Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

REVIEW

www.steel-research.de

A Review on the Modeling of Gaseous Reduction of Iron


Oxide Pellets
Ariyan Zare Ghadi,* Mohammad Sadegh Valipour, Seyed Masoud Vahedi,
and Hong Yong Sohn

gasified coal can be built inexpensively


The investigations on direct reduced iron (DRI) have commonly been made at two and in a wider range of annual capacities.
different scales: the pellet scale and the macroscopic or reactor scale. The The total output from direct reduction facil-
reduction of a single pellet is typically investigated for developing kinetics of ities has grown from practically nil in 1970
to 87 million tons in 2017.[1]
chemical reactions accompanied by heat and mass transport, whereas in the
A DRI shaft furnace is a chemical reactor
macroscopic view, the reduction process is investigated in the reactor scale. that prepares a set of conditions for convert-
However, these two aspects are not independent of each other and usually the ing iron oxides to sponge iron by gas–solid
kinetic parameters that are used in the reactor modeling come from the single reactions. It is mainly composed of three
pellet study. This article reviews the most relevant models in the literature with parts: 1) an upper part, containing pellet
regard to direct reduction reactions in the pellet scale together with more recently charging system; 2) a middle part, including
a reaction zone as well as a reducing gas
developed new approaches. This review also presents a critical assessment of the
charging system; and 3) a lower part, con-
previously proposed models in terms of which models are applicable under what taining DRI cooling and discharge systems.
conditions and to what types of the solid structure. The flaws and pitfalls in some With respect to heat and mass transfer, the
of these models are also pointed out with cautions. The present article is broadly middle part is the most important section
divided into experimental and mathematical sections. that strongly affects the performance as well
as the operation of the furnace.
As shown in Figure 1, iron oxide pellets
are continuously charged from the top of
1. Introduction the furnace and move downward by gravity along the middle part
of the DRI shaft furnace, known as the moving-bed reactor. The
An alternative process in the production of iron for steelmaking strongly reducing hot gas mixture (rich in hydrogen and carbon
is direct reduction without iron melting based on a reducing gas monoxide) is admitted to the lower section of the reduction zone
mixture of hydrogen and carbon monoxide, as the obtained to supply the necessary heat for preheating and to reduce the
sponge iron competes favorably with scrap in the electric arc charged pellets. The reducing gas mixture is typically produced by
furnace steelmaking. In the direct reduction process, iron oxides the partial reforming of natural gas mixed with a portion of the top
are reduced at moderate temperatures and the product, direct gas from the furnace. Finally, the oxygen in the iron oxide pellets is
reduced iron (DRI), has essentially the same size and shape removed by reducing agents (hydrogen and carbon monoxide)
as the original raw pellet. Unlike the iron blast furnace, which according to the following set of heterogeneous chemical reactions.
requires large capital investments and high-quality metallurgical 3Fe2 O3 þ H2 ! 2Fe3 O4 þ H2 O (1)
coke, a direct reduction plant utilizing reformed natural gas or
Fe3 O4 þ H2 ! 3FeO þ H2 O (2)
Dr. A. Zare Ghadi, Dr. M. S. Valipour, S. M. Vahedi
FeO þ H2 ! Fe þ H2 O (3)
Faculty of Mechanical Engineering
Semnan University
3Fe2 O3 þ CO ! 2Fe3 O4 þ CO2 (4)
P.O. Box 35131-19111, Semnan, Iran
E-mail: A.zareghadi@semnan.ac.ir
Fe3 O4 þ CO ! 3FeO ! CO2 (5)
Prof. H. Y. Sohn
Department of Metallurgical Engineering FeO þ CO ! Fe þ CO2 (6)
University of Utah
Salt Lake City, Utah, USA However, some homogenous reactions may also occur among
Prof. H. Y. Sohn the gaseous species in the reducing gas mixture. The most
Department of Chemical Engineering important of these are as follows
University of Utah
Salt Lake City, Utah, USA 2CO ↔ C þ CO2 (7)
The ORCID identification number(s) for the author(s) of this article
H2 O þ CO ↔ H2 þ CO2 (8)
can be found under https://doi.org/10.1002/srin.201900270.
DOI: 10.1002/srin.201900270 3H2 þ CO ↔ CH4 þ H2 O (9)

steel research int. 2019, 1900270 1900270 (1 of 16) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
l

www.advancedsciencenews.com www.steel-research.de

Ariyan Zare Ghadi received his bachelor


degree in Mechanical Engineering
from Babol Noshirvani University of
Technology in Iran. He completed
his master’s degree and Ph.D. in
Mechanical Engineering, Energy
Conversion field, at Semnan in Iran.
His research interests include
computational fluid dynamics, transport
phenomena in porous media, modeling
of DRI Processes, Nanofluid flow, discrete element method
(DEM), and discrete particle model (DPM).

Mohammad Sadegh Valipour is an


associate professor of mechanical
engineering at Semnan University. He
received his bachelor’s degree in
mechanical engineering from Sistan &
Baluchestan University. He completed
his master’s degree and Ph.D. at Sharif
University of Technology. He followed
his studies as a visiting researcher at
Nagoya University. He is an expert in
CFD, modeling of heterogeneous gas–solid reactions,
transport phenomena in porous media, energy saving,
Figure 1. Schematic configuration of the middle part (reacting zone) of energy modeling, and modeling of DRI processes. Director
the DRI shaft furnace as a moving-bed reactor. of Semnan University Technology Incubator and Dean of the
Mechanical Engineering Faculty at Semnan University are
some of his important activities in recent years.
The modeling of iron ore pellets undergoing reduction Seyed Masoud Vahedi graduated in
processes in a chemical reactor has received considerable atten- Mechanical Engineering from Semnan
tion due to the economic relevance of the product obtained.[1–4] University, Semnan, Iran, in
Developments in the field of direct reduction technology have September 2016. His thesis dealt with
consistently progressed to advance in the knowledge of the kinet- the numerical simulation of polymer-
ics of the reduction process.[5–8] Since long ago, efforts have been based drug-eluting stents (DESs).
devoted to the determination of the controlling step in the After completing master’s degree, due
reaction–transport process that takes place in the pellet. These to his coding and statistical skills, he
investigations can be categorized into two major groups: experi- was hired in some research groups.
mental and mathematical modeling work, each of which will be He was involved in the Edelman
discussed below. group at Massachusetts Institute of Technology (MIT) to
work on Drug-Eluting Stents. Meanwhile, he started
collaboration with Semnan University on Modeling of
2. Experimental Section Gaseous Reduction of Iron Oxide Pellets. His main
research interests are numerical simulation of heat and
Based on experimental observations, Joseph[2] developed reduc- mass transfer, especially in biological systems, and multi-
ibility tests on several natural ores. Some of the results of his phase flows.
experiments are shown in Figure 2. These results indicated that
the porosity of iron ore particles is one of the most important
factors in controlling reducibility. He showed that the reducibil- In this way, oxygen removed from the surface of the wustite
ity expressed as the reciprocal of time required for 90% reduction results in an increase in the concentration of ferrous iron
varied directly with the porosity. (Feþþ ). A Feþþ concentration gradient is built across the wus-
Edstrom[3] proposed a reduction mechanism based on solid- tite layer. Some of the ferrous ions and electrons migrate to
state diffusion of ferrous ions through the wustite layer according nucleation sites where they precipitate as metallic iron, but
to the reaction types (3) and (6). The reactions are the summation other ferrous ions and electrons diffuse across the wustite
of the following typical subreactions and magnetite layers where they react with magnetite to pro-
duce wustite and magnetite and with hematite to produce
FeO ¼ Feþþ þ O , O þ CO ¼ CO2 þ 2e,
(10) wustite and magnetite according to the following reactions,
and Feþþ þ 2e ¼ Fe respectively

steel research int. 2019, 1900270 1900270 (2 of 16) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
l

www.advancedsciencenews.com www.steel-research.de

Figure 2. Effect of pellet porosity on the rate of reduction for different iron ore samples. Reproduced with permission.[2] Copyright 1936, The Minerals,
Metals & Materials Society.

