45 (2Combustion characteristics in an iron ore sinteringbedevaluation of fuel substitution

Won Yang a,1, Sangmin Choi a,, Eung Soo Choi b, Deog Won Ri b,Sungman Kim b

a Department of Mechanical Engineering, Korea Advanced Institute of Science and Technology, Guseong-dong, Yuseong-gu,Daejeon, South Korea

b POSCO, Technical Research Labs, Goedong-dong, Nam-gu, Pohang, Gyeongbuk 790-785, South Korea

Received 14 May 2005; received in revised form 15 December 2005; accepted 12 January 2006Available online 9 March 2006

Abstract

In an iron ore sintering bed, combustion of solid fuel supplies heat needed for sintering of fine particles anddetermines the quality of the sintered ores and the productivity of the process. Coke has been widely used asan ideal fuel for this process, but recent attempts to partially replace coke with a less expensive fuel have beenconsidered effective in the real applications. This paper reports simulation results of mathematical modeling of bedcombustion as well as experimental observation from scale-downed pot tests. Combustion characteristics of thesolid fuel bed are also described and fuel substitution from coke to anthracite coal is evaluated. A numerical modelof the 1-D unsteady level considers the processes in the bed of particles: drying, devolatilization, char reactions,gaseous reactions, heat transfer, and geometrical changes of the bed material. The model can treat the solid materialas multiple solid phases whose contents include fine particles of iron ore, limestone, coke, and coal. Quantitativeparameters are newly defined for characterization of the bed combustion. These include flame front speed, sinteringtime, duration time in the combustion zone, combustion zone thickness, melting zone thickness, and maximumtemperature. Relationships between these parameters and the mechanism of combustion propagation are alsoinvestigated. The results show quantitatively that temperature profiles, combustion propagation, and thickness ofthe combustion zone in the bed are dominated by combustion-related operating parameters. This description isapplied to case studies for various coke contents, air supply rates, and fuel characteristics. It is shown that thedifferent reactivity of anthracite in comparison with coke can significantly affect the bed combustion in terms ofcombustion propagation and the quality of the sintered ores in the upper part of the bed. 2006 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

Keywords: Coke; Anthracite; Combustion characteristics; Iron ore sintering; Sintering pot

* Corresponding author.E-mail address: smchoi@kaist.ac.kr (S. Choi).

1 Current address: Industrial Equipment Team, KoreaInstitute of Industrial Technology, 35-3 Hongchon-Ri,Ipchang-Myun, Chonan-Si, Chungnam 330-825, South Ko-rea.

1. Introduction

Coal combustion in a mixed bed of solid materialis commonly utilized in ironmaking processes suchas iron ore sintering and blast furnace processes. Typ-ically, coal is supplied as coke, which is produced bycoking coal in an oven. At present, a limited amountCombustion and Flame 10010-2180/$ see front matter 2006 The Combustion Institute.doi:10.1016/j.combustflame.2006.01.005006) 447463www.elsevier.com/locate/combustflamePublished by Elsevier Inc. All rights reserved.

and FSintering is a preprocess to form iron ore pow-der (typically less than 1 mm in size) into aggregatesof porous chunks of iron ore (typically larger thana few centimeters, with physical properties suitablefor operation in a blast furnace). Coal occupies atmost 4% of solid material in the form of coke fine.Coke combustion provides iron ores with the energyrequired for sintering. Fig. 1 shows a schematic dia-gram of the sintering process. Various kinds of ironores, coke, and limestone are mixed with water in arotating drum and form pseudo-particles. They are fedinto a sintering bed having a length of 100 m and aheight of 0.6 m and move along a grate at a con-stant speed. The top of the bed is heated and ignitedby a gas burner during 100 s. Once ignited, cokecombustion is sustained for 2530 min to the end ofthe process without any external heat source. Surfacemelting and sintering occur in the high-temperatureregion. Combustion air is supplied by a suction fan,and the air exchanges heat with hot agglomerated oresover the combustion zone. After passing through thecombustion and sintering zone, the hot gas is used forevaporation of the raw mix and transfers the steamdownstream. The steam is condensed in the cold re-

grate length and width are neglected, an iron ore sin-tering process can be considered as 1-D unsteady(along the direction of bed height) when the timeaxis can be transformed into length by multiplica-tion of the traveling speed by the time, which is usu-ally constant during the whole process. For this rea-son, many researchers have numerically simulated thesintering process using 1-D unsteady partial differ-ential equations [18] and have performed lab-scaleexperiments known as sintering pot tests [14,6].Results of the tests can provide designers and oper-ators with useful information on combustion condi-tions, flue gas compositions, and quality of the prod-uct, which can be used for optimization of the wholesintering process. This approach has been reported inthe field of combustion modeling of solid fuel bedssuch as waste incinerators [811] and coal/biomasscombustion/gasification systems [1218]. Most previ-ous works treated the solid material as a single solidphase. However, this approach does not appear to beappropriate in the combustion modeling of a sinter-ing bed, in which various kinds of solid material areuniformly mixed and the size of each solid mater-ial can significantly affect the combustion conditions448 W. Yang et al. / Combustion

