Available online at www.sciencedirect.

com
Available online at www.sciencedirect.com
ScienceDirect
ScienceDirect
Energy
Available
Available Procedia
online
online 00 (2017) 000–000
atatwww.sciencedirect.com
www.sciencedirect.com
Energy Procedia 00 (2017) 000–000
www.elsevier.com/locate/procedia
ScienceDirect
ScienceDirect
www.elsevier.com/locate/procedia

Energy
EnergyProcedia
Procedia120 (2017) 000–000
00 (2017) 588–595
www.elsevier.com/locate/procedia

INFUB - 11th European Conference on Industrial Furnaces and Boilers, INFUB-11

INFUB - 11th European Conference on Industrial Furnaces and Boilers, INFUB-11
Review on mathematical models for travelling-grate iron oxide
Review on mathematical models for travelling-grate iron oxide
pellet induration
The 15th International Symposium on furnaces
District Heating and Cooling
pellet induration furnaces
MarianaAssessing the
M. O. Carvalho feasibility
a,b,
a,b,
*, Débora G.of using
Faria b
b the
, Manuel heat
García demand-outdoor
Pérezaa, Marcelo Cardosobb,
Mariana M. O. Carvalho *, Débora EsaG.K.Faria , Manuela García Pérez , Marcelo Cardoso ,
Vakkilainen
temperature function for Esa a long-term
K. Vakkilainen district
a heat demand forecast
a
Lappeenranta University of Technology - LUT Energy, Lappeenranta 20 FI-53581, Finland
b a
Lappeenranta
Federal University
a,b,c University
of Minas Gerais a of Technology
(UFMG) a - LUT
- Av. Antônio Energy,
Carlos 6627,Lappeenranta
Pampulha,
b 20 FI-53581,
Belo Horizonte,Finland
cMG 31270-901, Brazilc
b I. Andrić *, A. Pina , P. Ferrão , J. Fournier ., B. Lacarrière , O. Le Corre
Federal University of Minas Gerais (UFMG) - Av. Antônio Carlos 6627, Pampulha, Belo Horizonte, MG 31270-901, Brazil
a
IN+ Center for Innovation, Technology and Policy Research - Instituto Superior Técnico, Av. Rovisco Pais 1, 1049-001 Lisbon, Portugal
b
Veolia Recherche & Innovation, 291 Avenue Dreyfous Daniel, 78520 Limay, France
Abstract c
Abstract Département Systèmes Énergétiques et Environnement - IMT Atlantique, 4 rue Alfred Kastler, 44300 Nantes, France