Feþþ þ 2e þ Fe3 O4 ¼ 4FeO reduction proceeds topochemically, he initially concluded


(11)
and Fe þþ
þ 2e þ 4Fe2 O3 ¼ 3Fe3 O4 that the overall rate of reduction was controlled by the interfacial
chemical reactions. Later, Warner[12] indicated the significance of
However, Lu[4] claimed that the solid-state diffusion of ferrous gaseous diffusion through the iron product layer. Subsequently,
ions through wustite is much faster than the gaseous diffusion Olsson and McKewan[13] conducted an experimental study
rate of either hydrogen or carbon monoxide through the pores of under similar conditions and showed that pore diffusion had
the ore particles at the temperature normally used in a direct a significant effect on the overall rate under the specific set of
reduction (DR) process so that solid-state diffusion is not usually conditions.
a rate-controlling step. It is noted that such a claim of the rate- The overall reduction rate of a spherical pellet can be
controlling step should not be considered universally acceptable expressed by the following shrinking core model
because it depends on a number of operating conditions such as
 
temperature, pellet size, and pellet porosity and morphology. 1
r p d0 1  ð1  FÞ3 ¼ κ1 t (12)
Also, the nucleation and growth of iron crystals result in the
shrinkage of solid volume and an increase in the porosity of
the metallic phase and facilitate the diffusion of the reducing where r p , d0 , F, and κ 1 are, respectively, the original radius of
gases from the outer surface to wustite–iron interface. pellet, molar density of the solid, overall extent of reaction,
Edstorm[3] has shown that the volume of pellet increases by and the rate constant. Figure 3 shows a part of McKewan’s
about 25% by the transformation of hematite to magnetite experimental data obtained under isothermal conditions for
and by another 7–13% by the transformation of magnetite to different pellet diameters, indicating that Equation (12) is
wustite. satisfied.
Indeed, the reduction of iron oxide pellet, such as other Turkdogan and Vinters[14,15] and Turkdogan et al.[16] reported
gas–solid reactions, contains the following steps: 1) transfer of on an extensive experimental study on the reduction of hematite
reactants across the boundary layer outside the pellet; 2) diffu- by hydrogen. They illustrated that three regimes may exist in the
sion of the gaseous reactants through the porous solid product reduction of iron ore pellets: uniform internal reduction, limiting
layer; 3) interfacial chemical reactions; 4) diffusion of the gaseous mixed control, and control by gas diffusion in the porous iron
products through the porous solid product layer; and 5) transfer layer. In the uniform mode of reduction, the time of reduction
of the gaseous products across the boundary layer. McKewan[9–11] is independent of pellet size. When the reduction is completely
conducted an experimental investigation on the rate-controlling controlled by pore diffusion, the time of reduction is proportional
mechanism for dense hematite pellet. While recognizing that the to the square of pellet diameter.

steel research int. 2019, 1900270 1900270 (3 of 16) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
l

www.advancedsciencenews.com www.steel-research.de

Figure 3. Hematite pellet reduction by hydrogen (left) reduction rate and Equation (12) (right) Equation (12) for different pellet radius. Reproduced with
permission.[9] Copyright 1960, Association for Iron & Steel Technology.

According to the literature, structural deviations may either


augment the diffusional resistance through sintering or make
a more porous structure, which would increase the effective
diffusivity.[17] From the study conducted by Turkdogan and
Vinters,[14] it is observed that porosity increases with increasing
reduction temperature, and accordingly, effective diffusivity
increases with increasing temperature. This finding is not uni-
versal and may be changed by varying experimental conditions.
Trushenski[18] carried out a set of experimental runs to indi-
cate that the nontopochemical model is appropriate to describe
the reaction rates of porous pellets. He considered 1) the change
of particle size during reduction; 2) local changes in void fraction;
and 3) the reduction reactions in any region. He also concluded
that the rate constant of FeO-Fe reaction is higher than that of
Fe3O4-FeO, which is higher than that of Fe2O3-Fe3O4 reaction at
any certain temperatures.
Szekely and El-Tawil[19] reported on the reduction of hematite
disks with a mixture of hydrogen and carbon monoxide. Figure 4
shows a sample of their measurements. They observed that the
time required to attain a given extent of reduction is markedly
dependent on the composition of the reducing gas: hydrogen
results in much shorter reduction times than carbon monoxide. Figure 4. Effect of gaseous composition on the time required to attain a
Furthermore, this dependency of reduction time on composition given degree of reduction. Reproduced with permission.[19] Copyright
is markedly nonlinear. 1976, Springer Nature.
Towhidi and Szekely[20] reported on a comprehensive experi-
mental study on the rate of the reduction of hematite pellet, using
pure hydrogen or a mixture of hydrogen and carbon monoxide as the overall rate of reduction was markedly dependent on gas
the reducing gas. For the runs with hydrogen as the reducing composition and temperature. The region of diffusion control
agent, they found that the overall rate was diffusion controlled was much less clearly defined than in the runs carried out with
for conversion levels between about 0.5 and 0.9; in contrast, both pure hydrogen. Based on morphological studies, they observed
diffusion and chemical effects may be operative below conver- that in their experimental conditions, no significant iron phase
sion levels of 0.5. However, the precise value of these limits appeared until virtually the whole pellet was reduced to wustite.
depended on temperature and particle size. In the runs carried A typical experimental observation by these authors is given in
out with hydrogen–carbon monoxide mixtures, they found that Figure 5.

steel research int. 2019, 1900270 1900270 (4 of 16) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
l

www.advancedsciencenews.com www.steel-research.de

Figure 5. Reduction rate versus time and the effect of (left) temperature on the reduction rate by hydrogen (right) gas composition on the rate of
reduction by a mixture of hydrogen and carbon monoxide at 1173K. Reproduced with permission.[20] Copyright 1981, Taylor & Francis.

