Minerals Engineering 63 (2014) 45–56

Contents lists available at ScienceDirect

Minerals Engineering

journal homepage: www.elsevier .com/locate /mineng

Numerical study of hot charge operation in ironmaking blast furnace

0892-6875/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.mineng.2013.11.002

⇑ Corresponding author. Tel.: +61 2 9385443; fax: +61 2 93856565.E-mail address: s.kuang@unsw.edu.au (S.B. Kuang).

S.B. Kuang a,⇑, Z.Y. Li a, D.L. Yan b, Y.H. Qi b, A.B. Yu a

a Laboratory for Simulation and Modelling of Particulate Systems, School of Materials Science and Engineering, The University of New South Wales, Sydney, NSW 2052, Australiab State Key Laboratory for Advanced Iron and Steel Processes and Products, Central Iron and Steel Research Institute, Beijing 100081, China

a r t i c l e i n f o

Article history:Available online 28 November 2013

Keywords:Extractive metallurgyMathematical modellingBlast furnaceMultiphase flow

a b s t r a c t

Charge of hot coke and iron-bearing materials into an ironmaking blast furnace (BF) may bring significantenergy and environmental benefits to the BF process. However, there is little information about the quan-titative effects of hot charge operation on BF flow and performance. This paper presents a numericalstudy of multiphase flow, heat and mass transfer in a BF by a process model. The applicability of themodel in predicting BF performance is first confirmed by different applications. It is then used to studythe effects of hot charge operation at different temperatures. The results are analyzed in detail withrespect to BF flow and performance. It is shown that compared to the conventional operation, hot chargeoperation can lead to an increased productivity, decreased coke rate and CO2 emission, and at the sametime, increased gas pressure and top gas temperature. These effects vary with hot charge temperature.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Blast furnace (BF) ironmaking is the most important technologyby which iron is rapidly and efficiently reduced from iron-bearingmaterials (Biswas, 1981). Its primary energy source and reducingagent are mainly coal in form of coke and pulverized coal, whichis finally released as CO2 to the environment. Also, BF ironmakingsystem consumes 70% of the energy input in an integrated steel-making works. Therefore, BF, as the core of the system, is usuallyfeatured with extensive energy consumption and massive green-house gas emission. Furthermore, such a reactor demands a signif-icant amount of coke to maintain adequate furnace permeabilityand provide thermal and chemical energy sources. The consumedcoke, as a kind of noble material, shares a large portion (�25%) ofthe production cost of hot metal (HM). As such, coke rate (cokeconsumption per tonne of hot metal, also referred to as CR for con-venience) is critical to BF performance with regard to energy effi-ciency, CO2 emission and production cost.

In recent years, various technologies have been developed toimprove BF performance. These include, for example, top gas recy-cling (Austin et al., 1998; Nogami et al., 2006; Chu and Yagi, 2010;Helle et al., 2010), injection of pulverized coal, hydrogen bearingmaterials, natural gas, and coke oven gas (Slaby et al., 2006; Liet al., 2007; Shen et al., 2009), charge of novel burden materialssuch as scrap and carbon composite agglomerate (Nogami et al.,2006; Kawanari et al., 2011), and hot charge (Biswas, 1981).Although being examined at various levels, many of these technol-

ogies are still on trial, with the long-term practical feasibility lar-gely remaining unknown. This is especially true for hot chargeoperation, where coke and iron-bearing materials, usually referredto as burden, are alternatively charged into a BF at a higher tem-perature than the ambient temperature as used in a conventionaloperation. This high temperature may be achieved through the fol-lowing two ways. One directly charges the hot stock materialsfrom the upstream of the BF, which avoids the massive energy lossrelated to cooling process. Another makes use of the unutilizedsensible heat and chemical energy of materials within the inte-grated steelmaking works to preheat the burden materials to a cer-tain temperature before charging. With the help of the extra heatinput from the furnace top, hot charge operation may have greatpotential in improving BF performance. However, to date, ourknowledge about the effect of such a technology on BF flow andperformance is little, especially at a quantitative level. This prob-lem is further complicated by the fact that conveying and chargingsystems required by hot charge operation to withstand the hightemperature environment at the furnace top have not been fullyestablished yet.

On the other hand, in order to secure a successful running ofnew operations, it is a necessary prerequisite to predict and under-stand BF flow and performance over a wide range of conditions.This is difficult to achieve experimentally, because ironmaking BFis a very complex multiphase reactor accompanying with hightemperature and hazardous conditions. In principle, this problemcan be overcome by numerical simulation. In this direction, variousmathematical models have been developed in the past decades todescribe localized or global particulate and multiphase flow behav-iors in BFs (see, for example, the recent review by Dong et al.

Nomenclature

aFeo activity of molten wustiteAc effective surface area of coke for reaction, m2

cp specific heat, J kg�1 K�1

d diameter of solid particle, mDe

g;CO effective diffusivity of carbon monoxide, m2 s�1

Ef effectiveness factors of solution loss reactionfo fraction conversion of iron oreF interaction force per unit volume, kg m�2 s�2

g gravitational acceleration, m s�2

hij heat transfer coefficient between i and j phases,W m�2 K�1

H enthalpy, J kg�1

DH reaction heat, J kg�1

k thermal conductivity, W m�1 K�1

kf gas-film mass transfer coefficient, m s�1

ki reaction constant of ith chemical reaction, m s�1

K1 equilibrium constant of indirect reduction of iron ore byCO

Mi molar mass of ith species in gas phase, kg mol�1

Msm molar mass of FeO or flux in solid phase, kg mol�1

p pressure, Papct percentagepr prandtl number, cp lK�1

R� reaction rate, mol m�3 s�1

S source termShr shrinkage ratio defined as the ratio of the decreased vol-

ume, caused by softening and melting, to the originalvolume occupied by iron-bearing material

