone is assigned common
oped model should meet the initial design goals; however, parameters describing its physical properties, i.e., specific
for practical applications of the models several assumptions heat Cp, density ρ , thermal conduction k, molar mass M, etc.,
and simplifications are often needed in order to either which are the average values proportional to the physical
achieve those goals, or to meet other requirements related to properties of its constituents. The zones are schematically
the model in general (stability, robustness, etc.). This is presented in Fig. 1.
especially true for the electric arc furnace modeling, which
combines different hydraulic, electrical, thermal, chemical 2.2. Heat Transfer Model
and other processes. Therefore, to develop an applicable, When operating the EAF all three types of heat transfers,
robust and fairly accurate model, certain assumptions and i.e., conductive, convective and radiative, are present. Heat
simplifications need to be made, as some of the modeled conduction occurs between the zones in the furnace, which
processes cannot be completely described by conventional are in a direct contact with each other. Convection of the
laws and include different forms of randomness. The heat is present between the gas zone and the surrounding
assumptions and simplifications regarding the heat- and zones, while radiation prevails between the surfaces that are
mass-transfer models proposed are: in the field of vision of each other. Considering all three
• Due to the lack of some online measurements (mass mechanisms of heat transfer, the EAF thermal model is
balance, temperatures, etc.), the developed model, describ- based on the heat-transfer or energy-balance equations,
ing the related processes, should follow the fundamental which are determined for each of the previously mentioned
(ideal) laws of heat and mass transfer, without considering EAF zones.
the possible random processes, which cannot be modeled by
conventional principles. The identifiable parameters of the Solid scrap zone (sSc)
model are determined using the initial and endpoint mea- The solid-scrap zone receives the heat from the arcs Qarc, liq-
surements of the process values, while the non-identifiable uid metal QlSc-sSc, liquid slag QsSc-lSl, gas zone QsSc-gas, oxygen
parameters are obtained from the conclusions of different burners QCH4(PART 2) and CO post-combustion QCO-post(PART 2);
practical studies, investigating heat and mass transfers. while its energy is exchanged with the solid slag QsSc-sSl and
• In order to combine the existing electrical and hydraulic lost to the water-cooled panels QsSc-water. Also, a large por-
models with the proposed model and to avoid possible numer- tion of the heat is exchanged through the radiation QsSc-RAD,
ical problems due to the stiffness of the combined model, the which is dependent on the thermal conditions, emissivity
integration method should remain the same as for the electri- coefficients, surface areas and view factors in the furnace,
cal model,1) i.e., fixed-step Euler, sample time Ts = 10–4 s. which are addressed in Section 2.2.1. The energy received
• Any occurring discontinuities that are necessary for from the oxygen burners, arcs and CO post-combustion is
the proper model response should be modeled in a manner dependent on the exposure coefficient KsSc-lSc (Eq. (32)),
that does not cause numerical and/or stability problems. which determines the fractions of the arc and burner energy
• The modeled EAF is divided into different zones, transferred either to the solid or liquid metal zones. The
where the constituents of each zone possess similar or equal
thermal, chemical and physical properties. Each zone is con-
sidered as homogenous, both in the constituent’s properties,
as in temperature levels, e.g., the temperature and composi-
tion of the steel scrap is considered equal throughout the sol-
id-scrap zone, even though the actual EAF operation is far
from this assumption. However, since the proposed model
does not focus on the mass and heat transfers inside each
zone, which would require a more complex 3-dimensional
modeling, but rather concentrates on the zones as a whole,
such an assumption seems reasonable.
Considering the above-mentioned assumptions and sim-
plifications, the EAF layout is divided into different zones,
intended for the mass- and heat-transfer models, as follows:
• solid scrap zone (sSc), containing the elements such Fig. 1. Schematic representation of the EAF zones.

403 © 2012 ISIJ

 ISIJ International, Vol. 52 (2012), No. 3

energy balance for the solid steel zone can therefore be rep-
resented by Eq. (1):
QsSc = (Qarc + QCH4 + K post QCO − post ) ⋅ (1 − K sSc −lSc ) + QlSc − sSc
−QsSc − sSl −QsSc −lSl −QsSc − gas −QsSc − water −QsSc − RAD ,
.......................................... (1)
where Kpost denotes the efficiency of CO post-combustion.
CO post-combustion energy, which is not reflected to the
steel zone, is otherwise used to heat the gas zone. The neg-
ative sign preceding QsSc-RAD denotes that the solid-scrap
zone receives radiative energy.
The energy added to the solid steel zone from the arcs is
directly dependent on the arc powers. As suggested, the arc
energy transferred to the steel through conduction amounts
to 15–20% of the total arc power, 75–85% of the arcs’ ener- Fig. 2. Comparison between the measured (suggested)8) and the
gy is transferred to the furnace surfaces by radiation6) and tangent-hyperbolic approximated burner efficiency. X-axis
some of the energy is lost to gas and electrodes (2–5%).3,7) label shows the percentage of the meltdown time as pro-
In our case the energy dissipated from the arcs by conduc- posed by Ref. 8) and the corresponding temperature of solid
tion (Qarc) is proportional to arcs’ powers and represents steel as can be obtained from the model.
20% of the total power, the energy transferred by radiation
(Qarc-RAD) is assumed to represent 75% of arc power, 2.5%
of the arc power heats the gas zone (Qarc-gas), while the peratures of the solid and liquid metal zones, respectively.
remaining 2.5% of the energy is lost to electrodes and is thus Like Eq. (4), the energy exchange between the solid
neglected in further calculations. Qarc can therefore be scrap, solid slag and liquid slag zones can be defined by
defined by Eq. (2): Eqs. (5) and (6):
Qarc = 0.20 ⋅ Parc , ............................ (2) QsSc − sSl = msSc − sSl K therm 2 K area 2 (TsSc − TsSl ), ......... (5)
where Parc represents the sum of arc powers. Radiative heat QsSc −lSl = msSc −lSl K therm 3 K area 3 (TsSc − TlSl ), ......... (6)
from the arcs is included in QsSc-RAD and is addressed in Sec-
tion 2.2.1. where msSc-sSl and msSc-lSl represent the masses of either the
The energy added to the solid steel zone from the oxygen solid steel or the solid or liquid slag, depending on which
burners QCH4 is proportional to the burners’ power and we one is smaller; Ktherm2 and Ktherm3 represent the heat-transfer
suggest that their efficiency is described by the following coefficients; Karea2 and Karea3 represent the area coefficients
Eq. (3): between the solid steel and the solid or liquid slag zones;
⎛ TsSc, TsSl and TlSl represent the temperatures of the solid steel
⎛ 1 300D C ⎞ ⎞
QCH4 = ΔHTD − n K burn ⋅ ⎜ 0.35 + 0.65 tanh ⎜ − 1 ⎟ ⎟ , ... (3) and the solid and liquid slag zones, respectively.
⎜ ⎟ The energy exchanged between the solid scrap and the sur-
⎝ ⎝ TsSc ⎠⎠
D
where ΔHT − n (PART 2)
represents the energy provided by the rounding gas QsSc-gas can be described by Eq. (7) as follows:
burners, Kburn represents the approximate burner efficiency m
QsSc − gas = sSc K therm 4 (TsSc − Tgas ) ⋅ (1 − K sSc −lSc ), ... (7)
(0.7)6) and together with (0.35 + 0.65 tanh( 1300
D