Al-Kahtany and Rao,[21] Rao and Moinpour,[22] and Rao and (ε ¼ 0.08) compacts had a minimum reduction rate around
Al-Kahtany[23] performed experiments to obtain intrinsic kinetic 650  C. This was related to sintering and densification of the
data, without any interfaces caused by mass transfer effects. To formed iron phase around the unreduced oxide grains. By
accomplish the aforementioned goal, experiments were con- increasing porosity or decreasing grain size of compacts, the
ducted at the lowest possible temperature and using very thin slowing down in the rate was lessened till it vanished in the
solid samples. When the reducing gas contains carbon monox- highly porous compacts.
ide, carbon deposition is more likely due to the catalytic effect of Paul and Mukherjee[27] conducted experimental studies on the
reduced iron. Therefore, this reaction may affect the reduction kinetics of reduction of iron ore pellets by carbon monoxide
rate of hematite pellet, based on the results of Towhidi and under isothermal and nonisothermal conditions. They indicated
Szekely[24] on the influence of carbon deposition by Reaction that nonisothermal reduction kinetics of iron ore pellets is not
(7) on the reduction kinetics of hematite with a mixture of carbon controlled by a single rate-controlling step. They showed three
monoxide and hydrogen. They concluded that there was no controlling mechanisms and first-order reaction.
carbon deposition above 900  C or in the absence of carbon mon- Most of the previous experimental studies on the reduction
oxide. Below 900  C, carbon deposition may occur and prevent the of pellets used single or two-component gas mixture (e.g., H2,
conversion to reach completion. They also claimed that the maxi- H2—N2, H2—CO, CO, and CO—CO2) as the reducing gas.
mum rate of carbon deposition occurred between 500 and 600  C. However, in the industrial practice of direct reduction, it is neces-
When the hydrogen content of the reducing gas is high, the reduc- sary to understand the reduction rate and product quality in a mix-
tion reaction is more important than carbon deposition; in con- ture of H2—H2O—CO—CO2. Proctor et al.[28] performed an
trast, when the carbon monoxide content of the reducing gas is experimental investigation on the reduction of hematite pellets with
high, carbon deposition may overwhelm the reduction process. gaseous mixtures of H2—H2O—CO—CO2. They claimed that the
El-Geassy and Rajakumar[25] experimentally investigated the instantaneous rate of reduction of the pellet depends on its oxygen
reduction behavior of wustite micropellets. They determined that content, the reduction potential of the gas phase, the temperature,
the rate of reduction of wustite by H2 was higher than that by CO. the physical characteristics of the ore, and the structural changes
It was concluded from their results that small additions of H2 to during the reduction. Using X-ray measurements of the reduced
CO resulted in significant augmentation of the reduction rates. iron, they confirmed that the hematite was reduced completely
This was related to the increased formation of iron nuclei on the to wustite before metallic iron was formed. They reported that
wustite surface in the presence of H2. Finally, a correlation the chemical reaction in the wustite–iron interface was rate-
between the rate of reduction and the contents of H2 and CO controlling in the range of 30%<F<90%, described by
was obtained.
1
El-Geassy and Nasr[26] investigated the influence of the struc- r p d0 f 1 ¼ κ1 t, f 1 ¼ 1  ð1  FÞ3 (13)
ture on the kinetics of the reduction of hematite compacts by H2.
They found that the reduction rate of highly porous compacts Also, diffusion through the porous iron layer was more
(ε ¼ 0.54) was increased with temperature until completion, suitable for controlling the reduction rate in the range of
whereas the reduction of less porous (ε ¼ 0.35) and dense 50%<F<95%, described by

steel research int. 2019, 1900270 1900270 (5 of 16) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
l

www.advancedsciencenews.com www.steel-research.de

 
F 2 Da Costa et al.[31] carried out an experimental study to obtain
r 2p d0 f 2 ¼ κ 2 t, f 2 ¼ 0.5   0.5ð1  FÞ3 (14) kinetic parameters of iron oxides reduction. The discrepancies of
3
the rate constants and the activation energies obtained from var-
They indicated that neither chemical control nor diffusion ious experimental studies of iron ore reduction in the literature
control was sufficient to describe the complete course of reduc- motivated them to investigate these parameters. These inconsis-
tion alone. They thus proposed a combination correlation of two tencies are attributed to the large variation of experimental
aforementioned rate controlling mechanisms as follows conditions (temperature and pressure, gas flow rate, and gas
  composition) and to the initial material (mineralogical composi-
κ2 t κ2 tion, crystal and pellet size, porosity, and pore distribution of the
¼ f þf2 (15)
r 2p d0 κ1 r p 1 ore). The sample was examined by X-ray diffraction (XRD), scan-
ning electron microscopy (SEM), and Mössbauer spectrometry to
Bonalde et al.[29] conducted an experimental study on the reac- describe the morphological progress during the reduction and
tion mechanism of iron ore reduction. Their results showed that ascertain the rate-controlling mechanisms. Partially reduced
the reduction by H2 was mixed-controlled for the first 5 min and samples at 800  C were observed by SEM. The initial hematite
pore diffusion was rate-controlling afterward. The same behavior cube was made up of fairly large, dense grains (Figure 7a).
was seen when using pure CO as the reducing gas, but the After some time, the surface was covered with minor pores
change in the rate-controlling step occurred around 20 min after (Figure 7b), but the sample mostly kept its initial structure.
the start of reduction. Pores became larger when wustite appeared (Figure 7c), but
Piotrowski et al.[30] investigated the kinetics of the reduction of no significant conversion was detected in the microstructure.
hematite to wustite by experimental study. They found that the When the quantity of iron in the sample became considerable
Avrami–Erofe’ev equation of coupled nucleation and growth pro- (Figure 7d), the original grainy microstructure broke up into
cesses was applicable to describe the initial steps of the reduction smaller particles and was gradually converted to the structure
mechanism. The general equation they applied is as follows of sponge iron (Figure 7e).
Kazemi et al.[32] developed a new thermogravimetric setup to
½ lnð1  FÞ1=n ¼ kt (16) study the reduction of iron ore under a well-controlled experi-
mental condition. In conventional methods, the time required
where F is reaction extent, t is the time of reaction, and n is the to replace the inert gas with the reactant gas results in a lower
exponent associated with the nuclei shape. They calculated n reduction rate in the initial stage. Their results indicated that the
as 1.6. However, as Figure 6 shows, the selected model had a rate of reduction of iron ore is higher than the rates reported in
considerable deviation from experimental data in the final stage the literature. The microphotograph (Figure 8) showed two par-
of the reduction. Then, they found out at this stage of the reaction tially reacted regions with a clear boundary, an exterior zone,
the diffusion was rate-controlling and utilized the diffusion- which included mainly iron and an insignificant quantity of iron
controlling model for the final stage of the reduction. oxide, and an unreacted interior core. In addition, the outer zone

Figure 6. Comparison of experimental data with the two applied kinetic models: Avrami–Erofe’ev phase change model (n ¼ 1.6) and 1D diffusion model
(parabolic law) a) (T ¼ 775  C), b) (T ¼ 850  C). Reproduced with permission.[30] Copyright 2007, Elsevier B.V.

steel research int. 2019, 1900270 1900270 (6 of 16) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
l

www.advancedsciencenews.com www.steel-research.de

Figure 7. Microstructure conversion of the iron ore during reduction at 800  C by H2: a) initial; b) after 89 s; c) after 206 s; d) after 400 s; e) fully reduced;
f ) after 400 s, cross-section of polished sample. Reproduced with permission.[31] Copyright 2013, Elsevier B.V.