Iron oxide pellets are currently one of the main feedstocks in iron-making processes, and increasing over the years. They are mostly
Iron oxidein
produced pellets are currently
moving-grate one of the
induration main feedstocks
furnaces. in iron-making
Several mathematical processes,
models and increasing
describing the pelletover the years.
induration They are
process havemostly
been
produced
published in moving-grate induration furnaces. Several mathematical models describing the pellet induration
Abstractover the last 50 years. Despite being a topic of research for a long period, a relative small amount of journal papers process have been
is
published
available inover
the the
main last 50 years.
academic Despite being
databases. a topicthe
In addition, of model
research for a long have
assumptions period,
notachanged
relative considerably
small amountover of journal papers is
time, regardless
available
the
District in
developmentthe main
heating academic
ofnetworks
faster aredatabases.
computational
commonly Inaddressed
addition,
tools. the
The purpose model
in theof this assumptions
paper is
literature astoonehave
provide not changed
a review
of the considerably
regarding
most effective overfor
the available
solutions time, regardless
mathematical
decreasing the
the development
models
greenhouse gas of faster computational
for travelling-grate
emissions iron
fromore tools. The
the induration
building purpose
furnaces,
sector. These ofsystems
this paper
pointing is toopportunities
at require
some provide a review
andregarding
high investments aspects tothe
which are available
improve
returned themathematical
accuracy
through of
the heat
models
these for travelling-grate
sales.models.
Due to the changediron ore induration
climate conditionsfurnaces, pointing
and building at some opportunities
renovation policies, heatand aspectsintothe
demand improve
future the accuracy
could of
decrease,
these models.
prolonging the investment return period.
©The
2017 Thescope
main Authors. Published
of this by
paper isby Elsevier
to Elsevier
assess theLtd.
feasibility of using the heat demand – outdoor temperature function for heat demand
©
© 2017
2017 The Authors.
The Authors. Published
Published by Elsevier Ltd.
Ltd.
Peer-review
forecast. Theunder responsibility
district of of
Alvalade, the organizing
located
Peer-review under responsibility of the organizing committee
in Lisbon of
of INFUB-11.
(Portugal),
committee was used as a case study. The district is consisted of 665
INFUB-11
Peer-review
buildings thatunder
varyresponsibility of the organizing
in both construction period andcommittee
typology. of INFUB-11.
Three weather scenarios (low, medium, high) and three district
Keywords: Ironscenarios
renovation oxide pellets;
weremoving-grate;
developed travelling-grate; induration deep).
(shallow, intermediate, furnace;To
mathematical modelling
estimate the error, obtained heat demand values were
Keywords: Iron oxide pellets; moving-grate; travelling-grate; induration furnace; mathematical modelling
compared with results from a dynamic heat demand model, previously developed and validated by the authors.
The results showed that when only weather change is considered, the margin of error could be acceptable for some applications
1.(the
Introduction
error in annual demand was lower than 20% for all weather scenarios considered). However, after introducing renovation
1.scenarios,
Introduction
the error value increased up to 59.5% (depending on the weather and renovation scenarios combination considered).
TheThe roleof
value of slope
iron oxide pellets
coefficient in the global
increased steel production
on average has grown
within the range over
of 3.8% upthe last per
to 8% decades. Currently,
decade, they aretoone
that corresponds the
Themain
the role
ofdecrease of
theiron oxide
in feedstocks
number pellets
heatinginhours
inofironmaking,theand
global
of the steel production
shareduring
22-139h is expectedhas grown
to increase
the heating over
season as the last
recent
(depending decades.
technologies Currently, ofthey
- e.g. direct
on the combination are one
reduction
weather and
of the main
- renovation
emerge feedstocks
in scenarios in ironmaking,
the steel considered).
industry. On theand the hand,
other share function
is expected to increase
intercept as recent
increased technologies
for 7.8-12.7% - e.g. (depending
per decade direct reduction
on the
- coupled
emerge scenarios).
in the steelThe
industry.
values suggested could be used to modify the function parameters for the scenarios considered, and
improve the accuracy of heat demand estimations.

© 2017 The Authors. Published by Elsevier Ltd.

* Corresponding
Peer-review author.
under Tel.: +358 50
responsibility of4675 127
the Scientific Committee of The 15th International Symposium on District Heating and
E-mail address:
* Corresponding
Cooling. [email protected]
author. Tel.: +358 50 4675 127
E-mail address: [email protected]
1876-6102
Keywords:©Heat
2017demand;
The Authors. Published
Forecast; Climatebychange
Elsevier Ltd.
1876-6102
Peer-review©under
2017responsibility
The Authors. of
Published by Elsevier
the organizing Ltd. of INFUB-11.
committee
Peer-review under responsibility of the organizing committee of INFUB-11.

1876-6102 © 2017 The Authors. Published by Elsevier Ltd.

Peer-review under responsibility of the Scientific Committee of The 15th International Symposium on District Heating and Cooling.
1876-6102 © 2017 The Authors. Published by Elsevier Ltd.
Peer-review under responsibility of the organizing committee of INFUB-11
10.1016/j.egypro.2017.07.180
 Mariana M. O. Carvalho et al. / Energy Procedia 120 (2017) 588–595 589
2 Carvalho, M. M. O et al. / Energy Procedia 00 (2017) 000–000

Different types of indurating machines can be used to produce iron ore pellets, such as travelling-grates, grate-kiln
furnaces, and pellet shaft furnaces. This paper focuses on the travelling-grate, which is the most widely used approach.
Fig. 1 shows a simplified scheme of a typical travelling-grate pelletizing process. In this system, fresh, green pellets
undergo a sequence of stages, usually updraft drying (UDD), downdraft drying (DDD), preheating, firing, post-firing
and several cooling zones.

air
green
flow
pellets cooling
UDD DDD firing zone cooling zone I zone II

air burned
inlet pellets
Fig. 1. Scheme of the iron ore pellet induration process in a travelling-grate.