T
C
− 1)) represents mEAF
where msSc represents the current mass of solid steel; mEAF
sSc
the hyperbolic-tangent approximation which accounts for
decreasing burner efficiency with increasing temperature of represents the EAF mass capacity; Ktherm4 represents the
the steel. Hyperbolic-tangent burner efficiency approxima- thermal conductance coefficient; TsSc and Tgas represent the
tion is derived from the paper of Bergman and Gottardi,8) temperatures of the solid steel and the gas zones, respective-
who suggest that the burner efficiency is decreasing propor- ly; KsSc-lSc represents the exposure coefficient of the solid or
tionally to the % of the meltdown time. Since the current liquid metal (Eq. (32)). Coefficient Ktherm4 was obtained from
temperature of the steel is related to the % of the meltdown the approximate heat transfer coefficient for steel7,9,10) taking
time, we suggest that the burner efficiency is approximated into account the properties of the specific EAF. Together with
m
by the temperature of the solid steel. The comparison fraction m sSc this factor accounts for decreasing heat transfer
between the measured8) and the approximated burner effi- to the gas zone with decreasing mass of solid steel.
EAF

ciency can thus be represented by Fig. 2. A portion of the solid steel energy is lost due to the cooling
Modern EAF assemblies use the advanced technology of (QsSc-water) of the furnace walls, which was modeled by Eq. (8):
QsSc − water = K water1 (TsSc − Twall ) sSc ⋅ ⎛⎜ 1 − e ⎞,
carbon monoxide (CO) post-combustion. The idea of the T −
msSc

⎟ ..... (8)
mEAF

post-combustion is to oxidize the carbon monoxide, which Tmelt ⎝ ⎠

is produced by chemical reactions in the bath, to carbon where Kwater1 represents the thermal conductance coefficient,
dioxide (CO2) and use the energy of the chemical reaction which was obtained from approximate heat transfer coeffi-
with oxygen, instead of emitting the CO to off-gas. The cient of the furnace walls7) taking into account the dimen-
energy added to the solid steel zone from the CO post- sions of the specific furnace; TsSc, Twall and Tmelt represent the
combustion QCO-post is dependent on the efficiency (Kpost) of temperatures of the solid steel, the cooling panels and the
the post-combustion and is discussed in part 2(PART 2).
m
− m sSc
steel’s melting point, respectively; and (1 − e EAF
) represents
The energy flow between the solid and liquid scrap zones
QlSc-sSc can be described by Eq. (4) as follows: the exponential decrease of the cooling power with the
m
− m sSc
QlSc − sSc = mlSc − sSc K therm1 K area1 (TlSc − TsSc ), ......... (4) decreasing mass of solid steel. The expression (1 − e EAF
)
where mlSc-sSc represents the mass of either the solid or liquid might be unusual in the EAF modeling practice; however,
metal, depending on which one is smaller; Ktherm1 represents as the presented paper studies each EAF zone separately
the heat-transfer coefficient; Karea1 represents the area coef- (rather than a whole) the needed cooling powers are directly
ficient between both zones; TlSc and TsSc represent the tem- dependent on the mass of each zone. The expression there-

© 2012 ISIJ 404

 ISIJ International, Vol. 52 (2012), No. 3

fore, accounts for higher or lower cooling power needed for can be described with Eq. (12) as follows:
the each zone, when there is more or less material present. m
QlSc − gas = lSc K therm 7 (TlSc − Tgas ) ⋅ K sSc −lSc , ...... (12)
A slight non-linear mass dependency could be, without sig- mEAF
nificant loss of
m
accuracy, replaced with completely linear where mlSc represents the current mass of the solid steel;
dependence ( m ); however, exponential behavior has been
sSc

EAF
mEAF represents the EAF mass capacity; Ktherm7 represents
chosen since the melting of the scrap exhibits minor non- the thermal conductance coefficient; TlSc and Tgas represent
linear dynamics (due to cone-frustum void, Fig. 3). Similar the temperatures of the liquid metal and gas zones, respec-
expressions can also be found in Eqs. (13), (15) and (18), tively; KsSc-lSc represents the exposure coefficient. Coeffi-
where their interpretation is similar as here. cient Ktherm7 was obtained similarly to coefficient Ktherm4 in
Eq. (7) from the approximate heat transfer coefficient for
Liquid metal zone (lSc) steel7,9,10) and the properties of the specific EAF. Together
m
Considering the liquid metal zone, it receives the ener- with fraction m lSc this factor accounts for increasing heat
gy from the arcs Qarc, the oxygen burners QCH4(PART 2), CO transfer to the gas zone with increasing mass of liquid steel.
EAF

post-combustion QCO-post(PART 2) and chemical reactions Similarly to solid scrap zone, the liquid metal zone also
QlSc-chem(PART 2); while its energy is lost to the solid steel QlSc-sSc, loses energy due to the cooling (QlSc-water) of the furnace
solid slag QlSc-sSl, liquid slag QlSc-lSl, water-cooled panels walls, which can be described with Eq. (13):
QlSc-water and gas zone QlSc-gas. Like with Eq. (1), KsSc-lSc rep-
QlSc − water = K water 2 (TlSc − Twall ) lSc ⋅ ⎛⎜ 1 − e mEAF ⎞⎟ , ...... (13)
T − lSc
m