Figure 8. SEM microphotographs of industrial pellets, T ¼ 1123 K sample reduced to a) 50%, b) 30%, c) 50%, d) 80%. Reproduced with permission.[32]
Copyright 2014, John Wiley & Sons.

had fewer minor pores and large pores compared with the center mixture or an increase in reaction temperature increased the
of the sample when the reduction advanced. reduction reaction. The effective diffusion coefficient and the
Zuo et al.[33] conducted experimental work to study the reduc- rate constant of chemical reaction were increased with tempera-
tion kinetics of wustite pellets with H2–CO mixtures. They cal- ture or hydrogen content. The addition of a small amount of CO
culated the kinetic parameters by utilizing the unreacted in the H2 reduced the gas effective diffusion coefficient consid-
shrinking core model (USCM). They also studied the effect of erably. The reduction of FeO by a H2—CO mixture was mixed-
gas ratio at different temperatures. As expected, their results controlled; in the first stage of reaction, the chemical reaction was
showed an increase in hydrogen content in the reducing gas rate-controlling; then a combination of chemical reaction and gas

steel research int. 2019, 1900270 1900270 (7 of 16) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
l

www.advancedsciencenews.com www.steel-research.de

diffusion was rate-controlling and near the end of the reaction, understanding of the interactions of various rate steps and
only gas diffusion controlled the rate of reaction. how they are combined in the quantitative description of the
Guo et al.[17] experimentally studied the kinetics and mecha- overall rate is necessary for correctly analyzing the reduction
nisms of direct reduction of iron ore-biomass composite pellets of iron oxide pellets. This has led investigators to develop math-
with H2. XRD, SEM, and Brunauer–Emmett–Teller (BET) tests ematical models as discussed in the next section.
were adopted to explore the effect of biomass in pellet during the
reduction. They found out that the biomass not only amplified
the reduction extent of the pellets but also enhanced the reduc- 3. Mathematical Models
tion velocity index (RVI) of the pellets. The BET test showed that
the presence of biomass led to increasing the contact area A reliable mathematical model can not only deal with complex
between iron oxide and reducing gas during reduction. They systems of reduction but also reduce the cost of experimental
selected the unreacted core model to estimate activation energy study. The mathematical models that have been reported in
of the reduction. They observed that chemical reaction was the the literature for modeling the reduction reaction in the pellet
rate-controlling step because correlation factors related to chem- may be divided into three groups: one interface shrinking core
ical reactions were higher than the ones related to diffusion, as model, three interfaces shrinking core model, and homogeneous
shown in Figure 9. It is emphasized, however, that goodness of model. In the following, these models will be discussed
such match of conversion model is not sufficient for determining separately.
the controlling mechanism,[34] especially considering the uncer-
tainty and scatter in experimental data. A number of other cor- 3.1. One Interface Shrinking Core Model
roborating pieces of evidence should be established for this such
as the effect of pellet size and the time for reaction compared In this case, a simple model based on reaction at the iron–wustite
with the time under pore diffusion control with the effective dif- interface is applied for the removal of oxygen from the pellet
fusivity estimated from the pellet structure including porosity. core. Also, both intrapellet and interphase resistances to mass
Finally, they calculated the activation energy of reduction and transfer are incorporated. In Figure 10, a schematic configura-
concluded this value decreases from 122 kJ mol1 for pellets tion of the pellet reduction process is presented. It is possible
without biomass to 111 kJ mol1 for ones with biomass. to recognize the unreacted iron oxide core, the iron layer already
As it was previously stated, a large number of experimental formed, and the external gas film. Because of its simplicity, this
investigations were formerly presented with a single or multiple model has often been used in analyzing the gaseous reduction of
reacting agents. Some investigations were done for a deeper metal oxides, even for an initially porous pellet. It must, however,
understanding of the reduction phenomena and for estimating be emphatically noted that this model is applicable only to a very
the controlling step. Depending on the experimental conditions, dense solid undergoing a one-step reaction. An indiscriminate
pellet properties, and the methodology of the experiment, a application of this model can result in erroneous conclusions
wide variety of kinetic models has been reported. A clear regarding the effect of pellet size and temperature, among

Figure 9. Comparisons of reduction curve under H2 atmosphere at 1123K between chemical-reaction-controlled and diffusion-controlled models.
a) Oxide pellets with biomass; b) oxide pellets without biomass. Reproduced with permission.[17] Copyright 2015, Elsevier B.V.

steel research int. 2019, 1900270 1900270 (8 of 16) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
l

www.advancedsciencenews.com www.steel-research.de

model in which the gaseous transport is suggested as a major


controlling step. Lu[37] applied a pseudo-steady model wherein
the pore diffusion and irreversible first-order chemical reaction
in series were assumed as controlling the rate. Subsequently,
Spitzer et al.[38,39] showed the impossibility of defining a single
step having the absolute control of the overall reduction rate.
Therefore, a model that included all the aforementioned rate-
controlling steps was reported by Spitzer et al.[38] Isothermal
and isobaric conditions together with the pseudo-steady state
were the major assumptions they used for formulating this
model. Experimental data reported by McKewan[10] were used
for validation of their results. Finally, they showed that a pore-
diffusion effect was likely to exist and that both diffusion and
chemical reaction should be considered simultaneously.
The model equations applied to the gas phase, using the
pseudo-steady-state assumption and an equilibrium-limited first-
order chemical reaction rate are given below (Spitzer et al.,[39]
Kam,[40] Negri et al.,[41] and Feinman and Macrae[42])
 
kj C bp
Ṙj ¼ 4πr 2i C bR  (17)
ϕj Kej
  
ϕp
ϕj ¼ 1 þ kj ϕR þ (18)
Kej
    
1 rc r r
ϕi ¼ þ c 1  c , i ¼ P, R (19)
km,i r p De,i rp

where r i is radial position of the reaction interface, kj is the reac-


tion-rate constant, ϕj is the effectiveness factor, km,i is the exter-
nal mass-transfer coefficient, De,i is the effective diffusion
coefficient, P is the gaseous product species, and R is the gaseous
reactant species. The reaction rate term is expressed in terms of
the effectiveness factor, considering both kinetic and transport
Figure 10. Schematic configuration of dense pellet reduction used for one resistances of the reaction and diffusion processes. The model
interface shrinking core model. is completed via the solid balance equation

dr c X
dc0 ¼  Ṙj ; r c ¼ r p at t¼0 (20)
others. Based on the representation shown in Figure 9, oxygen dt j
removal from the wustite iron interface of a dense particle must
be included in the following process steps: 1) gaseous reactant where dc0 , r c , and Ṙ and t are, respectively, molar density of the
transport from the bulk gas phase to the external surface of solid, the radial position of the reaction interface, the rate of reac-
the solid; 2) diffusion of the reactant through the iron layer tion of gaseous species, and reaction time. The degree of reduc-
toward the unreacted core surface; 3) the chemical reaction tion for this model, defined as the oxygen fraction consumed
between the reducing agent and the solid oxide to form iron during the reaction process, can be calculated by the following
and the gaseous product; 4) diffusion of the latter through the  3
iron layer toward the external surface of the solid particle; and r
F ¼1 c (21)
5) gaseous product transport from the solid surface to the bulk rp
fluid phase.
Steps 1) and 5) are referred to as gas-film resistance. Similarly, Yagi et al.[43,44] presented a model with one global chemical
steps 2) and 4) are identified as shell-layer diffusion resistances reaction and a single reducing agent, under isothermal pellet
and step 3 as the reaction interface resistance. These steps offer conditions. Their simplified scheme includes the modeling of
resistance in series to the overall reaction. If one is considerably all transport resistances assuming the same values of both dif-
slower than the others, it may be identified as the “rate-controlling fusivities and transport coefficients for reactants and products.
step.” A single first-order reversible reaction with respect to the gas
McKewan[10] and Themelis and Gauvin[35] developed a simple phase is considered. The model was applied to isothermal and
mathematical model based on interfacial chemical reaction and nonisothermal fixed-bed reactors.
concluded that the iron oxide reduction is dominantly controlled Kam[40] introduced a generalization of the Spitzer model to the
by chemical reaction. Bogdandy et al.[36] applied a pseudo-steady case of a mixture of reducing agents (hydrogen and carbon