Nomenclature

CZT Combustion zone thickness, cmCp Specific heat, J/kg KDTCZ Duration time in combustion zone, sFFS Flame front speed, cm/minfv Volume fractionh Convective heat transfer coefficient,

W/m2 Kh Thermal enthalpy per unit volume, J/m3k Thermal conductivity, W/m KMg Volumetric mass generation rate by

gaseous reaction, kg/m3 sMs Volumetric mass generation rate by

solidgas reaction, kg/m3 sMaxT Maximum temperatureMZT Melting temperatureq Heat per unit volume, W/m2S General source termST Sintering timeT Temperature, Kt Time, s

of coke is being replaced with less expensive coal.However, coke is still widely used because of itsunique combustion properties, as well as its mechani-cal and structural properties.gion of the bed in the lower part. These coal combus-tion characteristics in the bed significantly influencethe quality and productivity of sintered ore. Thus, itlame 145 (2006) 447463

v Velocity, m/sY Mass fractiony Vertical distance, my Fraction of heat absorbed by solid General diffusion coefficientH Reaction heat, J/kgP Pressure difference, Pa Density, kg/m3 General scalar quantity

Subscriptsconv Convectiong GasI, J Index of solid phasek Solid component or gas speciesrs Solidgas reactionrg Gaseous reactionrad Radiations Solid

is important to control the combustion-related para-meters, which should be set by investigation of coalcombustion characteristics in the iron ore bed.

If the transfer phenomena along the directions ofand productivity of the process. Therefore, some re-searchers proposed a 1-D unsteady model [7] of solidbed combustion that can treat solid material as mul-

and

ring p

ity and characteristics of the products. For this reason,

some additional parameters need to be introduced forcharacterization of the combustion zone, and numeri-cal study would extend the applicability of the avail-able parameters.

One of the most important issues in the ironmak-ing process, including sintering, is fuel substitutionfor coke, which has traditionally been used. Coke de-rived from softening coal is more reactive, but is con-siderably more expensive. For this reason, coke hasbeen replaced in part with powdered anthracite as alow-cost fuel. However, the difference in reactivityof anthracite must be evaluated to avoid undesirableeffects in terms of productivity and quality of the sin-tered ore.

In this study, a systematic method for analyzingthe computational and experimental results of the ironore sintering process is proposed. Experiments andnumerical test cases are designed to simulate sintering

characteristics of an iron ore sintering bed are de-scribed in terms of temperature and solid componentdistribution within the bed.

Fig. 2 shows a conceptual diagram of the com-bustion characteristics in an iron ore sintering bedin terms of temperature distribution. Definitions ofthe quantitative parameters are also introduced forcharacterization of solid bed combustion. Combus-tion/sintering is initiated from the top of the bed. Asthe combustion zone propagates downward, its thick-ness increases by accumulation of combustion heat.The solid bed is divided into four regions by temper-ature and mass fraction of char within the bed: a rawmix zone, a combustion zone, a melting zone, anda completely sintered zone. The combustion zone isdefined as the area in which the temperature of thesolid phase, which includes coal char, is high enoughto obtain oxidation and char mass fraction is suffi-ciently larger than a small value . This means that theW. Yang et al. / Combustion

Fig. 1. Schematic diagram of sinte

tiple solid phases. Thus, different fuel characteristicssuch as particle size and composition can be effec-tively simulated. However, a more systematic, quan-titative analysis is needed for interpretation and eval-uation of the results for practical design and opera-tion. In particular, combustion characteristics shouldbe described quantitatively for evaluation of the op-erating conditions of the process. Flame front speed(FFS) is considered an important quantitative parame-ter for characterization of the bed combustion [7,9]. Itcan describe the propagation speed of the combustionzone, which represents the productivity of the sinteredores. However, features of the combustion/reactionzone itself are also important with respect to the qual-pot tests and fuel contents and air supply are paramet-rically studied. Some quantitative parameters of bedcombustion are introduced and the test results are dis-Flame 145 (2006) 447463 449

rocess in the ironmaking process.

cussed in conjunction with these parameters. Quan-titative discussion is designed to show the effects offuel replacement of coke with coal.