resents the exposure coefficient between the solid and liquid

Tmelt ⎝ ⎠
metal zones. The liquid metal zone also exchanges a signif-
icant amount of energy through radiation QlSc-RAD, which is where Kwater2 represents the thermal conductance coefficient;
addressed in Section 2.2.1. The energy balance for the liquid TlSc, Twall and Tmelt represent the temperatures of the liquid
metal zone can be represented by Eq. (9): metal, cooling panels and steel melting point, respectively;
QlSc = (Qarc + QCH4 + K post QCO − post ) ⋅ K sSc −lSc + QlSc − chem
m
− m lSc
and (1 − e EAF
) represents the exponential increase of the
−QlSc − sSc − QlSc − sSl − QlSc −lSl − QlSc − gas − QlSc − water − QlSc − RAD . cooling power with the increasing mass of liquid metal,
.......................................... (9) described in Eq. (8).
The arcs’ conductive (Qarc) and the burners’ (QCH4) energy
input is obtained with Eqs. (2) and (3), while the CO post- Solid slag zone (sSl)
combustion (QCO-post) and chemical energy (QlSc-chem) equa- The solid slag zone receives the energy from the liquid met-
tions are addressed in part 2(PART 2). Like with the solid zone, al QlSc-sSl, exchanges the energy with the solid steel QsSc-sSl and
the actual energy from the arc, burners and post-combustion loses its energy to the cooling panels QsSl-water. The energy-
transferred to the liquid metal is proportional to the expo- balance equation is as follows 14:
sure coefficient KsSc-lSc (Eq. (32)). The energy flow between QsSl = QsSc − sSl + QlSc − sSl − QsSl − water . ............. (14)
the solid and liquid metal zones QlSc-sSc is obtained using Eq. Since the solid slag zone is usually not in direct contact with
(4), while the energy exchange between the liquid metal, liq- the liquid slag, gas zone, burners or the arcs, the energy
uid slag and solid slag can be determined with the following exchange between those zones is ignored. The energy flows
Eqs. (10) and (11): QsSc-sSl and QlSc-sSl are obtained with Eqs. (5) and (10),
QlSc − sSl = mlSc − sSl K therm 5 K area 5 (TlSc − TsSl ), ....... (10) respectively. The energy loss due to the cooling QsSl-water of
the furnace walls can be described with Eq. (15):
QlSc −lSl = mlSc −lSl K therm 6 K area 6 (TlSc − TlSl ), ........ (11)
QsSl − water = K water 3 (TsSl − Twall ) sSl ⋅ ⎛⎜ 1 − e mEAF ⎞⎟ , ...... (15)
T − sSl
m

where mlSc-sSl and mlSc-lSl represent the masses of either the

Tmelt ⎝ ⎠
liquid metal, solid or liquid slag, depending on which one
is smaller; Ktherm5 and Ktherm6 represent the heat-transfer where Kwater3 represents the thermal conductance coefficient;
coefficients; Karea5 and Karea6 represent the area coefficients TsSl, Twall and Tmelt represent the temperatures of the solid
between liquid metal and solid slag zones; TlSc, TsSl and TlSl slag, cooling panels and steel melting point, respectively;
represent the temperatures of the liquid metal, solid and liq-
m
− m sSl
and (1 − e EAF
) represents the exponential decrease of the
uid slag zones, respectively.
Like with the solid scrap zone, the energy exchanged cooling power with the decreasing mass of solid slag,
between the liquid metal and the surrounding gas QlSc-gas, described in Eq. (8).

Liquid slag zone (lSl)

The liquid slag zone receives energy from the liquid met-
al QlSc-lSl; while its energy is lost to solid steel zone QsSc-lSl,
gas zone QlSl-gas and cooling panels QlSl-water. The energy-
balance equation can be written as follows 16:
QlSl = QlSc −lSl + QsSc −lSl − QlSl − gas − QlSl − water . ....... (16)
The heat exchange between the liquid slag and gas zones
( QlSl − gas ) can be defined with the following Eq. (17):
m
QlSl − gas = lSl K therm8 (TlSl − Tgas ) ⋅ K sSc −lSc , ...... (17)
mEAF
where mlSl represents the current mass of the liquid slag;
mEAF represents the EAF mass capacity; Ktherm8 represents
the thermal conductance coefficient; TlSl and Tgas represent
the temperatures of the liquid slag and gas zones, respec-
tively; KsSc-lSc represents the exposure coefficient. Similarly
Fig. 3. Four different phases of the scrap-melting process relevant to coefficients Ktherm4 and Ktherm7, Ktherm8 is obtained
from.7,9,10) Together with fraction m lSl the factor accounts
m
for obtaining the surface areas and view factors.
EAF

405 © 2012 ISIJ

 ISIJ International, Vol. 52 (2012), No. 3

for increasing heat transfer to the gas zone with increasing zone has no energy relation to the gas zone. The energy loss
mass of liquid slag. due to the cooling Qgas-water of the furnace roof and walls, can
The energy loss due to the cooling QlSl-water of the furnace be described by Eq. (22):
walls can be described with Eq. (18): ⎛ A1 A2 ⎞
Qgas − water = K water 5 ⎜ (Tgas − Troof ) + (Tgas − Twall ) ⎟,
QlSl − water = K water 4 (TlSl − Twall ) lSl ⋅ ⎛⎜ 1 − e mEAF ⎞⎟ , ......... (18)
T − lSl
m
A1 + A2 A1 + A2 ⎠
⎝
Tmelt ⎝ ⎠ ........................................ (22)
where Kwater4 represents the thermal conductance coefficient; where Kwater5 represents the thermal conductance coeffi-
TlSl, Twall and Tmelt represent the temperatures of the liquid cient; Tgas, Troof and Twall represent the temperatures of the
slag, cooling panels and steel melting point, respectively; and gas and the roof- and wall-cooling panels, respectively.
mlSl
− A1 A2
(1 − e mEAF ) represents the exponential increase of the cooling Fractions A1 + A2
and A1 + A2 define the portion of the gas energy
power with the increasing mass of liquid slag, described in transferred either to roof or wall zone, depending on their
Eq. (8). surface areas A1 and A2.
Like as in Eq. (14), the liquid slag is not in direct contact
with the solid slag; therefore, the energy exchange between 2.2.1. Radiative Heat Transfer
those zones is ignored. Since the electric arcs tend to dis- The radiative heat transfer represents a significant amount
place the liquid slag to reach the liquid metal zone, the ener- of the total heat transferred in the EAF. For the purpose of
gy received from the arcs is neglected. One would expect this study the furnace is considered as an enclosure with N
that the energy-balance equation for the liquid slag zone surfaces, all treated as gray bodies. The gray-body radiosity
would also include the radiative heat exchange. Even though Ji can be determined with Eq. (23):11)
this would be sound from the physical point of view, it is
N
very difficult to determine the radiative relations between J i = ε iσ SBTi 4 + (1 − ε i )∑ (VFij ⋅ J j ), ........... (23)
the liquid slag and the other surfaces, mostly in the condi- j =1
tions before the flat-bath period. For this reason we decided
not to consider the radiative heat transfer in the case of the where ε i represents the body’s emissivity [0–1], σ SB repre-
liquid slag zone. Consequently, the parameters regarding the sents the Stefan-Boltzmann constant, Ti represents the
liquid slag zone had to be adjusted accordingly. On the other body’s temperature, VFij represents the view-factor from
hand, the liquid slag zone has a direct impact on the radiative surface i to surface j [0–1] and Jj represents the radiosity of
heat transfer of the liquid metal zone, which is discussed later. the body j. The first part of Eq. (23) represents the gray-
body radiation according to the Stefan-Boltzmann law,
Gas zone (gas) while the second part of the equation represents the incident
The gas zone receives energy from the arcs Qarc-gas, the radiation, which is the sum of the radiations from all the oth-
oxygen burners QCH4-gas(PART 2) and the CO post-combustion er surfaces per unit of surface area i.
QCO-post(PART 2), depending on the post-combustion inefficien- Obtaining the radiosity for all the bodies in the furnace,
cy (1- Kpost), exchanges the energy with solid steel QsSc-gas, radiative heat transfer Qi-RAD for the surface i can be com-
liquid metal QlSc-gas and liquid slag QlSl-gas; while its energy puted with Eq. (24):11)
is lost to the cooling panels Qgas-water. The energy-balance N
equation is as follows 19: Qi − RAD = Ai ∑ VFij ( J i − J j ), ................. (24)
Qgas = Qarc − gas + (1 − K post )QCO − post + QCH 4 − gas j =1