Sh�r normalized shrinkage ratio, Sh�r ¼ Shr=Shr;max,Shr,max = 0.7

T temperature, Ku interstitial velocity, m s�1

Vb bed volume, m3

Vg gas volume, m3

Volcell volume of control volume, m3

yi mole fraction of ith species in gas phasey�i mole fraction of ith species in equilibrium state

Greek lettersC diffusion coefficientI identity tensor/ general variableU shape factora specific surface area, m�2 m�3

b mass increase coefficient of fluid phase associated withreactions, kg mol�1

d distribution coefficiente volume fractiong fractional acquisition of reaction heatl viscosity, kg m�1 s�1

q density, kg m�3

s stress tensor, Pax mass fractionnore, ncoke local ore, coke volume fraction

Subscriptse effectiveg gasi identifier (g or s)i, m mth species in i phasej identifier (g or s)k kth reactions solidsm FeO or flux in solid phase

Superscriptse effectiveg gass solid

46 S.B. Kuang et al. / Minerals Engineering 63 (2014) 45–56

(2007)). Generally speaking, the existing approaches can be classi-fied into two categories: continuum approach at a macroscopic le-vel and discrete approach at a microscopic level. The former issuitable for process modelling and applied research because ofits computational convenience and efficiency. Indeed, most of theBF modelling is based on continuum approach (Dong et al.,2007). However, to date, comprehensive numerical studies of hotcharge operation in BF have not been found in the literature.

This paper presents a numerical study of BF flow and perfor-mance at different hot charge temperatures by a continuum-basedprocess model. It is organized as follows. First, the numerical mod-el is introduced. The applicability of the model is examined by dif-ferent applications. On this base, the effects of hot charge onprocess performance are quantified, followed by a detailed studyof the flow and heat and mass transfer for better understanding.The findings from this study should be useful not only for estab-lishing a full picture about the hot charge operation but also fordeveloping some guides for possible implementation of this tech-nology in practice.

2. Model description

The current BF process model is a two-dimensional (2-D) math-ematical model which considers mass, momentum and enthalpyconservations for gas and solid phases at steady state. It is in prin-ciple the same as that recently reported by Dong et al. (2010). For

brevity, we only describe the key features of this model below,with the new developments emphasized.

2.1. Governing equations

Table 1 summarizes the governing equations for fluid flow aswell as heat and mass transfer considered in this study. Gas is de-scribed by the well-established volume-averaged, multiphase, Na-vier–Stokes equations (Dong et al., 2007). Solids are assumed to bea continuous phase that can be modelled based on the typical vis-cous model used in multiphase flow modelling (Austin et al., 1997),coupled with the method proposed by Zhang et al. (1998) fordetermination of the deadman boundary. General convection–dif-fusion equations are applied to describe heat and mass transferamong the phases.

2.2. Momentum, heat and mass transfer between phases

The gas–solid interaction as gas flows through a packed bed isdescribed by the Ergun’s expression (1953):

Fsg ¼ �Fg

s ¼ �ðaf qg jusg j þ bf Þus

g : ð1Þ

The interphase mass transfer, which occurs due to reactions andphase changes, is evaluated from simple mass balances. Accord-ingly, three crucial chemical reactions and two important phase

Table 1Governing equations.

Equations Description

Mass conservation r � ðeiqiuiÞ ¼ Si; where Si ¼ �P

kbi;kR�kr � ðegqgugugÞ ¼ r � sg � egrpþ qgeg g þ Fs

g

Momentumconservation

sg ¼ eglg ½rug þ ðrugÞT � � 23 eglgðr � ugÞI

Gasr � (esqsusus) =r � ss - esrps + qsesg

ss ¼ esls½rus þ ðrusÞT � � 23 eslsðr � usÞI

Solidr � ðeiqiuiui;mÞ �r � ðeiCirui;mÞ ¼ Sui;m

If /i,m is Hi,m, Ci ¼ kicp;i

Heat and speciesconservation

Sui;m¼ dihijaðTi � TjÞ þ gi

PkR�kð�DHkÞ

If /i,m is xi,m, Ci = qiDi, Sui;m¼P

kai;m;kR�kwhereui;m ¼ xg;CO;xg;CO2 ;xs;Fe2 O3 ;xs;Fe3O4 ;xs;FeO;xs;flux

Phase volume fractionP

iei = 1State equation p =

Pi(yiMi)RTg/Vg

Table 2Chemical reactions.

Reaction formula Reaction rate Refs.

Fe2O3(s) + CO(g) = Fe(s) + CO2(g) R�1 ¼12noreeore PðyCO�y�COÞ=ð8:314TsÞ

d2ore=De

g;CO ½ð1�f0Þ�1

3�1�þdorefk1ð1þ1=K1Þg�1Muchi(1967)

FeO(l) + C(s) = Fe(l) + CO(g) R�2 ¼ k2AcVb

aFeO,AcVb¼ 0:468½eucokedcoke�

Omori(1987)

C(s) + CO2(g) = 2CO(g) R�3 ¼6ncokeecoke pyco2

=ð8:314TsÞdcoke=kfþ6=ðqcoke Ef k3Þ

Omori(1987)

FeO(s) ? FeO(l) R�4 ¼ hTi�Tmin;sm

Tmax;sm�Tmin;smi10H

xsmuiqiei dAMsmVolcell

Austinet al.(1997)

Flux(s) ? Slag(l)

S.B. Kuang et al. / Minerals Engineering 63 (2014) 45–56 47

changes are considered, including indirect reduction of iron, directreduction of wustite, solution loss reaction, and melting of Fe andFeO, as listed in Table 2. Because the hydrogen proportion in thecurrent simulation system is small, hydrogen reduction and gas–water reactions are not considered.