Several mathematical models describing the induration process have been developed over the last 50 years. Despite
being a topic of research for a long period, a relative small amount of journal papers is available in the main academic
databases. Model assumptions have not changed considerably over time, regardless of the faster computational tools
available. Some of these assumptions and findings even seem to be contradictory; for instance, some authors consider
the solid heat conduction negligible [1, 2], while others have demonstrated that the temperature within a pellet may
vary by 100 ºC [3]. In addition, the role of the grate bars and pellet cars, as well as the hearth layer in the heat transfer
zone, is not clear in many cases. Also, some points such as ore and coal composition, and fuel addition in the firing
zone have not been fully addressed nor yet understood.
The purpose of this paper is to provide a literature review of the available mathematical models for travelling-grate
iron ore induration furnaces. From this study, a flexible code has been developed, allowing physical properties,
compositions and reaction rates to be easily changed. In this way, different modeling approaches can be compared. In
the future, this model shall be utilized to evaluate the impacts of fossil fuel substitution by renewable sources, including
air emission considerations.

2. Mathematical model

The phenomenological description of agglomeration processes consists of heat and mass transfer equations,
reaction rates, and thermochemical properties estimation. In general, the aim of the induration process model is to
predict temperatures, flow rates and compositions of the gas and solid phases along the moving grate. During the
review, it was noticed that most of papers were published on conference proceedings, making it sometimes difficult
to retrieve the original studies. Apparently, the seminal papers published on the topic date back to 1975, when
Hasenack et al. [4] and Voskamp and Brasz [5] proposed steady-state models of iron ore pellet moving beds. Since
then, a few improvements have been included in their models. The following assumptions were always used:

 The process can be described in two dimensions;

 The grate speed is constant;
 Plug-flow is assumed for the gas phase;
 The solid phase consists of a porous packed bed with uniform particle size and physical properties;
 The heat losses to the side walls and surroundings are negligible.
590 Mariana M. O. Carvalho et al. / Energy Procedia 120 (2017) 588–595
Carvalho, M. M. O et al. / Energy Procedia 00 (2017) 000–000 3

In this section, a comprehensive analysis of the existing models is provided. Each equation is presented and
discussed, focusing on the following considerations and their impact in the results.

2.1. Energy balances

The gas moves across the pellet bed while heat is exchanged by convection with the solid phase. Water vapor flows
into the gas phase due to evaporation in the drying zones, oxygen is absorbed from the gas stream for combustion and
magnetite oxidation reactions, and carbon dioxide is released into the gas due to combustion and calcination. A
complete form of the gas-phase energy balance including the sensible heat involved in mass exchange between gas
and solid phases is shown in Eq. 1:

Gc p ,g
Tg
z

4

i 1
GX i
z 
Tg
Ts

c pi dT  ha Tg  Ts r r
p
 (1)

The solid-phase energy balance, shown in Eq. 2, contains all the terms related to the enthalpies of reaction, as
well as the pellet heat conduction term. In this form, the balance is expressed by a partial differential equation (PDE),
requiring a more complex solution strategy. The boundary conditions are given by Eqs. 2a and 2b, and a uniform
pellet temperature profile is usually taken as initial condition.

Ts   2T 2 Ts  6
 s 1   b c p ,s  s  2s     Rn H r ,n (2)
t  r r r  n1

s
Ts
r r  rp

 ha Tg  Ts r r
p
 (2a) Ts
r r 0
0 (2b)

If the pellet heat conductivity s is large enough, the temperature gradient inside the solid phase can be neglected,
and an average value for the temperature throughout the whole pellet is calculated. In this case, Ts would be uniform
along the pellet radius and the solid-phase energy balance is simplified, Eq. 3:

Ts 6
 s 1   b c p ,s  ha Tg  Ts    Rn H r ,n (3)
t n 1

Most authors have adopted this approximation. On the other hand, some papers point out that heat convection and
diffusion phenomena are of the same order of magnitude [4], and that the temperature inside a single pellet may vary
as much as 100 ºC from the center to the surface [3], indicating thus that such simplification would not be valid.
The energy balances assume that the heat associated with chemical reactions is released into or absorbed from the
solid phase, eliminating the enthalpy terms from the gas-phase balance equation. The reason behind this approximation
is that, from all reactions, only water condensation may take place on the gas phase, and the heat distribution between
solid and gas phases cannot be measured or determined accurately. Thurlby [6] accounted for the latent heat of
condensation in the gas phase, whereas Majumder et al. [7] considered the heat of water evaporation/condensation in
both gas and solid phase energy balances, with opposite signs. Alternatively, Barati [2] included a coefficient i to
represent the fraction of heat of reaction retained by the solid phase. The value of the coefficient was equal to 0.5 for
the heat of condensation and to 1.0 for all others, meaning that half of the heat released by water condensation is
retained by the gas phase and all heat released (or absorbed) by other reactions is retained by (or taken from) the solid
phase. The author noticed, however, that the actual value of this coefficient is not relevant because the heat is
immediately distributed in both phases, due to a high heat transfer coefficient h.
 Mariana M. O. Carvalho et al. / Energy Procedia 120 (2017) 588–595 591
4 Carvalho, M. M. O et al. / Energy Procedia 00 (2017) 000–000

Several correlations are available for estimating the heat transfer coefficient in packed beds. Choosing an adequate
equation is relevant, as Wynnyckyj and Batterham [8] observed that different correlations may result in heat transfer
coefficients which may vary by a factor of 10. Most authors have used the so-called modified Ranz-Marshall equation
[9—11] (Eq. 4), or similar correlations to determine the heat transfer coefficient in the pellet bed. Seshadri and da
Silva Pereira [12] tested seven different heat transfer coefficient correlations for packed bed against experimental data
from a pilot scale pot grate and concluded that the modified Ranz-Marshall equation was, in fact, the most appropriate
in their case. Hasenack et al. [4] used a Chilton-Colburn j-factor correlation as a function of the Reynolds number that
results in a similar equation. This approach was also tested by Seshadri and da Silva Pereira [12], but proved to be
inferior under their experimental conditions. Küçükada et al. [3] also employed a similar equation, but using a
correction of 1.1 multiplying the Reynolds number. On the other hand, Kuuni and Suzuki [13] have observed that for
low Peclet numbers, e.g. Pe < 10, the Ranz-Marshall equation fails to accurately predict the heat transfer coefficient
when compared to various experimental data published in literature. To mitigate this problem, they have proposed a
simple model which takes a channeling length factor  into consideration (Eq. 5).

1/ 2
hd p  Re  
Nu p   2  0.6 p  Pr1 / 3 (4) Nu p  Pep (5)
g  b  61   b 

Heat transfer by radiation is not explicitly considered in any of the papers revised. Some authors claim the heat
transfer coefficient h already contains radiation effects, arguing that the models have been extensively validated by
high-temperature tests in which radiation effects are present [2,14]. Küçükada et al. [3] considered convection to be
predominant over radiation and conduction, while noting that these and other neglected phenomena may affect the
results. Conflictingly, Sadrnezhaad et al. [15] emphasized that radiation is, in fact, negligible due to a small difference
between the gas-phase and wall temperatures.
The heat transferred to the pellet cars and grate bars is the only source of loss accounted for in the travelling-grate
models. Nevertheless, several papers do not consider this factor explicitly [8, 15, 16]. Voskamp and Brasz [5] consider
a constant value for the grate bars specific heat capacity in each zone. In addition, the grate bars are one layer high
(i.e. the domain is discretized into N + 1 layers, from which the first bottom one always correspond to the grate bars),
and the surface area is estimated in 20 m2. Hasenack et al. [4] recognized the necessity to account for losses to the
grate bars and pellet cars, whilst these were not implemented into the model. Thurlby [6] uses a two-layer domain for
calculating the heat transferred to the grate bars, but use the same heat transfer coefficient and surface area for the
pellet bed.
Thurlby et al. [1] proposed a simpler correlation to estimate the convective heat transfer coefficient in the grate
bars, Nu = Re1/2, from pot grate experiments. In the paper, the authors mentioned that a lower coefficient is used for
the supporting structure. Additionally, the study recognized the importance of grate bars and supporting structures in
the furnace heat transfer, and attributed four layers in the domain discretization for it. Barati [2] used the same
correlation; nonetheless, the actual height and surface area were not clarified in any of these papers.