steel research int. 2019, 1900270 1900270 (9 of 16) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
l

www.advancedsciencenews.com www.steel-research.de

monoxide). Two global reduction reactions are considered pro-


ceeding independently without interaction between the gaseous
species (water-gas shift reaction is ignored). A significant contri-
bution to this kind of mathematical model of hematite reduction
was introduced when Hughes and Kam[45] include the water-
gas-shift reaction (WGSR) in the model formulation. This gas-
phase reaction is strongly catalyzed by iron. Thus, the model
considers the heterogeneous reaction front moving with the
reduced iron/oxide interface, whereas the WGSR takes place
within the solid product layer. The resulting equations for the
kinetic expression are as follows
 
C CO CH2 O C CO2
Ṙw ¼ kw  (22)
C H2 Kew

Based on a pseudo-steady-state assumption, a quasi-


homogeneous phase for the solid product layer, such as the case
of the catalytic reaction system, the material balance for gaseous
species is written as follows
 
1 d dCi
r 2
Dei ¼ υiw Ṙw
r 2 dr dr
C i ¼ C i,b , r ¼ r p (23)
dC i
Dei ¼ υij Ṙj,c , r ¼ rc
dr

The interface reaction kinetics was assumed reversible and


first-order, and external gas-film resistances were neglected. An
approximate analytical solution was obtained after linearization
of the pseudo-homogeneous reaction rate term is performed.
The pellet model was then used by the authors to simulate an
isothermal moving-bed reactor.

3.2. Three Interface-Shrinking Core Model

It has been experimentally shown that the reduction of hematite Figure 11. Schematic representation of the three interface-shrinking
to iron in the pellet simultaneously includes three separate inter- core model. Reproduced with permission.[45] Copyright 1982, American
faces not only for a porous pellet but also for a dense pellet: Chemical Society.
hematite–magnetite, magnetite–wustite, and wustite–iron.[4]
However, in the case of a dense pellet, the thickness of each layer
is narrower than in the case of a porous pellet. Hence, a compre- (magnetite–wustite interface). Here, the process is repeated with
hensive model is needed to describe this complicated situation, a portion reacting at the front and the remainder moving
as illustrated in Figure 11. through the magnetite layer toward the third reaction front
In this case, the successive reactions are modeled using a suc- (hematite–magnetite interface). The gaseous product follows
cession of moving fronts. A pseudo-steady-state assumption is the same pattern of motion but in the opposite sequence and
applied for modeling the gas phase behavior, an instantaneous direction. The velocity of each interface is determined by the
match between the flux of species (both arriving and living front) value of the specific kinetic constant for the corresponding
and chemical reaction consumption is also assumed for the inter- superficial reaction as well as by the value of the solid reactant
mediate fronts. In many cases, the gas mixture does not have concentration available at each particular front. With a first-
enough reducing power to transform wustite to iron, but it order reversible reaction at each interface, the mathematical
can still accomplish the reduction of magnetite or hematite. formulation of these dependencies is given by
Spitzer et al.[39] introduced a model following the aforemen-  
tioned description to study iron oxides reduction with hydrogen dr j kj Cp,j
¼ C R,j  , j ¼ h, m, w (24)
as the reducing agent. The overall reaction was described as a dt d0,j Kej
sequence of steps, starting with the transport of the reducing
gas through the film and iron layer, toward the first reaction where, do,h ¼ 0.1111do , do,m ¼ 0.1889do , and do,w ¼ 0.7do .
front (wustite–iron interface). A portion of the reactant reaching Spitzer et al.[40] compared this model prediction with experimen-
this boundary reacts right at this front; the remainder continu- tal data of McKewan[11] that was obtained by means of metallo-
ous through the wustite layer toward the second front graphic techniques. They concluded that predictions of the

steel research int. 2019, 1900270 1900270 (10 of 16) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
l

www.advancedsciencenews.com www.steel-research.de

three-interface model closely matched the experimental data reactants and products. They applied their model to a moving-
(Figure 12). bed reactor. Also, Negri et al.[57] assumed three interface reaction
The aforementioned mathematical model contains a number fronts including WGSR as a side reaction in the iron layer only.
of parameters and thus can fit the experimental data, but it con- The effect of WGSR on the rate of reduction in the pellet was
tains a fundamental and clear flaw of assuming distinct, sharp investigated only in the reactor modeling.
interface between different solid phases. Because the suboxides
and the product iron are porous, such sharp interfaces where 3.3. Porous Solid Model
chemical reactions take place would not, in general, be formed;
rather, the reactions would take place simultaneously with diffu- For iron oxide pellets with limited porosity, the reduction is
sion wherever there is porosity. This picture will be valid only in assumed to proceed topochemically and the analysis based on
the asymptotic case of complete rate control by pore diffusion of the three interface model has been used in the literature. It is
all the reactions. As will be discussed in the subsequent section, a again noted that this model contains a critical flaw and as a result,
realistic model addressing this issue has subsequently been its application becomes a more or less curve-fitting method that
developed by Sohn and Chaubal.[46] cannot be applied under the range of conditions outside those
Hara et al.[47] developed a similar model based on the same used in the particular experiments. In a pellet, the chemical reac-
physical depiction of the process; an identical sequence of resis- tion and gaseous diffusion proceed simultaneously, and thus the
tances is considered. The difference is the assumption of equal shrinking core model is not appropriate.
values for the transport properties of reactants and products of A number of researchers recognized the differences in the
the reducing process. This model suffers from the same flaw as reaction behavior of an initially nonporous and an initially
that of Spitzer et al.[39] This model has been applied to direct porous solid. In the latter, the chemical reaction and diffusion,
reduction in moving-bed reactors.[47–54] in general, take place simultaneously within any volume
Tsay et al.[55] developed a generalization of Spitzer’s model for elements in the pellet. In the field of iron oxide reduction,
the two reactant gas mixture (including hydrogen and carbon Tien and Turkdogan[58] formulated a mathematical model based
monoxide), based on the following assumptions: 1) chemical on this concept. They considered two separate zones of a pellet:
reactions proceed independently and the rate of reduction at the completely reduced iron layer and the incompletely reduced
any reaction interface is the sum of the rates by both reactants; iron oxide core. As will be seen subsequently, this is one special
2) all catalytic effects are neglected; 3) WGSR is also neglected; case of the grain model of gas–solid reactions developed by
and 4) an effective diffusivity for pseudo-binary diffusion is used. Szekely and coworkers,[34,59] which in turn is a special case of
They showed that the presented model had a reasonable the more general development on fluid-solid reactions formu-
agreement with experimental measurements of Szekely and lated by Sohn.[60–62]
El-Tawil.[19] Negri et al.[56] also analyzed the reduction of a porous Let us consider the grain model that encompasses the model
pellet based on the three interface models. They considered by Tien and Turkdogan[58] as a special case. This model considers
different diffusivities and mass-transfer coefficients for the that 1) a pellet is made up of grains of uniform size; 2) the initial
physical structure is maintained throughout the reaction; 3) the
reaction of each grain proceeds as a microscopic shrinking core;
and 4) gaseous diffusion through the product layer of the grains
is not considered because each individual grain is very fine and
also the product layer is porous. According to aforementioned
assumptions governing equations for nonisothermal, nonsteady
reduction of pellet based on the grain model are as follows:
Molar balance for gaseous reactants