The gas–solid convective heat transfer coefficient hgs is calcu-lated using the Ranz–Marshall equation:

hgs ¼ ckg

dsð2:0þ 0:6Re0:5

g Pr0:333Þ: ð2Þ

As the permeability of the softening phase is lost within thecohesive zone, heat transfer between gas and particles herechanges to between gas and slab. Thus, the corresponding heattransfer coefficient is expressed as follows (Maldonado, 2003):

hg�slab ¼kg

dsð0:203Re0:33

g Pr0:33 þ 0:22Re0:8g Pr0:4Þ: ð3Þ

Heat loss through BF wall is characterised by Newton’s law ofcooling, in which the temperature gradient normal to furnace wallis employed to quantify the heat transfer amount. Considering thecurrent BF refractory materials and wall thickness, the heat con-ductivity coefficient of BF wall is set to 5 W m�1 K�1.

2.3. Modelling of cohesive zone (CZ)

The softening and melting zone within a BF, i.e., the so-calledCZ, contributes significantly to the process complexity. This zoneis of critical importance for efficient operation of the BF becauseits shape and position determine the permeability, fluid flow, gasusage, thermal and chemical efficiency, and hot-metal quality inthe furnace. Within this region, iron-bearing materials aregradually transformed from the lumpy, to the softening, to the

half-molten states before finally melting down. Instead of passingthrough the low permeability portion of this region, the reducinggas can flow radially through adjacent coke layers, which form alow resistance path between dripping and lumpy zones. This find-ing implies a possible retardation of heating and reduction rates forore in the CZ, with direct gas contact only occurring at the ore-cokeinterface. Therefore, careful consideration must be given to theinternal structure of the CZ.

Different numerical CZ treatments have been proposed, such asisotropic and anisotropic nonlayered treatments and layered treat-ment. In this study, the layered treatment recently proposed byDong et al. (2010) is adopted. In such a model, the CZ is treatedas layered structure and subdivided into different states accordingto the BF dissection studies reported in the literature. To achievethis, the stratified structure of coke and ore layers is first calculatedbased on the solid flow field. Then, the CZ is defined to start andfinish within the temperature range of 1473–1673 K. Finally, theCZ region is divided into alternate coke and softening and meltingore layers.

For simplicity, the following three states are specified for theore layer: (1) state I, 0.7 < Sh�r < 1.0 corresponds to the portion withmolten state and liquid source in which the ore layer voidage isoccupied fully by the liquid phase; (2) state II, 0.5 < Sh�r < 0.7 corre-sponds to the combined portion with softening and melting of oresin which the pressure drop may have increased significantly; and(3) state III, 0 < Sh�r < 0.5 corresponds to the softening stage inwhich macropores of the bed remain open so the variation of thepressure drop is limited. In the CZ, the solid conduction and gas–solid heat transfer coefficient are specified according to the differ-ent heat and mass transfer mechanisms in each of the states, asdiscussed by Dong et al. (2010). Note that in this study, a similartreatment is extended to the lumpy zone (the region above theCZ) consisting of ore and coke layers.

2.4. Modelling of variation of stockline

The coke charged into a BF is consumed mainly through the fol-lowing ways: chemical reactions such as direct reduction of wus-tite and solution loss reaction, combustion in the raceway infront of tuyeres, and carburization, as illustrated in Fig. 1a. Thecoke consumption due to the combustion is constant because ofthe constant oxygen in the blown-in blast in a normal operation.The coke consumption due to the carburization is small and canbe largely regarded as constant. But the coke consumption bychemical reactions varies with in-furnace state. To adapt to thisvariation, the solid charge rate needs to be changed by varyingthe stockline in BF practice to ensure that the furnace is not over-loaded. To date, modelling of varying stockline has not been dis-cussed in the literature and is not clear. In this study, thisproblem is overcome by introducing a simple numerical treatmentaccording to the nature of BF operation, as discussed in thefollowing.

Generally, in a BF process model, it is difficult to directly modelthe raceway. Alternatively, the condition of reducing gas generatedin the raceway is approximated using the heat and mass balanceand treated as the input of gas phase, with the outlet for solids(coke) introduced. The outlet boundary essentially represents thesimplified boundary of the raceway. Thus, the solid flowrate atthe outlet is the rest of the charged coke after chemical reactionsand carburization, and should correspond to the tuyere cokeamount (the amount of coke combusted in the raceway per tonneof hot metal). If not, the coke balance is not satisfied. In this case,we can adjust the productivity and thus the solid charge rate atthe furnace top to control the coke input. This is to some degreesimilar to the variation of stockline in practice. Here, the adjustedamount of productivity is determined by definition according to

(b)(a)Fig. 1. Schematic illustrations of: (a) coke balance in a BF, and (b) solution procedure of the present BF process model.

48 S.B. Kuang et al. / Minerals Engineering 63 (2014) 45–56

the difference between the tuyere coke amounts obtained respec-tively according to the blast conditions and solid flowrate. This en-sures that the proposed method can quickly obtain the expectedproductivity to satisfy the coke balance. In this study, the numberof adjustments is less than four for all the cases simulated.

3. Numerical solution

The well-established sequential solution procedure is employedto calculate the coupled fluid flow, heat and mass transfer throughthe following steps as shown in Fig. 1b:

(1) The input conditions in simulation are determined accordingto the global heat and mass balance under given material,operational and geometrical conditions.

(2) The initial gas and solid flow, temperature, as well as con-centration fields are determined.

(3) The layered structure is determined based on the burdendistribution at the furnace top (e.g., the ore and coke batchweights as well as the radial distribution of ore–coke ratio)and the timelines of solid flow.

(4) The flow, temperature, and concentration fields are calcu-lated without considering the chemical reactions until anapproximate convergence (i.e., high tolerance) in terms offlow, mass and energy residuals is obtained.