.... (19) where Ai represents the surface i area, VFij represents the
+QsSc − gas +QlSc − gas +QlSl − gas −Qgas − water . view factor from surface i to surface j and Ji and Jj represent
The coefficient of post-combustion efficiency (Kpost) in Eq. the radiosities of surfaces i and j, respectively.
(19) shows that the energy of the CO post-combustion, which For greater transparency of the following equations, a
is not consumed by either solid or liquid metal, is transferred similar notation to MacRosty3) was used, numbering the sur-
to the gas zone and mostly represents a loss of useful energy. faces relevant for the radiative heat transfer as follows: 1-
The gas-zone interactions with the solid steel QsSc-gas, liq- roof, 2-walls, 3-solid scrap, 4-liquid metal and 5-arcs. The
uid metal QlSc-gas and liquid slag QlSl-gas zones have already arcs are considered to be a black bodies and transparent
been presented with Eqs. (7), (12) and (17). The energy dis- when receiving the radiative energy.
sipated from the arcs and burners also heats the gas zone, According to the above equations, radiosities for all the
which is obtained by Eqs. (20) and (21): surfaces can be obtained with Eq. (25), with the following
simplifications regarding the incident radiation: the roof and
Qarc − gas = 0.025 ⋅ Parc , ...................... (20) walls receive the radiative heat from all the other surfaces;
⎛ ⎛ ⎛ 1300D C ⎞ ⎞ ⎞ the solid steel and liquid metal do not exchange heat by
QCH 4 − gas = ΔHTD − n ⎜ 1 − K burn ⋅ ⎜ 0.35 + 0.65 tanh ⎜ − 1⎟ ⎟ ⎟ , means of the radiation, since they are in direct contact and
⎜ ⎜ ⎟
⎝ ⎝ ⎝ TsSc ⎠ ⎠ ⎟⎠ conductive heat transfer prevails.
........................................ (21) J roof = J1 = ε1σ SBT14 + (1− ε1 )(VF1− 2 J 2 + VF1−3 J 3 + VF1− 4 J 4 + VF1− 5 J 5 ),
where Parc represents the total arc powers and 0.025 repre-
sents the approximated fraction of the arcs’ energy heating J wall = J 2 = ε 2σ SBT24 + (1− ε 2 )(VF2 −1 J1 + VF2 −3 J 3 + VF2 − 4 J 4 + VF2 − 5 J 5 ),,
D
the gas zone; ΔHT − n (PART 2) represents the burner power and
D
J sSc = J 3 = ε 3σ SBT34 + (1 − ε 3 )(VF3−1 J1 + VF3− 2 J 2 + VF3− 5 J 5 ),
(1 − K burn ⋅ (0.35 + 0.65 tanh( 1300 C
− 1)) describes the hyperbolic
J lSc = J 4 = ε 4σ SBT44 + (1 − ε 4 )(VF4 −1 J1 + VF4 − 2 J 2 + VF4 − 5 J 5 ),
T sSc
tangent approximation of the relation between the solid
scrap temperature and the burner efficiency (similar to Eq. ........................................ (25)
(3)). Meaning that the decreased burner efficiency due to where ε are the emissivities of the surfaces.
higher scrap temperature heats the gas zone instead of the At this point the radiative power of the arc is also needed,
steel. Due to the small masses of the solid slag, which is which is determined with Eq. (26). Since the conductive
usually placed on the bottom of the furnace, the solid slag heat transfer from the arcs is defined as 0.20 of the total arc