2.2. Mass balances

To handle the constant variation in the composition, component mass balances are required, considering that
changes in the concentrations affect not only reaction rates but also the thermochemical property estimations. Yet,
most authors do not explicitly calculate component mass balances for all chemical species present in the system. The
seminal papers usually calculate the variations in the gas and solid phase of moisture content and solid phase magnetite
content, if present [1, 5, 14]. Hasenack et al. [4] also considered the oxygen concentration in the gas phase to estimate
the rate of magnetite oxidation at temperatures above 500 ºC. The most probable reason for neglecting mass
components is to reduce the number of differential equations to be solved, therefore reducing computational effort.

Some papers particularly present component mass balances for both gas and solid phases [2, 6, 7, 16]. In the gas
phase, the overall mass balances account for changes in the gas superficial mass flow rate (Eq. 6), and component
592 Mariana M. O. Carvalho et al. / Energy Procedia 120 (2017) 588–595
Carvalho, M. M. O et al. / Energy Procedia 00 (2017) 000–000 5

mass balances calculate the resulting composition of the gas phase (Eq. 7). Is it important to notice that the gas
superficial flow rate is not kept constant in Eq. 7, as considered by several authors [2, 7]. Keeping G constant in Eq.
7 results in constant mass fraction of non-reacting compounds (e.g. N2) and others, depending on the grate zone. The
domain discretization reduces the effect of such approximation by making G constant just over one iteration cell.
Solid-phase mass balances are similar to the ones written for the gas phase, with the solid bulk density, b replacing
the gas superficial mass flow rate and the t coordinate replacing z. However, all articles revised consider b uniform
throughout the grate, resulting in Eq. 8. The uniform density approximation is less critical when applied to the solid
phase than to the gas phase, as not only the variations are smaller but also the thermochemical properties are usually
estimated as a function of the temperature alone. Yet, water evaporation and carbon combustion may reduce the green
pellet specific mass by more than 10 %, not to mention calcination. On the other hand, estimating solid density
variation may be challenging, because it involves other factors such as bed void fraction and pellet porosity which are
usually, but not always, kept uniform.

G 4
X i 1  G  X j 10
  Ri (6)  Ri  X i (7)  s (1   b )  Rj (8)
z i 1 z G  z  t j 1

2.3. Pressure drop

In most papers revised, the Ergun equation [17] is used to estimate the relation between pressure drop and gas
flowrate across the pellet bed. This semi-empirical correlation is shown in Eq. 9, in which the first term on the right
represents the viscous forces and the second one the inertial contribution to the pressure gradient. Wynnyckyj and
Batterham [8] have demonstrated that, even though the latter term is usually the largest, both terms are of the same
order of magnitude under typical pelletizing conditions. Thus, ignoring the viscous term as in the model of Pomerleau
et al. [16] may lead to inaccurate results and should be avoided.

P 1501   b  1.751   b  2
2
  gG  G  A1 g G  A2G 2 (9)
z d p 2  b3 g d p b3 g

Despite being the most common pressure drop correlation used for packed beds, the Ergun equation has some
limitations. For instance, it assumes isothermal incompressible flow when integrating the mechanical energy balance.
Thus, Voskamp and Brasz [5] and Hasenack et al. [4] have used the formulation proposed by Szekely and Carr [18],
which includes temperature-dependent fluid density and viscosity (Eq. 10). Although the non-isothermal formulation
has proven to be more accurate, the authors themselves observed that if the packed bed can be divided into small
enough sections, the original equation could be satisfactory, provided density and temperature are recalculated at the
integral mean temperature in every iteration.