(5) Thermochemical behaviors are taken into account, whichleads to the CZ determination, and with this information,fluid flow, heat and mass transfer, as well as chemical reac-tions are recalculated until the CZ position is converged ordoes not change much.

(6) The coke balance is examined to adjust the productivity. Ifthe difference between the tuyere coke amounts calculatedrespectively according to the blast conditions and solid flow-rate is less than 0.1 kg/tHM, the simulation stops. If not, goto Step 1 and repeat the whole procedure.

In the above proposed procedure, because the fluid flow andheat and mass transfer are solved in sequence, their numericalsolutions are in principle similar to those used for a single-phaseflow, whose convergence conditions have been well discussedand documented in the literature (Ferziger and Peric, 2002). How-ever, a general convergence criterion as used in CFD cannot ensurethat the present model generates physically meaningful results. Infact, our tests indicated that the convergence condition of the mod-el is mainly determined by the position of CZ where both flow andthermochemical behaviors are significantly different from those inother regions of BF. Therefore, in this study, the criterion for theconvergence is set to 10 pct of the relative CZ positions in two con-secutive iterations. The error is reasonable considering that theproblem is complicated and the variation of the process perfor-mance predictions within this error is small.

4. Simulation conditions

Fig. 2 shows the computational domain and an enlarged areapresenting the representative computational cells. A BF geometrywith the hearth diameter of 10.6 m and inner volume of 2327 m3

is used in the simulation. Assuming the symmetrical distributionof process variables, only half the BF is considered in the simula-tion and is treated as a two-dimensional slot model for simplicity.

S.B. Kuang et al. / Minerals Engineering 63 (2014) 45–56 49

The whole computational domain is divided into 519 � 118 non-uniform control volumes in the Cartesian coordinates. Each com-putational cell is around 5 cm � 6 cm, which should ensure thatthe mesh is small enough to properly capture the complicated mul-tiphase flow in different layers in the lumpy and CZ regions. Notethat the thickness of ore or coke layer at the furnace top is around50 cm. Our trial numerical results indicate that this mesh size givesmesh-independent numerical solutions.

The hot blast (air) is blown into the BF at the flowrate of 4917Nm3/min and temperature of 1050 �C. Corresponding to the blastconditions, the components, temperature and flowrate of thereducing gas generated in the raceway are determined using themass and heat balance, and used as the inlet conditions of gasphase, as listed Table 3. Note that there are no injectants (e.g. pul-verised coal) through the tuyeres, and the water content in theblast is negligible under the condition considered. Thus, the reduc-ing gas mainly consists of CO and N2.

Solids (ore, coke and flux) are charged from the furnace top withthe ore batch weight of 30 tonnes, coke rate of 500 kg/tHM. Theircomponents listed in Table 3 are needed to determine solid heatcapacity and thermal conductivity, as suggested by Austin et al.(1997). The typical radial distributions of ore–coke ratio and sizesof ore and coke as used by Dong et al. (2010) are adopted in thisstudy. This information, together with the productivity, determinesthe solid inlet velocity that is assumed constant along the horizon-tal direction. Note that the productivity is a simulation output inthe present model, which to some degree varies around a target va-lue depending on the in-furnace state under a given condition, asdiscussed in Section 2.4. The solid temperature at the furnace topvaries from 300 to 1300 K to cover a wide range of hot charge tem-perature (HCT). Note that HCT is the only independent variableconsidered in the present study.

With the input information of solids, the physical properties ofthe mixture of ore and coke, such as particle diameter and porositycan be obtained in the computational domain for modelling theflow and heat and mass transfer of solids. In the present study,the ore-to-coke ratio is low at the center, and the coke particle size

(a)Fig. 2. Computational domain (a), and its

is small near the furnace wall. This burden distribution provides ahigh permeability at the furnace center. Initially, the particle prop-erties imposed at the furnace top are extended to the lower part sothat particle size and porosity distributions in the entire furnacecould be obtained. During the calculation, with the identificationof the lumpy zone, CZ, dripping zone, and deadman, the propertiesin these regions are re-estimated to replace the initial values basedon the following rules: (1) in the lumpy zone, porosity in the ore orcoke layer is a function of particle size (see Table 3); (2) in the CZ,porosity and particle size of iron-bearing materials are functions ofnormalized shrinkage ratio as used by Dong et al. (2010); (3) in theCZ and dripping regions, porosity in the areas occupied by coke iscalculated in the same way as used for coke layers in the lumpyzone; and (4) in the deadman, coke size and porosity are assumedas ds = 0.02 m and e = 0.65 according to a normal BF practice.

5. Results and discussion

5.1. Model applicability

It is necessary to examine the applicability of the proposedmathematical model before its application for numerical experi-ments. This is here done using three cases, focused mainly onprocess performance because it is impossible to measure thein-furnace flow and heat and mass transfer at this stage of develop-ment. Three major parameters used to describe process perfor-mance are considered: utilization efficiency and temperature oftop gas as well as productivity. The top gas utilization efficiencyis defined as the volume percentage of CO2 in the mixture of COand CO2 in the gas flowing out of the furnace top. It reflects theutilization efficiency of reducing gas CO inside a BF. The top gastemperature is usually used to assess the thermal energy utiliza-tion efficiency of the furnace. The productivity is the amount ofhot metal produced per furnace volume per day.

In the first case, the applicability of the proposed model in pre-dicting process performance at different coke rates is examined.Note that adjustment of coke rate is a basic BF operation frequently

(b)representative grid arrangement (b).

Table 3Inlet conditions in the base case simulated.