© 2012 ISIJ 406

 ISIJ International, Vol. 52 (2012), No. 3

power, the radiative heat transfer, considering the losses and kg

energy transferred to the gas (0.05), is defined as 0.75 of the to the bulk densities of the solid ( ρsSc = 900 ) and liquid
m3
total arc power:
( ρ lSc = 7 000 kg ) metals.12)
Qarc − RAD = 0.75 ⋅ Parc , ....................... (26) m3
Later, radiative heat flows can be obtained for all the sur- To determine the surface areas and view factors, the fol-
faces using Eq. (27): lowing EAF geometry and associated annotation was used
as shown in Fig. 4.
Qroof − RAD = A1 ⋅ (VF1− 2 ( J1 − J 2 ) + VF1−3 ( J1 − J 3 ) + VF1− 4 ( J1 − J 4 )) − The dimensions of the exposed walls (hwall), scrap height
VF5−1Qarc − RAD , (hscrap1, hscrap2), bath height (hbath), slag height (hslag) and
cone frustum (hcone, rcone in, rcone out) are obtained from the
Qwall − RAD = A2 ⋅ (VF2 −1 ( J 2 − J1 ) + VF2 −3 ( J 2 − J 3 ) + VF2 − 4 ( J 2 − J 4 )) −
scrap and slag melting Eqs. (42), (43), (46) and (47) dis-
VF5− 2Qarc − RAD , cussed later. The arc length (harc) is obtained from the EAF
QsSc − RAD = A3 ⋅ (VF3−1 ( J 3 − J1 ) + VF3− 2 ( J 3 − J 2 )) − VF5−3Qarc − RAD , electric model proposed in our previous paper.1) The other
QlSc − RAD = A4 ⋅ (VF4 −1 ( J 4 − J1 ) + VF4 − 2 ( J 4 − J 2 )) − VF5− 4Qarc − RAD .
non-variable dimensions are: reaf out = 6.6 m, reaf in = 4.9 m,
........................................ (27) rhole = 3.4 m, heaf up = 2.9 m and heaf low = 1.0 m.
The view factors needed in the radiative heat-transfer Eq.
The negative value of the Qi-RAD indicates that the surface (27) are obtained from different sources that have studied
i receives the radiative energy from other surfaces, since Qi- radiative heat transfer,11,13,14) assuming the inner zones of the
RAD denotes the energy flow leaving the surface i. EAF to be as follows: roof = ring (outer diameter - reaf out,
inner diameter - rhole), walls = cylinder (height - hwall, diam-
View factors and surface areas eter - reaf out), steel scrap = cone frustum attached to coaxial
As can be seen in Eqs. (25) and (27) the amount of radi- ring (cone height - hcone, cone outer diameter - rcone out, cone
ative heat transfer between the surfaces is dependent on the inner diameter - rcone in, ring diameters - reaf out and rcone out),
surface emissivities, temperatures, surface areas and view liquid metal = disc (diameter reaf in). The arcs are assumed
factors from one surface to another. At this point the rela- to be cylindrical, although in reality the arcs tend to extend
tions between the view factors, the surface areas and the at the bottom, forming a parabolic shape, which behaves
geometry of the furnace and the steel are considered. chaotically.15)
As assumed, the EAF can be represented as an enclosure Since the energy dissipated from the electric arcs prevails
with N surfaces, meaning that the sum of all the view factors in a radiative form, the view factors from the arcs to other
from one surface to the others is 1, as given by Eq. (28): surfaces in the furnace play a significant role. When dealing
N with a flat-bath situation in an EAF, the view factors from
∑ VFi − j = 1. ............................. (28) the arcs could be obtained relatively easily. However, when
j =i considering the situation prior to the flat bath, i.e., solid
By definition, a view-factor reciprocity rule applies between scrap is present, the view factors from the arc tend to change
two surfaces, as follows 29:11) as the scrap melts and are more difficult to determine. Con-
Ai ⋅ VFi − j = A j ⋅ VFj −i , ...................... (29) sidering the paper published by MacRosty,3) the arc-to-
solid-scrap view factor (VF5-3) is approximated as an expo-
meaning that a view factor from surface i to j equals the nential function 0.9 − erb / 42 , where rb is a cone-frustum
view factor from surface j to i multiplied by the ratio of base radius. It is generally known that the electrodes bore
A
surface areas Aj . The definitions 28 and 29 are useful into the solid scrap at the beginning of each basket, meaning
when determining i
the view factors between complex sur- that the view factor from the arc to the scrap increases rap-
faces, as they can be obtained from other view factors in an idly as the electrode progresses its way into the scrap and
enclosure. then gradually decreases when the scrap surrounding the
All the surfaces in the furnace are considered as simple electrode melts. In this manner we propose the view factor
geometric shapes with different orientations, such as discs, from the arc to the solid (VF5-3) or liquid (VF5-4) scrap, con-
rings, cylinders, cone-frustums or different combinations of sidering the view factor reciprocity rule, to be determined as
these. During the melting process the geometric ratios follows with Eqs. (30) and (31):
between the surfaces change as a consequence of scrap VF5−3 = (1 − VF5−1 − VF5− 2 )K sSc −lSc , ............ (30)
melting and new scrap loading. Typical phases of the melt-
ing progress can be represented by Fig. 3, which shows that VF5− 4 = (1 − VF5−1 − VF5− 2 )(1 − K sSc −lSc ), ........ (31)
there are four phases of the scrap-melting process relevant
for determining the surface areas and view factors. After the
initial charge of the furnace, the electrodes bore themselves
into the scrap, creating a cone-frustum shaped void. As the
melting progresses, the radius of the frustum increases and
eventually reaches the edge of the furnace (phase 1). After-
wards, the frustum-shaped void starts to move downwards,
exposing more and more of the wall’s surface (phase 2).
When a sufficient amount of scrap melts, the level of the liq-
uid metal is reached by the arcs (phase 3). From this point
on, the energy flow into the liquid metal increases, while the
flow into the solid scrap decreases rapidly. When the melt-
ing process approaches an end, all of the scrap is molten and
slag starts to foam (phase 4). Also visible in Fig. 3 is the fact
that the initial (solid) and final (liquid) volumes of the metal
are not the same, which is due Fig. 4. EAF geometry used for computing the surface areas and
view factors.

407 © 2012 ISIJ

 ISIJ International, Vol. 52 (2012), No. 3

where KsSc-lSc is the exposure coefficient [0–1], which deter- sesses equal characteristics (see assumption), Cp,xx can be
mines the exposure of the solid or liquid metal and is excluded from the equation.
approximated with a tangent-hyperbolic function from the As was mentioned previously, besides the standard EAF
amount of solid and liquid metal (height), as shown in Eq. zones, two more zones are used for determining the heat
(32): losses due to the cooling of the furnace, i.e., the roof and the
wall zones. The temperatures of the roof and walls can be
K sSc −lSc = 0.5 ⋅ tanh(5(hbath − hscrap1 − hscrap 2 + hcone )) + 0.5, determined with Eqs. (39) and (40):
........................................ (32)
−Q1−RAD + A1 +1A2 Qgas−water − φ1−H2O C p, H2O (Troof − TH2O−in )
A
where hbath, hscrap1, hscrap2 and hcone define the solid and liquid dTroof
,
=
metal height and the depth of the electrodes in the cone frus- dt A1d1 ρ1C p,roof
tum, respectively, and are shown in Fig. 4. Briefly, the tan- ........................................ (39)
gent-hyperbolic relation between the steel heights deter-
dTwall −Q2−RAD + Qgas−water − φ2−H2O C p, H2O (Twall − TH2O−in )
A2
mines which zone (solid or liquid) is more exposed and A1 + A2
= ,
rather than being “on-off” assures continuous dynamics dt A2 d2 ρ2C p,wall
when the scrap cone void drops to reach the bath height. The
........................................ (40)
view factors VF5-1 and VF5-2 needed in Eqs. (30) and (31)
are obtained as a cylinder to coaxial ring (arcs to roof) and where Φ1-H2O and Φ2-H2O represent the water flow rates (con-
a cylinder inside or outside the radial cylinder (arcs to stant, approx. Φ1-H2O = 60 kg/s and Φ2-H2O = 130 kg/s),
walls). The view factors VF5-1 and VF5-2 are reduced by the Cp,H2O represents the water specific heat capacity, TH2O-in
(1-Kslag) factor, where Kslag denotes the slag impact on the represents the ingoing water temperature, A1 and A2 repre-
arc’s radiative heat to the roof and walls and is presented in sent the surface areas of the roof and the walls, d1 and d2
detail in part 2(PART 2). Some additional factors to further represent the thickness of the roof and the walls, ρ 1 and ρ 2
increase or decrease the factors depending on the solid scrap represent the material density of the roof and the walls,
distribution, which exceeds the scope of this paper, will be Cp,roof and Cp,wall represent the specific heat capacity of the
addressed in another paper. roof and walls, fractions A1/(A1+A2) and A2/(A1+A2)
define the portion of the gas energy transferred to roof and
2.3. Rate of Temperature Change wall zones, respectively. Other cooling losses (furnace
As a consequence of the heat-balance equations, the tem- hearth) can be defined by Eq. (41):
perature of each EAF zone changes according to the first-
order differential Eqs. (33) to (37). A methodology similar Qwater = QsSc − water + QlSc − water + QsSl − water + QlSl − water , ..... (41)
to Bekker2) and MacRosty3) was employed (and extended to At this point the model assumes invariable water flows Φ1-H2O
slag zones), since the relation between the energy and the and Φ2-H2O across the panels without the pressure drop; as
temperature is derived from the fundamental laws (see for the real processes the flows are slightly variable.16,17)
assumption): Determining the temperatures of the roof and walls is
dTsSc QsSc 1 − Tmelt
=
(
TsSc
)
, ...................... (33)
important when the EAF model is used for optimization pur-
poses. When applying the optimization, it could be possible
dt msSc C p, sSc that the routine parameterizes the model in a way that would
dTlSc QlSc increase the power of the arcs in order to reach the energy
= , ......................... (34) optimum; however, the temperatures of the cooling panels
dt mlSc C p,lSc should be taken into account, since the high radiative impact