G2   g ,0  H
  A1  g G  A2G 2 dz  
P( H )
ln     g dP (10)
 2   0 P ( 0)
b g 

Wonchala and Wynnyckyj [19] have observed that the kinetic energy term in Eq. 10 is, in fact, 3 to 5 orders of
magnitude smaller than the friction work and pressure energy terms. If the kinetic energy term can be neglected, the
resulting equation is separable, dismissing the need of an iterative procedure to find the solution. In turn, the gas phase
density could be expressed as a function of temperature and pressure by using any appropriate equation of state, e.g.
the ideal gas law.
 Mariana M. O. Carvalho et al. / Energy Procedia 120 (2017) 588–595 593
6 Carvalho, M. M. O et al. / Energy Procedia 00 (2017) 000–000

2.4. Drying process

The shrinking core model (SCM) is a common approach to describe water evaporation and chemical reactions in
in porous spherical particles. In basic terms, the SCM applied to gas-solid non-catalytic reactions proposes that a
reaction front travels from the particle surface towards its center, leaving a shell composed by inert material and
reaction products around an unreacted shrinking core [20]. The mass transfer coefficient is the result of three
resistances in series: (i) gas-film diffusion resistance; (ii) pore diffusion across the reacted shell; and (iii) chemical
reaction kinetics.
A three-stage SCM drying model seems to be the best description of the drying process [2, 4, 14, 21]. In the first
stage, the gas-film diffusion between the pellet surface and the bulk gas phase controls the drying process (as shown
in Eq. 11). Whenever the saturation concentration in the liquid film around the pellet surface is higher than the
concentration in the gas phase, the rate is positive, meaning that evaporation occurs. If the gas phase is already
saturated with water, condensation takes place, and the resulting drying rate is negative. During the first drying stage,
the pellet moisture content is implicitly assumed homogeneous. It continues until the water concentration reaches a
critical concentration, often around 120 kg/m3.
The second drying stage occurs when the pellet moisture content drops below the critical value. From this point, a
dry shell is formed around the pellets, and the pore diffusion resistance starts playing an important role in the drying
process. The resulting equation combines pore diffusion inside the pellet dry shell and film diffusion from the surface
to the gas bulk, see Eq. 12.
At some point of the drying process, the insufficient pellet temperature may be the main resistance to water
evaporation, and Eq. 12 is no longer able to accurately describe the drying rate. At this moment, the heat transfer is
limiting the drying rate. The relation describing the third step of the drying process is given by Eq. 13. Whenever the
drying rate given by Eq. 13 is smaller than the one resulting from Eq. 12, the third drying stage prevails, and the
process is limited by heat transfer. The wet bulb temperature Tw can be, however, difficult to estimate.

  ha Tg  Tw   rH 2O 
3

RH2O  a  k H2O C eq
H 2O C g
H 2O  (11) RH 2O 
a C Heq2O  C Hg 2O (12) RH 2O    (13)
1 rp ( rp  rH 2O ) H r ,H 2O  rp 

k H 2O rH 2O DHe 2O

The mass transfer coefficient, k H 2O is given by a similar expression used to estimate the heat transfer coefficient,
replacing the pellet Nusselt number Nu by its mass transfer correspondent - the Sherwood number Sh - and the Prandtl
number Pr by the Schmidt number Sc. The expression may also be used to calculate the mass transfer coefficients of
other components in the gas film surrounding the pellets.

2.5. Chemical reactions

Chemical reactions may occur during the induration process, depending on the iron ore and green pellet
composition. The SCM is also used to calculate the rates of reaction. Contrary to drying, the chemical reaction
resistance, as well as the film and pore diffusion, are relevant to the model. The expression is given by Eq. 14. In some
cases, however, the chemical reaction is the limiting step, resulting in Eq. 15. The rate constant is given by the
Arrhenius equation, shown in Eq. 16. All reactions considered in the models revised and the relevant parameters are
listed in Table 1.

 j a COeq  COg  (14) Rn  k 'n 1   b  s X j (15)

  Ea ,n 
Rn  2 2 k 'n  k0,n exp   (16)
 RTs 
2
1 rp ( rp  rj ) r p
 e
 2
kO2 rj D j rj k 'n
 Carvalho, M. M. O et al. / Energy Procedia 00 (2017) 000–000 7
594 Mariana M. O. Carvalho et al. / Energy Procedia 120 (2017) 588–595

Table 1. Main parameters used for reaction rate estimation.