Variables Values

GasInlet velocity, m/s 177Inlet gas component, mole percentage 34.959 pct CO; 0.0 pct CO2; 0.813 pct H2; 0.0Pct H2O; 64.228 pct N2

Inlet gas temperature, K 2313.6Top pressure, atm 2.0

SolidOre, t/tHM 1.64Ore components, mass fraction Fe2O3 0.6566; FeO 0.1576; CaO 0.0652; MgO 0.0243; SiO2 0.06; Al2O3 0.0295; MnO 0.0061; P2O5 0.008Average ore particle size, m 0.03Coke, kg/tHM 500Coke components, mass fraction C 0.857; Ash 0.128; S 0.005; H 0.005; N 0.005Average coke particle size, m 0.045Flux, t/tHM 0.0264Flux components, mass fraction CaO 0.438; MgO 0.079; SiO2 0.024; Al2O3 0.033; CO2 in CaO 0.344; CO2 in MgO 0.082Ore voidage 0.403(100dore)0.14

Coke voidage 0.153logdcoke + 0.742Average ore/(ore + coke) volume ratio 0.5923Burden temperature, K 300

0

5

10

15

20

25

30

35

40

45

420 500 580

Top

gas

uti

lizat

ion

effi

cien

cy (

%)

CR (kg/tHM)

0

100

200

300

400

500

600

700

420 500 580

Top

gas

tem

pera

ture

(K

)

CR (kg/tHM)

0

0.5

1

1.5

2

2.5

3

420 500 580

Pro

duct

ivit

y (t

HM

/day

/m3 )

CR (kg/tHM) CR=420 kg/tHM 500 kg/tHM 580 kg/tHM

Unit: K

(a) (b)

(c) (d)

Fig. 3. Effect of coke rate on BF performance: (a) top gas utilization efficiency, (b) top gas temperature, (c) productivity, and (d) solid temperature.

50 S.B. Kuang et al. / Minerals Engineering 63 (2014) 45–56

0

5

10

15

20

25

30

35

40

45

3663 4360 4884

Top

gas

uti

lizat

ion

effi

cien

cy (

%)

Blast rate (m3/min)

400

420

440

460

480

500

520

3663 4360 4884

Top

gas

tem

pera

ture

(K

)

Blast rate (m3/min)

0

0.5

1

1.5

2

2.5

3663 4360 4884

Pro

duct

ivit

y (t

HM

/m3 /d

ay)

Blast rate (m3/min) Blast rate=3663 m3/min 4360 m3/min 4884 m3/min

Unit: K

(a) (b)

(c) (d)

Fig. 4. Effect of blast rate on BF performance: (a) top gas utilization efficiency, (b) top gas temperature, (c) productivity, and (d) solid temperature.

S.B. Kuang et al. / Minerals Engineering 63 (2014) 45–56 51

practiced in ironmaking plants, and the effect of coke rate on pro-cess performance is well recognized. Fig. 3 shows the calculatedperformance parameters at different coke rates. Here, the simula-tion conditions are based on the base case with consideration ofcoke rate variation. As seen from Fig. 3(a)–(c), when coke rate is in-creased, both productivity and top gas utilization efficiency de-crease, however, the top gas temperature increases. This isbecause a larger coke rate leads to an increased amount of cokecombusted in the raceway, generating more heat and reducinggas CO. Fig. 3 also includes the solid temperature distribution atdifferent coke rates to show two un-expected in-furnace stateswidely encountered in practice. When coke rate is too high (e.g.CR = 580 kg/tHM in Fig. 3d), the BF is over heated and the CZ posi-tion is very high. Thus, the BF becomes too ‘‘hot’’. Conversely, whencoke rate is too low (e.g. CR = 420 kg/tHM in Fig. 3d), the thermalenergy is not enough to maintain a normal operation and the CZposition is very low and the BF may cool down. Overall, the pre-dicted results shown in Fig. 3 are consistent with the practicalapplication of BF (Biswas, 1981).

The second case considers process performance at differentblast rates. It is known that given a certain coke rate, the

productivity of a BF is determined by the amount of oxygen blownthrough the tuyeres. The more oxygen is blown into the furnace,the more coke is consumed and forms CO, and the higher produc-tivity becomes (Geerdes et al., 2009). This relationship is used toexamine the validity of the proposed model. For such a purpose,process performance is predicted at different blast rates based onthe condition in the base case, and the results are given in Fig. 4.As seen from this figure, the productivity increases with the blastrate (Fig. 4c). This is because a larger blast rate leads to an in-creased amount of oxygen blown into the furnace. Furthermore,it is observed that other performance parameters such as temper-ature and utilization efficiency of top gas (Fig. 4a and b) and thethermal state inside the BF as reflected by solid temperature distri-bution (Fig. 4d) are affected slightly by blast rate. This is in linewith the basic requirement in an intensified smelting operationfor high productivity. Again, the predicted results agree with thepractical application of BF (Biswas, 1981; Geerdes et al., 2009).

In the above two cases, only qualitative comparison is made be-cause it is difficult to find some published experimental or plantdata which include the details adequate enough for model valida-tion. Recently, the present model has been used to directly

Fig. 5. Calculated distributions of porosity for the BF operated at: (a) HCT = 300 K, CR = 500 kg/tHM; (b) HCT = 1300 K, CR = 500 kg/tHM; and (c) HCT = 1300 K, CR = 460 kg/tHM.

52 S.B. Kuang et al. / Minerals Engineering 63 (2014) 45–56

simulate a few BFs in operation. The measured and calculated re-sults in terms of productivity, top gas temperature, and top gas uti-lization efficiency are compared quantitatively. The predictionerrors are less than 10%, as reflected elsewhere (Kuang et al.,2012). Nonetheless, the above results shown in Figs. 3 and 4 sug-gest the proposed model can be used to describe the key behaviorsof BF, at least qualitatively. In the following, the effects of hotcharge operation are investigated based on the numerical resultsgenerated by the model.