∂ðεCj Þ
¼ ∇ ⋅ ðDeff ,j ∇C j Þ þ ð1  εÞðṘg,j Þ (25)
∂t

Molar balance for gaseous products

∂ðεCj Þ
¼ ∇ ⋅ ðDeff ,j ∇C j Þ þ ð1  εÞðṘg,j Þ (26)
∂t

Energy balance

∂T X
ðρCp Þeff ¼ ∇ ⋅ ðλeff ∇TÞ þ ð1  εÞ ðṘg,i ÞΔHi (27)
∂t i

For continuity of heat and mass fluxes on the surface of the


pellet
Figure 12. Comparison between experimental and three interface model
prediction. Reproduced with permission.[40] Copyright 1981, Institution of ∂Tðr p , tÞ
λeff ¼ hðTðr p , tÞ  T b Þ (28)
Chemical Engineers (IChemE). ∂r

steel research int. 2019, 1900270 1900270 (11 of 16) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
l

www.advancedsciencenews.com www.steel-research.de

2
∂C j ðr p , tÞ F p ð1  XÞFp  2ð1  X Þ
Deff ,j ¼ km,j ðC j ðr p , tÞ  yb,j Þ (29) pFp ðX Þ ≡ 1 
∂r Fp  2
8 2
Spherical symmetry at the center of the pellet >
> X for Fp ¼ 1 (40)
<
¼ X þ ð1  X ÞLnð1  X Þ for Fp ¼ 2
∂Tð0,tÞ >
>
λeff ¼0 (30) : 2
∂r 1  3ð1  X Þ3 þ 2ð1  X Þ for Fp ¼ 3

∂Ci ð0, tÞ The model by Tien and Turkdogan[58] for iron oxide reduction
Deff ,i ¼0 (31)
∂r corresponds to the case of Fg ¼ 1. The single reaction assumption
was justified by the authors given the fact that the rate-controlling
where the local reduction rate is calculated as follows for the case step was the reduction of wustite. The model was applied to the
of spherical grains analysis of fractional reduction for different temperatures and
  different pellet sizes. The result of the model was found to show
3kr,j 2 CP,j
Ṙg,j ¼ ð1  f Þ3 C R,j  (32) good agreement with experimental data, as indicated by the
rg Kej
example shown in Figure 13.
∂f X3kr,j  
CP,j
 Also, an unsteady and quasi-steady isothermal solution of
¼
2
ð1  f Þ C R,j 
3 (33) aforementioned equations was used by Usui et al.[63] for investi-
∂t j
ρ0 r g Kej gation of wustite pellet reduction with pure hydrogen. The
numerical results were compared with experimental data and
The overall fractional reduction of a pellet containing spherical there was a good agreement between model predictions and
grains is estimated by experimental results. They also reported that a quasi-steady
R rp numerical solution yielded results that are close to those of
4πr 2 f dr the unsteady numerical solution. This is expected in general for
X¼ 0
(34)
3 πR0
4 3 gas–solid reactions, as described frequently (see for example[34]).
As can be seen in Figure 14, a typical result by Usui et al.[63]
Here, f represents the local fractional reduction of grains that indicates that the assumption of constant effective diffusivity
is defined as f ¼ 1  ðr j =r g Þ3 . and porosity yielded satisfactory prediction.
Pseudo-steady (ignoring accumulation term in Equation (25) Valipour et al.[64] investigated an unsteady isothermal reduc-
and (26)) and isothermal conditions (ignoring Equation (27), (28), tion of hematite pellet by a mixture of hydrogen, water vapor,
and (30)) are frequently applicable in gas–solid reactions. carbon monoxide, and carbon dioxide. In this model, it was
Sohn and Szekely[59] subsequently generalized the aforemen- assumed that each grain is reduced topochemically and the rate
tioned equations to different shapes of the grains and also is controlled by the chemical reaction at the wustite–iron inter-
obtained the following approximate solution for the conversion face. They validated their model with experimental data from
versus time relationship for an isothermal system and all grain
and pellet shapes

t ≅ gFg ðX Þ þ σ̂ 2 ½P Fp ðX Þ þ 4X=Sh  (35)

in which Fg and Fp are, respectively, the shape factors for the


grains and pellet with the values of 1, 2, and 3, respectively,
for flat, cylindrical, and spherical shapes, and
   
bk Ag C
t ≡ C A0  CO t (36)
ρB Fg V g K
  2  
ð1  εÞF p k Ag Vp 1
σ̂ 2 ≡ 1þ (37)
2De FgV g Ap K

Sh ≡ ðkm =De Þð2F p V p =Ap Þ ¼ ShðDM =De Þ (38)

in which V and A are the initial volume and surface area of a


grain or the pellet.

g Fp ðX Þ ¼ 1  ð1  X Þ1=Fp (39)

which is the conversion function when pore diffusion resistance Figure 13. Comparison between experimental results and the porous
is negligible and the gas phase concentration is uniform in the solid model for different temperatures. Reproduced with permission.[58]
entire pellet at the bulk value. Copyright 1972, Springer Nature.

steel research int. 2019, 1900270 1900270 (12 of 16) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
l

www.advancedsciencenews.com www.steel-research.de

Figure 14. Effect of constant porosity, constant diffusivity, and quasi-steady Figure 15. Comparison of the model predictions and the experimental
assumption on the rate of reduction. Reproduced with permission.[63] data for the reduction of hematite to iron in a mixture of hydrogen and
Copyright 1990, The Iron and Steel Institute of Japan. carbon monoxide. Reproduced with permission.[65] Copyright 2007,
Springer Nature.