Chemical Reaction k0 Ea/R [K] Reference Reaction rate
2 FeOOH(s) → Fe2O3(s) + H2O(l) 3.04 · 106 12000 [1]
CaCO3(s) → CaO(s) + CO2(g) 8.3 · 106 18300 [6] Eq. 15
MgCO3(s) → MgO(s) + CO2(g) 2.5 · 105 11200 [6]
4 Fe3O4(s) + O2(g) → 6 Fe2O3(s) 3.47 · 106 19799 [22]
Eq. 14
C(s) + O2(g) → CO2(g) 595.6 · TS 17970 [2]

It is important to notice that not all the reactions are included in every model. For instance, from all the papers
revised, only Thurlby et al. [1] took goethite into consideration in the iron ore composition. With few exceptions (i.e.
Swedish ores), magnetite is a rather minor component in most iron ores; nonetheless, it is included in most models.
Calcination is rarely considered, and it is usually reaction-limited. Sadrnezhaad et al. [15], however, have used the
SCM also to represent limestone decomposition. Finally, carbon combustion rates are usually calculated using
experimental data for char. This approach is questionable, considering that some coals used in the process may have
fixed carbon content below 70 %. In addition, some other components such as Sulphur and nitrogen cannot be
considered.

2.6. Thermochemical properties and transport coefficients

The role of the thermochemical properties, such as specific heat capacity and density, as well as the estimation of
the transport coefficients (viscosity, thermal conductivity and diffusivity) in the model accuracy has not been discussed
in the revised papers. In several cases, average values or temperature-only dependent correlations have been used, not
considering the varying composition in both gas and solid phases. For instance, Voskamp and Brasz [5] considered
that all thermochemical properties were constant in each zone. Other authors used temperature-dependent correlations
for air and hematite to estimate the specific heat capacity and density of the gas and solid phases [1, 3].
The diffusion coefficient through the gas phase was given by Fuller's correlation [23] in all reviewed papers. This
equation, however, depends only on the gas phase temperature and pressure, not considering the continuously
changing composition. The effective diffusion coefficient, which represents a correction of the diffusion coefficient
to consider the pellet porosity and tortuosity, is the most common approach. This simplified correction implies that
the pores are large enough so that pore diffusion resistance (e.g. Knudsen diffusion) is not relevant.

3. Solution strategy and model validation

The finite difference method is utilized to solve the differential equation system in most of the papers revised.
Explicit scheme is usually preferred, because it is simpler to implement. However, the number of layers in the vertical
direction as well as the step sizes along the grate must be small enough to produce accurate results. Some numerical
integration techniques have also been proposed to solve systems which do not involve PDEs, i.e. not considering
temperature gradients within the pellets.
Validation is also an important issue in modeling travelling-grate induration furnaces. Usually, the only plant
measurements which are available are pressure drop across the bed and temperature in the windboxes. Although the
pressure drop is useful to estimate the gas flow, the gas temperatures above pellet bed and below the grate bars do not
provide enough data to validate the model. So far, no plant equipped with thermocouples inside the bed have been
reported, and data extracted from pot-grate pilot plants have been used instead.

4. Conclusions

The main mathematical models available in the literature for travelling-grate iron ore induration furnaces have been
revised in this study. In general, the phenomenological approach to this problem is well established, as all the models
consist of mass and energy balances, reaction rates, and thermochemical properties estimations. On the other hand,
 Mariana M. O. Carvalho et al. / Energy Procedia 120 (2017) 588–595 595
8 Carvalho, M. M. O et al. / Energy Procedia 00 (2017) 000–000

few advances were made recently, neglecting modern and powerful computational tools. In addition, the impact of
some assumptions is still unclear. Yet, the main issue is still the lack of reliable experimental data to adjust and validate
the results. Statistical treatment of experimental data has not been discussed up to this moment.
Other aspects of iron ore pellet induration modelling were not covered in this paper. For instance, variable pellet
bed void fraction and porosity, equations to predict spalling and burned pellet quality have not been discussed here,
even though several publications addressing this issue could be found. Additionally, the role of sintering kinetics in
pellet induration is not mentioned, despite being an important aspect in modeling sintering strands. With the increasing
importance of iron ore pellets in iron and steel manufacturing, more attention should be given to this topic in the
future.

Acknowledgements

The authors would like to thankfully acknowledge the public research funding provided by: TEKES, the Finnish
Funding Agency for Innovation, under the ‘Neo-Carbon Energy’ project (number 40101/14) and CAPES Foundation
(Ministry of Education of Brazil) under the process number BEX 7306/15-6.