5.2. Effect of hot charge operation on BF performance

Fig. 5 shows the representative porosity distributions withinthe BF charged with ambient-temperature (Fig. 5a) and hot(Fig. 5b) burdens, respectively, corresponding to conventionaland hot charge operations. According to such distributions, four re-gions can be identified including lumpy zone, CZ, dripping zoneand stagnant zone (or deadman). The results reveal that favourableinverse-V shaped CZs are expectedly obtained in both cases; how-ever, the CZ position rises significantly due to the hot charge oper-ation. This suggests that excessive energy is available in the hotcharge operation. Note that given the same CZ, a BF operated underdifferent conditions could lead to similar flow and thermal condi-tions at the lower part of the furnace, producing hot metal withsimilar qualities. Based on such an understanding, some amountof coke is reduced to achieve the same CZ profile and position asthe conventional operation (or original operation without hotcharge), and the result is given in Fig. 5c. Here, the conventionaloperation representing a normal operation in BF practice is treatedas the base case. As seen from this Fig. 5c, a nearly same CZ profileand position can indeed be obtained for the operations with andwithout hot charge. Notably, to achieve this, the coke rate for thehot charge operation needs to be reduced by 40 kg/tHM. This rep-resents a significant amount of coke reduction, leading to muchless CO2 emission and production costs. Evidently, this largeamount of coke reduction is due to the high hot charge tempera-ture (=1300 K) used in the considered case.

Fig. 6 shows the influence of hot charge operation on processperformance in terms of coke rate, productivity, top gas utilizationefficiency, and top gas temperature at different burden tempera-tures. Here, hot charge operations are considered, respectively, atthe same coke rate (without coke reduction) and the same CZ (withcoke reduction) as the conventional operation. It can be seen fromFig. 6 that when hot charge temperature is increased, the coke ratedecreases and slows down at the same CZ (Fig. 6a). Conversely, thetop gas temperatures under two different conditions of hot chargeoperation both initially increase and then slow down (Fig. 6d). Theresults suggest that a higher hot charge temperature brings morebenefits in view of coke reduction but more challenges in convey-ing/charging gas and burden materials of high temperature. More-over, the gradual decrease of coke rate indicates that the benefitsare less significant when charge temperature is too high. In fact,a too high hot charge temperature is also unfavourable consideringthat it may lead to decreased permeability and deteriorate BF per-formance as a result of powder generation due to degradation ofraw materials and/or softening of ore and flux. Similar problemsmay occur to other BF operations. Modelling of such an adverse ef-fect in a BF process model should be useful but challenging, and de-serves comprehensive studies in the future. At present, it isneglected in this study. On the other hand, it is of interest to notethat the productivity (Fig. 6b) and top gas utilization efficiency(Fig. 6c) show different trends under two different conditions. Gi-ven the same coke rate, with increasing hot charge temperature,the productivity initially remains the same and then increasesslightly, and the top gas utilization efficiency decreases, particu-larly at relatively high hot charge temperatures. Conversely, giventhe same CZ, both productivity and top gas utilization efficiencyrapidly increase and then slow down.

5.3. Effect of hot charge operation on in-furnace state

Process performance is essentially governed by the in-furnacestate such as flow and heat and mass transfer. Analysis of the in-furnace state can generate some insights into BF flow and thermal

430

440

450

460

470

480

490

500

510C

oke

rate

(kg

/tH

M)

HCT (K)

At the same CZ

At the same coke rate

2

2.1

2.2

2.3

2.4

2.5

2.6

300 400 500 600 700 800 900 1000 1100 1150 1200 1250 1300

300 400 500 600 700 800 900 1000 1100 1150 1200 1250 1300

Pro

duct

ivit

y (t

HM

/m3 /d

ay)

HCT (K)

At the same CZ

At the same coke rate

35

36

37

38

39

40

41

42

43

44

45

300 400 500 600 700 800 900 1000 1100 1150 1200 1250 1300

300 400 500 600 700 800 900 1000 1100 1150 1200 1250 1300

Top

gas

uti

lizat

ion

effi

cien

cy (

%)

HCT (K)

At the same CZ

At the same coke rate

0

200

400

600

800

1000

1200

Top

gas

tem

pera

ture

(K

)

HCT (K)

At the same CZ

At the same coke rate

(a)

(b)

(c)

(d)

Fig. 6. Process performance as a function of hot charge temperature: (a) coke rate, (b) productivity, (c) top gas utilization efficiency, and (d) top gas temperature.

S.B. Kuang et al. / Minerals Engineering 63 (2014) 45–56 53

behaviors, hence providing a better understanding of process per-formance as well as some information guiding the design and oper-ation of BF. Such analysis is difficult to achieve experimentally butcan be readily performed based on the numerical simulations. Forclarity, the analysis here mainly focuses on the conventional oper-ation and the hot charge operations at HCT = 1300 K with andwithout coke reduction, similar to Fig. 5.

Fig. 7 shows the gas and solid flow fields for the three cases con-sidered. As seen from the figure, the difference among differentcases for gas flow is mainly observed in the CZ region, where thegas flows mainly through the coke window (Fig. 7a). Comparedto the conventional operation, both hot charge operations regard-less of coke reduction generally lead to larger gas velocities, whichis not desirable for preventing erosion of wall refractory materials.On the other hand, all the solid flow fields show similar distribu-tions. The solid velocities are observed to considerably decreasein the dripping zone as only coke is left there. In the lumpy zone,the solid velocities in both hot charge operations are generally lar-ger than the conventional operation, corresponding to the produc-tivity shown in Fig. 6b. As such, the slip velocities between gas andsolid phases are increased by a hot charge operation, particularlywhen coke rate is high.