=
(
dTsSl QsSl 1 − Tmelt
sSl T
)
, ...................... (35)
of the arcs damages the furnace linings.

dt msSl C p, sSl 2.4. Mass Transfer Model

With the heat-balance and temperature-gradient equations
dTlSl QlSl determined, the mass-balance models for each of the EAF
= , ......................... (36)
dt mlSl C p,lSl zones can be defined. The melt rate of the solid scrap can
be defined using the following Eqs. (42) and (43):
dTgas Qgas TsSc
QsSc Tmelt
= , ........................ (37) = − , ............. (42)
dt mgas C p, gas m sSc
λsSc + C p, sSc (Tmelt − TsSc )
where Qxx denotes the energy provided to each of the zones, m lSc = −m sSc, ............................. (43)
Txx represents the temperatures of the zones, mxx represents where msSc and mlSc are the masses of the solid and liquid
the current masses of each zone and Cp,xx represents the metal and λ sSc is the latent heat of fusion of the scrap. The
specific heat capacity for each zone. Suffix xx denotes the TsSc
TsSl
appurtenant zone. The factors (1− TTsSc ) and (1− Tmelt ) represent factor Tmelt represents the fraction of energy that is melting
melt
the energy that is heating the zone and assures that its tem- the solid scrap. Other variables appearing in the equations
perature cannot rise above the melting point. were already addressed. Since the mass of Fe is much larger
Since each EAF zone is homogenous in all aspects (see than the other dissolved elements (C, Si, Mn, Cr and P), the
assumptions), whether the change of mass should appear due latent heat of fusion and the specific heat capacity is
to the addition of the material (charging a new basket, slag assumed to be equal to those of Fe.
addition, etc.), the temperature of the zone is re-determined At any point, the overall mass of liquid Fe can be
using Eq. (38): obtained from Eq. (43) subtracting all the masses of other
T m + Taddition maddition dissolved elements and adding mFe from Eq. (30) in the part
Txx = xx xx , ................ (38) 2(PART 2), which yields Eq. (44):
m xx + maddition
where xx represents the particular EAF zone and addition rep- mFe − total = mlSc − ∑ mi + mFe , i = C , Si, Mn, Cr , P .... (44)
resents the material added to the zone. Since each zone pos- i

© 2012 ISIJ 408

 ISIJ International, Vol. 52 (2012), No. 3

Equation (44) unlike Eq. (43) also includes mass changes

due to the progress of the chemical reactions. The total mass
of the liquid metal zone can be obtained by adding the actual
masses of the dissolved elements to the mass of the liquid
Fe defined in part 2(PART 2), which gives Eq. (45):
mlSc − total = mFe − total + ∑ mi , i = C , Si, Mn, Cr , P ....... (45)
i
where mi denotes the actual mass of the element in the bath.
Similarly, the melt rate of the solid slag can be defined
using Eqs. (46) and (47):
TsSl
QsSl Tmelt
m lSl = , ............... (46)
λsSl + C p, sSl (Tmelt − TsSl )
Fig. 5. Simulation timeline.
m sSl = −m lSl, ............................. (47)
where msSl and mlSl are the masses of solid and liquid slag
Table 1. Timeline of the simulation.
and λ sSl is the latent heat of fusion of the slag. The factor
TsSl Time Charge Qarc
Tmelt
represents the energy that is melting the solid scrap. The [s]
O rate O2 post C rate offgas CH4 rate Slag
[ton] [MW] 2
other variables appearing in the equations were already 1 40 48.9 0.00 0.00 0.00 10 0.60 0.00
addressed. 300 0 48.9 0.00 0.00 0.00 10 0.00 0.00
Like with the temperature-change equations, the mass of 400 0 49.1 0.00 0.00 0.00 10 0.00 1.05
each zone should be re-determined when different additions 800 0 44.4 0.00 0.00 0.00 10 0.00 1.05
are added to the zones 48:
900 25 44.4 0.00 0.00 0.00 10 0.55 1.05
m xx = m xx + maddition . ....................... (48) 1 000 0 44.4 0.95 0.20 0.30 30 0.55 1.05
1 200 0 45.3 0.95 0.20 0.30 30 0.00 1.05
3. Results and Discussion 1 400 0 50.3 0.95 0.20 0.30 30 0.00 1.05
1 500 20 50.3 1.20 0.35 0.35 25 0.55 1.05
The following section presents the simulation results
1 750 0 50.3 1.50 0.30 0.75 25 0.00 1.05
obtained from the developed mathematical models for the
1 900 0 50.0 1.50 0.30 0.75 20 0.00 1.05
mass and heat transfer and the chemical reactions from both
parts of the paper; however, only the results relevant to this 2 200 0 50.0 1.60 0.10 0.35 20 0.00 1.05
paper are presented in the following. Some of the obtained
results are compared with the available endpoint measure-
ments. Since the EAF idle data (basket loading, electrode
maintenance, etc.) does not represent any significant infor-
mation about the melting process (except the tap-to-tap
times, which are thus far not relevant for this study), it has
been removed from the measurements and is therefore not
used as a part of the model’s validation.