Towhidi and Szekely.[20] The effects of bulk temperature, gas


distribution within the pellet under their experimental condi-
concentration, gas utility, grain size, and porosity on the rate
tions. However, it influenced the concentrations of the gaseous
of reduction were investigated.
species noticeably.
A mathematical model was presented by Valipour and
Ghadi et al.[69] numerically evaluated the capability of the grain
Saboohi[65] to simulate the multiple heterogeneous reactions
model in comparison with the USCM for the reduction of wustite
with complicated physicochemical and thermal phenomena in
pellet to sponge iron. Their results indicated that unlike the grain
a single pellet and during its fall down a furnace. This model
model, USCM cannot appropriately predict the effect of gas mix-
included both heat and mass transfer phenomena of gaseous
ture parameters including temperature and pellet characteristics
species in a porous medium in addition to chemical reactions.
such as grain diameter, porosity, and tortuosity on the local
Iron oxide reduction was selected to validate the model and,
fractional reduction degree. They illustrated the wustite pellet
as Figure 15 shows, an excellent agreement was seen between
reduction pattern at different reduction degrees (Figure 18).
numerical and experimental studies.
According to their results attained by the grain model, changes
Valipour[66] also performed mathematical modeling of the
in the grain diameter, temperature, and tortuosity complicate the
unsteady reduction of hematite pellets with syngas for isothermal
transport of gas species within the pellet. An increase in temper-
and nonisothermal conditions. In this study, he investigated the
ature leads to the shrinkage of reaction zone. However, an
reduction degree and temperature distribution in the pellet in
increase in porosity makes the transport of gas species easier,
detail. He also studied gas composition variation and tempera-
making a wider reaction zone.
ture variation at three different positions in the pellet for both
isothermal and nonisothermal conditions. Figure 16 and 17
show the results of the study for temperature variations and 4. Conclusions
gas composition within the pellet versus time.
Valipour and Khoshandam[67] investigated the reduction of Since the development of the direct reduction technology of iron
wustite pellets by means of a nonisothermal transient model. ore as a promising and clean production method for ironmaking,
Gas species contained a mixture of hydrogen, carbon monoxide, many investigations were conducted to examine the kinetics and
water vapor, and carbon dioxide. The impacts of pellet porosity, mechanism of the reduction reactions over the last three decades.
diameter, gas composition, and temperature were evaluated. Therefore, the authors were motivated to carry out a review on
The effects of nonisothermality on the overall reduction were experimental and numerical studies on the direct reduction of
also investigated. iron oxides. By experimental studies, mineralogical composition,
The effect of water-gas-shift reaction on wustite reduction was crystal size, porosity, pore-size distribution, and specific surface
studied by Valipour and Mokhtari.[68] Their mathematical model area of the ore have been determined and the morphological
was based on a transient, nonisothermal reduction based on the progress of reduction examined. Through these efforts, research-
grain model. They found that the impact of WGSR was not sig- ers have defined the rate-controlling steps of the reaction during
nificant on the rate of wustite reduction and temperature the reduction process, which had often been a challenging issue

steel research int. 2019, 1900270 1900270 (13 of 16) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
l

www.advancedsciencenews.com www.steel-research.de

Figure 16. The temperature variation in three positions within the pellet versus time. (left) isothermal reduction (right) nonisothermal reduction.
Reproduced with permission.[66] Copyright 2009, Sharif University of Technology.

Figure 17. Gaseous reactant and product variation in three different positions within the pellet versus time. Reproduced with permission.[66] Copyright
2009, Sharif University of Technology.

in the literature. Also, activation energy and frequency factor of There are some opportunities to develop experimental and
the reactions were estimated by utilizing mathematical models. numerical studies about the direct reduction of iron ores in
Formulating various mathematical models and fitting them to the microscopic view as a future direction. As energy consump-
the experimental data has led to two results: 1) determining tion is a key factor in ironmaking, using iron ores with higher
the best conceptual model describing the physics of reduction reaction rates is extremely preferable. Therefore, searching for or
and 2) defining the rate-controlling regime during reduction. preparing new raw materials that create higher conversion rates
The discrepancies of activation energy of reductions in the liter- could be an interesting idea. For example, utilizing a composite of
ature can be attributed to the wide varieties of experimental iron ore and biomass[17] indicated an improvement in reduction
conditions and to raw material with different properties, but rate. A further example in this regard is the Flash Ironmaking
sometimes to the indiscriminate application of an incorrect Technology based on gaseous reduction of fine iron oxide
model. concentrate recently developed by Sohn and coworkers.[70–72]

steel research int. 2019, 1900270 1900270 (14 of 16) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
l

www.advancedsciencenews.com www.steel-research.de

Figure 18. Wustite pellet reduction pattern at various reduction degrees. Reproduced with permission.[69] Copyright 2016, Taylor & Francis.

For improving numerical approaches and mathematical model- [1] Midrex Tecnology Inc. World Direct Reduct Stat 2017.
ing, the authors have the following suggestions: 1) Almost all [2] T. L. Joseph, Trans. AIME 1936, 120, 72.
previous authors in the literature neglect convection mechanism [3] J. O. Edstrom, J. Iron Steel Inst. 1953, 175, 289.
due to gas flow. This can be done by applying the Darcy equation [4] W. K. Lu Direct Reduc. Iron Technol. Econ. Prod. Use. Iron Steel Soc.
within the pellet and relating pressure gradient to the velocity of 410 Common, Iron and Steel Society/AIME, 1999, pp. 43–57.
gases, as was proposed by Sohn and Chaubal.[46,73] 2) When [5] M. S. Valipour, Y. Saboohi, Model Simul. Mater. Sci. Eng. 2007, 15, 487.
[6] S. Wu, J. Xu, S. Yang, Q. Zhou, L. Zhang, ISIJ Int. 2010, 50, 1032.
operating conditions cause a significant temperature and/or con-
[7] J. Xu, S. Wu, M. Kou, K. Du, ISIJ Int. 2013, 53, 576.
centration variation across the pellet dimension, the reduction
[8] A. Z. Ghadi, M. S. Valipour, M. Biglari, Int. J. Hydr. Energy 2017, 42, 103.
behavior in 2D space may be needed to examine the effects of [9] W. M. McKewan, Trans. Am. Inst. Min. Metall. Eng. 1960, 218, 2.
transport phenomena more accurately.[74] 3) The review has [10] W. M. McKewan, Trans. Metall. Soc. AIME 1961, 221, 140.
shown that most conceptual models for the direct reduction of [11] W. M. McKewan, Trans. Met. Soc. AIME 1962, 224, 2.
iron ores are based on unreacted shrinking core and grain mod- [12] N. A. Warner, Trans. Met. Soc. AIME 1964, 230, 1631.
els. However, there are new conceptual models showing excellent [13] R. G. Olsson, W. M. McKewan, AIME Met. Soc. Trans. 1966, 236, 1518.
agreement with experimental data in the reactions of other mate- [14] E. T. Turkdogan, J. V. Vinters, Metall. Mater. Trans. B 1971, 2, 3175.
rials. These models could be used to better simulate the iron-ore [15] E. T. Turkdogan, J. V. Vinters, Metall. Trans. 1972, 3, 1561.
reduction process by incorporating the microscopic view.[75–77] [16] E. T. Turkdogan, R. G. Olsson, J. V. Vinters, Metall. Mater. Trans. B
1971, 2, 3189.
[17] D. Guo, M. Hu, C. Pu, B. Xiao, Z. Hu, S. Liu, X. Wang, X. Zhu, Int. J.
Conflict of Interest Hydr. Energy 2015, 40, 4733.
[18] S. P. Trushenski, K Li, W. O. Philbrook, Met. Trans. 1974, 5, 1149.
The authors declare no conflict of interest. [19] J. Szekely, Y. El-Tawil, Metall. Trans. B 1976, 7, 490.
[20] N. Towhidi, J. Szekely, Ironmak. Steelmak. 1981, 8, 237.
[21] M. M. Al-Kahtany, Y. K. Rao, Ironmak. Steelmak. 1980, 7, 49.
[22] Y. K. Rao, M. Moinpour, Metall Trans B 1983, 14, 711.
Keywords [23] Y. K. Rao, M. M. Alkahtany, Ironmak. Steelmak. 1984, 11, 34.
diffusion-controlled, direct reduced iron, flash ironmaking technology, [24] N. Towhidi, J. Szekely, Metall. Trans. B 1983, 14, 359.
grain models, pellet scale modeling, rate-controlling steps, sponge iron [25] A. A. El-Geassy, V. Rajakumar, Trans. Iron Steel Inst. Jpn. 2011, 25, 449.
[26] A. A. El-Geassy, M. I. Nasr, Trans. Iron Steel Inst. Jpn. 1988, 28, 650.
Received: June 9, 2019 [27] S. Paul, S. Mukherjee, Ironmak. Steelmak. 1992, 19, 190.
Revised: September 13, 2019 [28] M. J. Proctor, J. D. Smith, R. J. Hawkins, Ironmak. Steelmak. 1992, 19, 194.
Published online: [29] A. Bonalde, A. Henriquez, M. Manrique, ISIJ Int. 2005, 45, 1255.
[30] K. Piotrowski, K. Mondal, T. Wiltowski, P. Dydo, G. Rizeg, Chem. Eng.
J. 2007, 131, 73.