Fig. 8 shows the solid and gas temperature fields. Compared tothe conventional operation, both hot charge operations have high-er solid temperatures but mainly in the lumpy zone, which alsoshow more uniform distributions in this region (Fig. 8a). Such highsolid temperatures in the lumpy zone should promote solution lossreaction which is an intensively endothermic reaction, leading tomore coke consumption. In addition, it is noted that the solid tem-perature near the wall at the upper part of the furnace in the hotcharge operation with coke reduction is lower than the hot chargeoperation without coke reduction. This may be due to that the heattransfer from the hot burdens to the gas is intensified by the cokereduction, leading to less energy generated inside the furnace. Cor-responding to the solid temperature distributions, the gas temper-atures at the upper part of the furnace in both hot chargeoperations are higher than the conventional operation (Fig. 8b),achieved by the extra heat brought into the furnace by the hot bur-dens. This leads to gas expansion and larger gas velocities in thetwo cases as observed in Fig. 7a. Note that the gas or solid temper-atures in Fig. 8 (also Figs. 3d and 4d) are largely uniform in thedeadman or stagnant zone because here the gas flow through thisregion is small and the heat exchange between gas and solidphases is also small.

HCT=300K 1300 K 1300 KCR=500 kg/tHM 500 kg/tHM 460 kg/tHM

(b)

(a)

Fig. 7. Gas (a) and solid (b) flow fields, corresponding to Fig. 5.

HCT=300 K 1300 K 1300 KCR=500 kg/tHM 500 kg/tHM 460 kg/tHM

(b)

(a)

Fig. 8. Distributions of: (a) solid temperature, and (b) gas temperature, corre-sponding to Fig 5.

54 S.B. Kuang et al. / Minerals Engineering 63 (2014) 45–56

Fig. 9 shows the gas pressure distributions and reveals that hotcharge operations regardless of coke reduction generally result inlarger gas pressure in the entire furnace compared to the conven-tional operation. This should be attributed to the increased slipvelocity between gas and solid phases as a result of gas expansionand increased productivity. Note that a large pressure drop maycause unfavourable phenomena for smooth running of the BFincluding fluidization, hanging and channelling. Interestingly, it isobserved that coke reduction, coupled with hot charge operation,can to some degree decrease gas pressure, although coke adjust-ment itself may increase gas pressure significantly, if not con-trolled reasonably. It should be pointed out that the findings

from the analysis of in-furnace state can also apply to other hotcharge temperatures, whose results are not included in this paperfor brevity.

In order to better link the in-furnace state to the BF perfor-mance, analysis of coke consumption in the furnace is performedand the results are given in Fig. 10. As seen from this figure, thecoke rate is nearly equal to the sum of the coke consumptions bythe combustion in the raceway (=the amount of tuyere coke) andreactions. Overall, the tuyere coke is the dominating contributorto the coke rate. With increasing hot charge temperature, the

HCT=300 K 1300 K 1300 K

CR=500 kg/tHM 500 kg/tHM 460 kg/tHM

Fig. 9. Distributions of gas pressure, corresponding to Fig. 5.

200 400 600 800 1000 1200 1400100

120

140

320

360

400

440

480

520

560

Am

ount

of

coke

(kg

/tHM

)

HCT (K)

Filled: at the same CZHollow: at the same coke rate

Fig. 10. Different coke consumptions as a function of hot charge temperature (A:Coke charged, R: Coke consumed by reactions, and D: Coke combusted in theraceway).

S.B. Kuang et al. / Minerals Engineering 63 (2014) 45–56 55

tuyere coke amount decreases at the same CZ, however, it de-creases slightly at the same coke rate. Meanwhile, the coke con-sumptions by reactions under two different conditions of hotcharge operation both increase. It is known that the productivityof a BF increases with the decrease of tuyere coke amount in a sta-ble situation (Geerdes et al., 2009). This is because every charge ofcoke at the top of a furnace brings with it an amount of iron-bear-ing materials, and the hot metal is produced as soon as the coke isconsumed under the stable condition. On the other hand, the topgas utilization efficiency may decrease with the increase of tuyerecoke amount or coke consumption by reactions because morereducing gas CO is generated from coke. Based on this understand-ing, the calculated flow and thermal behaviors as well as processperformance at different hot charge temperatures can be furtherexplained using the results shown in Fig. 10. At the same CZ orcoke rate, the increased productivity at a higher hot charge tem-perature (Fig. 6b) is essentially attributed to the decreased tuyerecoke amount. The variation of tuyere coke amount at the sameCZ is more considerable than that at the same coke rate, leadingto more significant variation of the productivity.

Fig. 10 can also be used to explain the top gas utilization effi-ciency shown in Fig. 6c. At the same coke rate, the variation oftuyere coke amount is small, the decreased top gas utilization effi-ciency at a higher hot charge temperature is mainly attributed tothe increased coke consumption by reactions. However, with thesame CZ, the tuyere coke amount and coke consumption by reac-tions show opposite trends at different hot charge temperatures.Their effects on top gas utilization efficiency are cancelled out byeach other. This is particularly evident at a high temperaturewhere the decrease of tuyere coke slows down, whereas the in-crease of coke consumption by reactions largely remains linear.Note that the tuyere coke amount contributes to a larger portionof the coke rate and plays a dominating role. Consequently, the de-creased tuyere coke amount at a higher hot charge temperatureessentially accounts for the decrease of coke rate and thus the in-crease of top gas utilization at the same CZ. However, it is theopposite effect of the two coke consumptions that accounts forthe gradual variation of top gas utilization efficiency at a highhot charge temperature. In addition, it should be pointed out thata decreased tuyere coke amount leads to less heat brought fromthe lower part to the upper part of the furnace. This explainswhy given the same burden temperature, the solid temperaturesand CZ position generally decrease in the hot charge operation ata smaller coke rate, as shown in Figs. 5 and 8.