3.1. Simulation Timeline

Figure 5 and Table 1 show the timeline of the simulation,
relevant for obtaining the simulation results presented here
and in part 2. The actions presented in the timeline are
defined according to the actual EAF melting process
obtained from the operational data. All the units except Qarc
(MW), Charge (ton) and Time (s) are in kg/s.
Observing Fig. 5 and Table 1, Charge represents the scrap Fig. 6. Rate of change of masses in different EAF zones.
loading in tons (loaded at times 0 s, 900 s and 1 500 s), Qarc
represents the sum of the arc powers in MW, O2 rate repre-
sents the oxygen lancing rate, O2 post represents the oxygen and the melting rate decreases accordingly. Also visible, the
lancing rate for CO post-combustion, C rate represents the mass of the solid slag drops rapidly after the furnace is
carbon injection rate, offgas represents the offgas flow rate, turned on and remains close to 0 even though slag-forming
CH4 rate represents the oxygen burners injection rate and materials are charged into the furnace, which is a conse-
Slag represents the rate of addition of slag-forming materials. quence of the fast melting of those materials due to the low
rate of their addition compared to the overall mass in the
3.2. Mass Transfer Model EAF.
Figure 6 shows the results of the mass-transfer model for
each of the EAF zones. 3.3. Temperature Change Model
As can be seen, at times 900 s and 1 500 s when a new Figure 7 shows the temperature changes in the observed
basket of scrap is loaded, the mass of solid steel increases EAF zones. Similarly, as in Fig. 6, a decrease in the temper-
instantly. Shortly after each charge a rapid increase in liquid ature can be observed at times 900 s and 1 500 s when a new
steel can be observed, which is a consequence of high expo- basket of scrap is loaded. The drop in temperature in the liq-
sure of the solid steel to the arcs. When a significant amount uid metal zone is higher after the first charge, which affects
of solid steel melts more energy is transferred to liquid steel the solidifying of the liquid metal observable in Fig. 6. As

409 © 2012 ISIJ

 ISIJ International, Vol. 52 (2012), No. 3

Fig. 7. Rate of change of temperatures in different EAF zones. Fig. 8. Radiative heats between EAF surfaces.

proposed by Ref. 18), the temperature of the gas is usually

higher than the other zones, which also coincides with the
results of our model. The initial gas-zone temperature is 298
K, but this rapidly increases to around 1 000 K, which is a
consequence of the low heat capacity of the gases and the
high temperatures of the oxygen burners and electric arcs.
After each charge, the model assumes that the temperature
in the gas zone drops to that of the surroundings (298 K) and
all the other gases, except N2 and O2, are released from the
EAF freeboard.
As can be seen in Fig. 7, liquid steel temperature already
appears at time 0, even though no liquid steel is present yet.
This might seem strange; however, simulation procedure
assumes a small initial amount of liquid steel (1 kg), with
the temperature 1 809 K in order to prevent numerical diffi-
culties and simulation instability. Similarly, temperatures of Fig. 9. Energy exchange between EAF zones.
solid steel and slag remain, even though both have already
melted, which is also due to the simulation assumption that
the final mass of solid steel or slag cannot drop bellow 1 kg.
However, when solid steel and slag masses reach their min-
imum value, they are excluded from further calculations.

3.4. Heat Transfer Model

Figure 8 shows the radiative heat transfer between the
considered EAF surfaces. As can be observed, the radiative
power of the arcs (Parc-RAD) reaches up to 33 MW, which
amounts to approximately 75% of the active arc powers.
The negative characteristic of the curves denotes that the
associated surface is receiving the radiative heat, while the
positive characteristic denotes a loss of radiative power
from that surface. Observing Figs. 8 and 6 simultaneously,
we can see the relation between the radiative powers trans-
mitted to either of the zones to its mass. After t = 1 700 s a Fig. 10. Total energy input to each EAF zone.
sharp increase in the radiative power to the liquid zone
occurs, which is a consequence of the slag forming, observed after each charge, which is mostly a consequence
described in Eq. (59)(PART 2). of the increased energy transfer from the liquid and gas
Figure 9 shows the power impacts between the EAF zones zones to the solid scrap. Like with Fig. 9, lower powers to
and between the arcs and the liquid and solid metal zones. the slag zones are observable, where the reasons for that
Like with Fig. 8, a relation between the solid and liquid metal were already mentioned.
masses and the corresponding powers is observable. Figure 11 shows the temperatures of the cooling panels
As can be seen in Fig. 9 the energy flows between the and the total loss of energy (power) for each zone due to the
arcs, the solid and the liquid steel prevail over the energy cooling of the furnace vessel or due to the gases extracted
flows between the steel and the slag zones, which is a con- to offgas.
sequence of the larger masses of the steel zones compared It can be seen that a relatively constant power (energy) is
to the slag zones. Nonetheless, modeling the slag zones lost due to the extraction of gases. Shortly after the first and
separately from the steel zones is important, as each zone second charge, greater loss of energy and increased wall
possesses its own physical characteristics, which leads to temperature (Twall) occur to the liquid zone, as a conse-
different mass, heat and temperature profiles. quence of the flat-bath conditions to prevent the furnace
Figure 10 shows the total input energy (power) flows into from overheating. Also visible at times 900 s and 1 500 s is
each zone. a sharp increase in the gas cooling power (Pwater-gas) and the
A sharp increase of power to the solid steel zone can be roof-cooling temperature (Troof), which occur due to higher

© 2012 ISIJ 410

 ISIJ International, Vol. 52 (2012), No. 3

Fig. 11. Left panel: Temperatures of the roof- and wall-cooling panels; Right panel: Energy losses.

energy consumption, endpoint steel and slag composition,

are given in part 2 paper(PART 2) or were already presented.1)

4. Conclusion
In this paper an approach to the mathematical modeling
and experimental validation of heat and mass transfer and
the temperature profiles of the EAF processes is presented.
The obtained model is mainly developed in accordance with
the fundamental laws of heat and mass transfer. Parameter-
ization of the model is carried out using available operation-
al measurements and theoretical data for the EAF processes.
Some relations had to be modeled experimentally using
Fig. 12. Comparison between measured (m) and simulated (s) end- exponential and hyperbolic-tangent approximations as the
point values indicating minimum (Min.), average (Ave.) knowledge of the process is not defined mathematically or
and maximum (Max.) values. the process exhibits non-linear behavior, which is easier to
model in this manner. Regarding the presented results, the
Table 2. Comparison between the endpoint measured and simu- proposed model can be considered as being appropriate for
lated average values, including standard deviations. the aims of the study, as high levels of similarity were
achieved between the simulated results and both the theo-
Fe initial Fe end Power on time End steel retical and operational data available. The developed model
[ton] [ton] [min] temperature [K]
is primarily designed for the EAF energy and cost optimi-
Measured 85.3±1.4 81.1±0.9 45.2±3.4 1 961.0±11.6 zation and operator-training simulator. Having a relatively
Simulated 85.0±2.0 82.0±1.4 45.0±2.3 1 958.0±10.5 accurate model of the EAF processes, different scenarios
and optimization techniques can be performed to enhance
the actual EAF process. The obtained model can be used for
gas temperatures as a consequence of the oxygen burners simulating any similar EAF assembly; however, the param-
and the oxidation of combustible materials present in the eters need to be readjusted according to the particular oper-
solid steel. The phenomenon is also observable in the results ational data. Although several studies investigate the specific
presented in part 2(PART 2). Since the masses of the liquid and electric, hydraulic, mass, thermal and chemical processes of
solid slags are lower compared to the overall charge, lower the EAF, an approach as presented in this study, which
cooling powers affect the slag zones. would combine all the mentioned mechanisms of those pro-
cesses in a single model in this extent has not been found in
3.5. Model Validation the literature.
To further validate the developed EAF model and to
ensure its applicability for the initial goals of this study, the Acknowledgement
endpoint simulation results were compared to the measured The work presented in this paper was funded by Slovenian
operational data and are presented in Fig. 12 and Table 2. Research Agency (ARRS) project J2-2310 Monitoring and
The measured average values were obtained from the data Control of Steel Melt Quality in Electric Arc Furnace.
for 40 different heats, while the simulated values were
obtained from the proposed model, starting the simulation REFERENCES
with different initial conditions. At this point a comparisons
between initial and endpoint steel mass (first and second 1) V. Logar, D. Dovžan and I. Škrjanc: ISIJ Int., 51 (2011), No. 3, 382.
column), power on times (third column) and endpoint steel 2) J. G. Bekker, I. K. Craig and P. C. Pistorius: ISIJ Int., 39 (1999), No.
4, 23.
temperatures (fourth column) are given. 3) R. D. M. MacRosty and C. L. E. Swartz: Ind. Eng. Chem. Res., 44
As can be seen in Fig. 12 and Table 2, the measured and (2005), No. 21, 8067.
simulated endpoint data are very similar, which indicates the 4) J. M. M. Fernández, V. Á. Cabal, V. R. Montequin and J. V. Balsera:
Eng. Appl. Artif. Intell., 21 (2008), No. 7, 1001.
accuracy of the built model and its usability for further anal- 5) M. Kirschen, V. Velikorodov and H. Pfeifer: Energy, 31 (2006), No.
ysis, i.e., energy optimization and an operator-training sim- 14, 2926.
ulator. Other values, important for the validation, such as: 6) Y. N. Toulouevski and I. Y. Zinurov: Innovation in electric arc fur-