steel research int. 2019, 1900270 1900270 (15 of 16) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
l

www.advancedsciencenews.com www.steel-research.de

[31] R. A. Da Costa, D. Wagner, F. Patisson, J Clean Prod 2013, 46, 27. [54] E. D. Negri, O. M. Alfano, M. G. Chiovetta, Ind. Eng. Chem. Res. 1995,
[32] M. Kazemi, B. Glaser, D. Sichen, Steel Res. Int. 2014, 85, 718. 34, 4266.
[33] Z. H bin, C. Wang, D. Ji, J. K. Xin, X. R. Sheng, Int. J. Miner. Metall. [55] Q. T. Tsay, W. H. Ray, J. Szekely, AIChE J. 1976, 22, 1064.
Mater. 2015, 22, 688. [56] E. D. Negri, O. M. Alfano, M. G. Chiovetta, Lat. Am. Appl. Res. 1988,
[34] J. Szekely, J. W. Evans, H. Y. Sohn, Gas-Solid Reactions, Academic 18, 93.
Press, New York 1976. [57] E. D. Negri, O. M. Alfano, M. G. Chiovetta, Ind. Eng. Chem. Res. 1991,
[35] W. H. Themelis, N. J. Gauvin, Trans. Metall. Soc. AIME 1963, 227, 290. 30, 474.
[36] L. von Bogdandy, H.-J. Engell, The Reduction of Iron Ores: Scientific [58] R. H. Tien, E. T. Turkdogan, Metall. Trans. 1972, 3, 2039.
Basis and Technology, Springer, Berlin, Heidelberg 1971. [59] H. Y. Sohn, J. Szekely, Chem. Eng. Sci. 1972, 27, 763.
[37] W.-K. Lu, Trans. Met. Soc. AIME 1963, 227, 203. [60] H. Y. Sohn, Metall. Trans. B 1978, 9, 89.
[38] R. H. Spitzer, F. S. Manning, W. O. Philbrook, AIME Met. Soc. Trans. [61] H. Y. Sohn, Metall. Trans. B 1991, 22, 737.
1966, 236, 726. [62] H. Y. Sohn, Reaction Engineering Models. Treatise on Process
[39] R. H. Spitzer, F. S. Manning, W. O. Philbrook, Trans. Metall. Soc. Metallurgy, Elsevier, Oxford, UK; Waltham, MA, USA 2014,
AIME 1966, 236, 1715. pp. 758–810.
[40] E. K. T. Kam, Trans. IChemE. 1981, 59, 196. [63] T. Usui, M. Ohmi, E. Yamamura, ISIJ Int. 1990, 30, 347.
[41] E. D. Negri, O. M. Alfano, M. G. Chiovetta, Lat. Am. J. Heat Mass. [64] M. S. Valipour, M. Y. Motamed Hashemi, Y. Saboohi, Adv. Powder
Transf. 1985, 9, 85. Technol. 2006, 17, 277.
[42] J. Feinman, D. R. Mac Rae, Direct Reduced Iron: Technology and [65] M. S. Valipour, Y. Saboohi, Heat Mass. Transf. 2007, 43, 881.
Economics of Production and Use, Iron and Steel Society, [66] M. S. Valipour, Sci. Iran Trans. C Chem. Chem. Eng. 2009, 16, 108.
Warrendale, PA 1999. [67] M. S. Valipour, B. Khoshandam, Ironmak. Steelmak. 2008, 36, 91.
[43] J. Yagi, R. Takahashi, Y. Omori, Tetsu-to-Hagane 1971, 57, 1453. [68] M. S. Valipour, M. H. Mokhtari, Int. J. ISSI 2011, 8, 9.
[44] J. Yagi, A. Moriyama, I. Muchi, J. Jpn. Inst. Met. Mater. 1968, 32, 209. [69] A. Z. Ghadi, M. S. Valipour, M. Biglari, Ironmak. Steelmak. 2016, 43, 418.
[45] R. Hughes, E. K. T. Kam, in Chemical Reaction Engineering—Boston, [70] H. Y. Sohn, Steel Times Int. 2007, 31, 68.
ACS Publications Washington, DC 1982, pp. 29–38. [71] H. Y. Sohn, Y. Mohassab, J. Sustain. Metall. 2016, 2, 216.
[46] H. Y. Sohn, P. C. Chaubal, Trans. Iron Steel Inst. Jpn. 1984, 24, 387. [72] A. Abdelghany, D.-Q. Fan, M. Elzohiery, H. Y. Sohn, in AISTech 2018
[47] Y. Hara, M. Sakawa, S. Kondo, Tetsu-to-Hagané 1976, 62, 315. Proceedings, AIST, Tokyo, Japan 2018, pp. 683–906.
[48] Y. Hara, M. Sakawa, S. Kondo, Tetsu-to-Hagane 2017, 62, 324. [73] H. Y. Sohn, M. B. Aboukheshem, Metall. Trans. B 1992, 23, 285.
[49] J.-I. Yagi, J. Szekely, Trans. Iron Steel Inst. Jpn. 1977, 17, 569. [74] J. Shi, E. Donskoi, D. L. S. McElwain, L. J. Wibberley, Math. Comput.
[50] J.-I. Yagi, J. Szekely, Trans. Iron Steel Inst. Jpn. 1977, 17, 576. Model. 2005, 42, 45.
[51] J. Yagi, S. J. Ichiro, AIChE J. 1979, 25, 800. [75] M. Elzohiery, H. Y. Sohn, Y. Mohassab, Steel Res. Int. 2017, 88, 290.
[52] T. Yanagiya, J. Yagi, Y. Omori, Ironmak. Steelmak. 1979, 6, 93. [76] F. Chen, Y. Mohassab, T. Jiang, H. Y. Sohn, Metall. Mater. Trans. B
[53] Y. Takenaka, Y. Kimura, K. Narita, D. Kaneko, Comput. Chem. Eng. Process. Metall. Mater. Process. Sci. 2015, 46, 1133.
1986, 10, 67. [77] M. E. Choi, H. Y. Sohn, Ironmak. Steelmak. 2009, 37, 81.

steel research int. 2019, 1900270 1900270 (16 of 16) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

You might also like