6. Conclusions

The multiphase flow, heat and mass transfer in a BF have beenstudied by a BF process model, with special reference to hot chargeoperation. The model can satisfactorily predict BF performancesuch as productivity, top gas utilization efficiency and top gas tem-perature under different conditions, although further develop-ments may be needed, particularly for more complicated BFoperations, e.g. combustion of injected pulverized coal in the race-way. The findings from this study can be summarized as follows:

(1) Compared to the conventional operation, hot charge opera-tion can lead to an increased productivity, and decreasedcoke rate and CO2 emission, and at the same time, increasedgas pressure and top gas temperature. In general, withincreasing hot charge temperature, these effects are moreevident. However, the gains do not vary much at high hotcharge temperatures.

(2) Hot charge operation may lead to a higher solid temperatureand a larger gas pressure than the conventional operation,particularly at the upper part of the furnace. The effectsare unfavorable in view of coke consumption and smoothoperation. This problem can be overcome by decreasing cokerate to produce a cohesive zone of the same position. Thistechnique can also be used to quantify the coke reductionswith different hot charge temperatures.

(3) The coke reduction as a result of hot charge operation ismainly due to the reduced amount of coke combusting inthe raceway, which also accounts for the increases of gas uti-lization efficiency and productivity.

Finally, it should be pointed out that to date, hot charge opera-tion has not been practiced due to its high requirements in equip-ment as a result of high temperature environment at the furnacetop. If this problem can be overcome, and the properties of burdenmaterials do not change much, hot charge operation should bebeneficial to BF performance, as demonstrated in this study. There-fore, related aspects need to be investigated carefully. At this stage,caution should be taken in developing and implementing this tech-nology in practice.

56 S.B. Kuang et al. / Minerals Engineering 63 (2014) 45–56

Acknowledgements

The authors are grateful to the Australia Research Council (ARC)and Central Iron & Steel Research Institute (CISRI) for the financialsupport of this work, and the National Computational Infrastruc-ture (NCI) for the use of its high performance computationalfacilities.

References

Austin, P.R., Nogami, H., Yagi, J.I., 1997. A mathematical model for blast furnacereaction analysis based on the four fluid model. ISIJ International 37 (8), 748–755.

Austin, P.R., Nogami, H., Yagi, J.I., 1998. Prediction of blast furnace performance withtop gas recycling. ISIJ International 38 (3), 239–245.

Biswas, A.K., 1981. Principles of Blast Furnace Ironmaking: Theory and Practice.Cootha Publishing House.

Chu, M.S., Yagi, J.I., 2010. Numerical evaluation of blast furnace performance undertop gas recycling and lower temperature operation. Steel Research International81 (12), 1043–1050.

Dong, X.F., Yu, A.B., Yagi, J.I., Zulli, P., 2007. Modelling of multiphase flow in a blastfurnace: recent developments and future work. ISIJ International 47 (11), 1553–1570.

Dong, X.F., Yu, A.B., Chew, S.J., Zulli, P., 2010. Modeling of blast furnace with layeredcohesive zone. Metallurgical and Materials Transactions B – Process Metallurgyand Materials Processing Science 41 (2), 330–349.

Ergun, S., 1953. Pressure drop in blast furnace and in cupola. Industrial andEngineering Chemistry 45 (2), 477–485.

Ferziger, J., Peric, M., 2002. Computational Methods for Fluid Dynamics. Springer,New York.

Geerdes, M., Toxopeus, H., Vliet, V.D., 2009. Modern Blast Furnace Ironmaking: AnIntroduction. IOS Press BV, Amsterdam, the Netherlands.

Helle, H., Helle, M., Pettersson, F., Saxen, H., 2010. Multi-objective optimization ofironmaking in the blast furnace with top gas recycling. ISIJ International 50 (10),1380–1387.

Kawanari, M., Matsumoto, A., Ashida, R., Miura, K., 2011. Enhancement of reductionrate of iron ore by utilizing iron ore/carbon composite consisting of fine iron oreparticles and highly thermoplastic carbon material. ISIJ International 51 (8),1227–1233.

Kuang, S.B., Li, Z.Y., Yu, A.B., 2012. Process modelling and analysis of ironmakingprocesses for improving energy efficiency in Baosteel. Internal Report toBaosteel.

Li, J.X., Wang, P., Zhou, L.Y., Cheng, M., 2007. The reduction of wustite with highoxygen enrichment and high injection of hydrogenous fuel. ISIJ International 47(8), 1097–1101.

Maldonado, D., 2003. Heat and mass transfer in blast furnace: low temperaturecohesive layer experiments and mathematical modelling at low and hightemperates. PhD thesis, University of New South Wales, Sydney.

Muchi, I., 1967. Mathematical model of blast furnace. Transactions ISIJ 7 (1), 223–237.

Nogami, H., Yagi, J.I., Kitamura, S., Austin, P.R., 2006. Analysis on material andenergy balances of ironmaking systems on blast furnace operations withmetallic charging, top gas recycling and natural gas injection. ISIJ International46 (12), 1759–1766.

Omori, Y., 1987. Blast furnace phenomena and modelling. Elsevier Applied Science,London, UK.

Shen, Y.S., Guo, B.Y., Yu, A.B., Zulli, P., 2009. A three-dimensional numerical study ofthe combustion of coal blends in blast furnace. Fuel 88 (2), 255–263.

Slaby, S., Andahazy, D., Winter, F., Feilmayr, C., Burgler, T., 2006. Reducing ability ofCO and H2 of gases formed in the lower part of the blast furnace by gas and oilinjection. ISIJ International 46 (7), 1006–1013.

Zhang, S.J., Yu, A.B., Zulli, P., Wright, B., Tuzun, U., 1998. Modelling of the solids flowin a blast furnace. ISIJ International 38 (12), 1311–1319.