411 © 2012 ISIJ

 ISIJ International, Vol. 52 (2012), No. 3

naces. Springer-Verlag, Berlin Heidelberg, Germany, (2010). glasses, The Metals Society, London, UK, (1983).
7) D. Guo and G. A. Irons: 3rd Int. Conf. on CFD in the Minerals Pro- 21) E. T. Turkdogan: Fundamentals of Steelmaking, Institute of Materi-
cess Industries; CSIRO, Canberra, Australia, (2003), 651. als, London, UK, (1996).
8) M. Bergman and R. Gottardi: Ironmaking Steelmaking, 17 (1990),
No. 4, 282.
9) O. J. P. González, M. A. Ramírez-Argáez and A. N. Conejo: ISIJ Int.,
50 (2010), No. 1, 1. Appendix
10) M. R. R. I. Shamsi and S. K. Ajmani: ISIJ Int., 47 (2007), No. 3, 433.
11) R. Siegel and J. R. Howell: Thermal Radiation Heat Transfer, Taylor Table 3 gives the values of all parameters used in the
and Francis, Washington, USA, (2001). model including the corresponding units. Some of the val-
12) W. Kurz and D. J. Fisher: Fundamentals of Solidification, Trans Tech ues, such as: mEAF, A1, A2, d1, d2, Twater and TH2O-in are EAF
Publications, Ltd., Switzerland, (2005).
13) A. Feingold: J. Heat Transf., 100 (1978), No. 4, 742.
specific, while others can be obtained from different hand-
14) M. H. N. Naraghi and B. T. F. Chung: J. Heat Transf., 104 (1982), books and papers studying heat-transfer processes. Ktherm1,
No. 3, 426. Karea1, Ktherm2, Karea2, Ktherm3, Karea3, Ktherm5, Karea5, Ktherm6 and
15) J. Alexis, M. Ramirez, G. Trapaga and P. Jönsson: ISIJ Int., 40 Karea6 were obtained from the paper by Bekker;2) ε1, ε2, ε 3
(2000), No. 11, 1089.
16) J. A. T. Jones, B. Bowman and P. A. Lefrank: The Making, Shaping and ε4 were obtained from;19) Cp,sSc, Cp,lSc, Cp,sSl, Cp,lSl, Cp,gas,
and Treating of Steel; 10th ed., Chapter 10: Electric Furnace Steel- Cp,H2O, Cp,roof and Cp,wall were obtained from;2,20,21) λ sSc and
making, The AISE Steel Foundation, Pittsburgh, PA, USA, (1998), λ sSl were obtained from,12) Ktherm4, Ktherm7 and Ktherm8 were
525. obtained from approximate heat transfer coefficients for
17) S. Blažič, D. Matko and G. Geiger: Math. Comput. Simul., 64 (2004),
617. steel and slag7,9,10) normalized to the specific EAF properties
18) J. G. Bekker, I. K. Craig and P. C. Pistorius: Cont. Eng. Prac., 8 (dimensions), while Kwater1, Kwater2, Kwater3, Kwater4 and Kwater5
(2000), No. 4, 445. were obtained from approximate heat transfer coefficients
19) W. Trinks, M. H. Mawhinney, R. A. Shannon and R. J. Reed: Indus-
trial Furnaces, John Wiley & Sons, Inc., New Jersey, USA, (2004).
for furnace walls7) normalized to the specific EAF dimen-
20) E. T. Turkdogan: Physiochemical properties of molten slags and sions.

Table 3. Values of the parameters used in the model.

Ktherm1 Karea1 Ktherm2 Karea2 Ktherm3 Karea3 Ktherm4

kW m kW m kW m kW
0.2 2 0.008 2 0.2 2 0.12 2 0.05 2 0.12 2 57.5
m K kg m K kg m K kg K

Ktherm5 Karea5 Ktherm6 Karea6 Ktherm7 Ktherm8 Kwater1

kW m kW m kW kW kW
0.2 2 0.12 2 0.08 2 0.12 2 22.5 22.5 12
m K kg m K kg K K K

Kwater2 Kwater3 Kwater4 Kwater5 Tmelt Twater mEAF

kW kW kW kW
20 10 5 0.05 1 809 K 298 K 105.000 kg
K K K K

ε1 ε2 ε3 ε4 σ SB Cp,sSc Cp,lSc
J kJ kJ
0.85 0.85 0.80 0.40 5.67×10–8 0.039 0.047
m2 K 4s molK molK

Cp,sSl Cp,lSl Cp,gas Cp,H2O TH2O-in A1 A2

kJ kJ kJ kJ
0.025 0.047 0.030 0.075 298 K 25.5 m2 17.1 m2
molK molK molK molK

d1 d2 ρ 1&2 Cp,roof Cp,wall λ sSc λ sSl

kg kJ kJ kJ kJ
0.30 m 0.45 m 7 000 0.65 0.96 15.4 12.66
m3 kgK kgK mol mol

© 2012 ISIJ 412

View publication stats