References

[1] J. Thurlby, R. Batterham, R. Turner. Development and validation of a mathematical model for the moving grate induration of iron ore pellets,
Int. J. Miner. Process. 6 (1979) 43–64.
[2] M. Barati. Dynamic simulation of pellet induration process in straight-grate system, Int. J. Miner. Process. 89 (2008) 30–39.
[3] K. Küçükada, J. Thibault, D. Hodouin, G. Paquet, S. Caron. Modelling of a pilot scale iron ore pellet induration furnace, Can. Metall. Q. 33
(1994) 1–12.
[4] N. Hasenack, P. Lebelle, J. Kooy. Induration process for pellets on a moving strand, in: Mathematical process models in iron and steelmaking,
1975, pp. 6–16.
[5] J. H. Voskamp, J. Brasz. Digital simulation of the steady state behaviour of moving bed processes, Meas. Control 8 (1975) 23–32.
[6] J. Thurlby. A dynamic mathematical model of the complete grate/kiln iron-ore pellet induration process, Metall. Trans. B 19 (1988) 103–112.
[7] S. Majumder, P. V. Natekar, V. Runkana. Virtual indurator: A tool for simulation of induration of wet iron ore pellets on a moving grate,
Comput. Chem. Eng. 33 (2009) 1141–1152.
[8] J. Wynnyckyj, R. Batterham. Iron ore sintering and pellet induration processes, in: C. E. Capes (Ed.), 4th International Symposium on
agglomeration, AIME, Toronto, Canada, 1985, pp. 957–994.
[9] W. Ranz. Friction and transfer coefficients for single particles and packed beds, Chem. Eng. Prog. 48 (1952) 247–253.
[10] W. Ranz, W. Marshall. Evaporation from drops – part i, Chem. Eng. Prog. 48 (1952) 141–146.
[11] W. Ranz, W. Marshall. Evaporation from drops – part ii, Chem. Eng. Prog. 48 (1952) 173–180.
[12] V. Seshadri, R. O. da Silva Pereira. Comparison of formulae for determining heat transfer coefficient of packed beds, ISIJ Int. 26 (1986), 604–
610.
[13] D. Kunii, M. Suzuki. Particle-to-fluid heat and mass transfer in packed beds of fine particles, Int. J. Heat Mass Transfer 10 (1967) 845–852.
[14] V. Seshadri, R. O. da Silva Pereira. Mathematical simulation of induration of iron ore pellets in pot grate, in: C. E. Capes (Ed.), 4th
International Symposium on Agglomeration, AIME, Toronto, Canada, 1985, pp. 729–744.
[15] S. Sadrnezhaad, A. Ferdowsi, H. Payab. Mathematical model for a straight grate iron ore pellet induration process of industrial scale, Comput.
Mater. Sci. 44 (2008) 296–302.
[16] D. Pomerleau, D. Hodouin, E. Poulin, A first principle simulator of an iron oxide pellet induration furnace – an application to optimal tuning,
Can. Metall. Q. 44 (2005) 571–582.
[17] S. Ergun, Fluid flow through packed columns, Chem. Eng. Prog. 48 (1952) 89–94.
[18] J. Szekely, R. Carr, On nonisothermal flow of gases through packed, Trans. Metall. Soc., AIME 242 (1968) 918–921.
[19] E. Wonchala, J. Wynnyckyj, Nonisothermal flow of gases through packed beds, Metall. Mat. Trans. B 18 (1987) 279–280.
[20] O. Levenspiel, Chemical Reaction Engineering, 3rd ed., John Wiley & Sons, 1999.
[21] R. O. da Silva Pereira, V. Seshadri, Secagem de pelotas de minerio de ferro, Metalurgia ABM 41 (1985) 141–144.
[22] D. Papanastassiou, G. Bitsianes, Mechanisms and kinetics underlying the oxidation of magnetite in the induration of iron ore pellets,
Metallurgical Transactions 4 (1973) 487–496.
[23] E. N. Fuller, P. D. Schettler, C. J. Giddings, A new method for prediction of binary gas-phase diffusion coefficients, Ind. Eng. Chem. 58 (1966)
19–27.