2. Characterization of bed combustion

The key features of solid bed combustion thatdistinguish it from the combustion characteristicsof a single fuel particle are combustion propagationthrough the bed and temperature and thickness of thecombustion zone itself. These features are closely re-lated to combustion conditions such as air/fuel ratioand external heat supply, as well as the fuel charac-teristics of a single particle. In this study, combustionchar combustion occurs in the zone where the solidtemperature and are above 1000 K and 105. Com-bustion characteristics in an iron ore sintering bed can

and F

interine thicbetween the combustion zone and the raw mix.Based on this division of the solid bed, quantita-

tive parameters are defined to characterize the solidbed combustion. Flame front speed (FFS) was pre-viously introduced [7,9] to express the propagationspeed of the combustion zone in the solid fuel bed.Sintering time (ST) is also used to express the prop-agation speed. Definitions of these parameters aregiven in Table 1. FFS is defined as the distance be-tween two points along the vertical axis in the bedover the time consumed for propagation. These twopoints can be arbitrarily selected. The propagationtime is defined as the time consumed to reach 1000 Kbetween the two points. The FFS is mainly influencedby the char combustion rate and parameters of heattransfer and thermal diffusion within the bed. ST isdefined as the elapsed time until the flue gas tempera-ture reaches a maximum, which means that coke com-bustion is nearly completed, as shown in Fig. 2. It isinversely proportional to the flame front speed. ST is

the combustion zone itself, even though the char-acteristics can seriously influence the quality of theproduct from the sintering process. Therefore, inaddition to FFS and ST, new parameters are intro-duced to describe the characteristics of the combus-tion zone itself: duration time in the combustion zone(DTCZ), combustion zone thickness (CZT), meltingzone thickness (MZT), and maximum temperature(MaxT). These parameters can be determined fromthe simulation model. CZT and MZT are definedas the thickness of the combustion zone and melt-ing zone, respectively. They can vary with time andtend to increase as the process proceeds [7]. Since themelting zone is not necessarily included in the com-bustion zone, MZT can be larger or smaller than CZT.DTCZ is defined as the retention time above the ini-tiation temperature of char combustion (assumed tobe 1000 K in this study) when the mass fraction ofchar content is above (considered to be 105 in thisstudy), which is a function of time and location in the450 W. Yang et al. / Combustion

Fig. 2. Characterization of bed combustion in an iron ore sduration time in the combustion zone, CZT: combustion zonperature).

be significantly affected by the melting of iron ores,which influences the bed permeability and therebydisturbs airflow. Consequently, the amount of iron oremelting should be controlled to maintain appropriatecombustion conditions and to produce sintered oresof high quality. The melting zone is defined as the re-gion in which iron ore melting occurs: In this study,iron ore undergoes melting when the solid tempera-ture is above 1373 K [7]. When char combustion iscompleted and the temperature is above the meltingtemperature, which implies that the zone is locatedin the upper part of the melting zone, the hot bed isquenched by inlet air. The flame front is the boundarywell known in the field of ironmaking process as a keyparameter of productivity in the sintering process. Thetwo parameters, i.e., the FFS and the ST, can servelame 145 (2006) 447463

g bed (FFS: flame front speed, ST: sintering time, DTCZ:kness, MZT: melting zone thickness, MaxT: maximum tem-

as indices of combustion propagation characteristicsfor a single case, since they have one representativevalue for the whole process. Furthermore, they canbe obtained by experiment and can be used as para-meters for validation of the numerical model. Moredetailed information on the combustion propagationmechanism related to various modes of heat genera-tion and transfer can be obtained through calculationof the heat balance terms.

The above parameters cannot, however, adequatelydescribe all combustion conditions in the sinteringbed, since they consider only the propagation ofthe combustion zone, but not the characteristics ofbed. These parameters can be utilized to account forthe amount of combustion heat and the degree of heatconcentration in the bed. The maximum temperature

and

ustion

en twor pro

il theumove tustionzone

the i

zone

than

ratureFig. 3 shows a schematic diagram of the sinter-ing pot test rig. The bed has a diameter of 205 mmand a height of 600 mm. R-type thermocouples areinstalled along the axial direction of the bed: y =0.49 m (the upper part of the bed), y = 0.30 m (cen-ter), and y = 0.11 m (the lower part of the bed).Thus, the numerator of Eq. (1) in Table 1 is definedas the distance between the locations y = 0.49 mand y = 0.11 m. Flue gas temperature is measuredby a K-type thermocouple installed in the wind box.Hearth ores of diameter 10 mm are installed at thebottom of the pot to improve gas permeability. Theassembly of pseudo-particles for the experiment fol-lows the procedure shown in Fig. 1. Iron ores, coke,limestone, and other additives are mixed with waterin the rolling chamber, which produces various sizesof pseudo-particles. Iron ores consist of many kinds ofores such as hematite, magnetite, and limonite of vari-

Fig. 3. Schematic diagram and dimensions of the sinteringpot.

are obtained by a series of sieves (7.9, 5.0, 2.8, 2.0,and 1.0 mm). Table 2 shows the distributions of thediameters of the pseudo-particles. Note that the mass-averaged value of the particle sizes is 3.2 mm.

After the selected amount of raw material, 35 kgin this experiment, is fed into the pot, the bed is ex-posed to the LPG burner for 90 s and is allowed toignite. The pressure difference of the suction fan isset at 9810 Pa (1000 mm Aq) during ignition and at14,715 Pa (1500 mm Aq) afterward. Gas inlet veloc-ity, flow rates, and flue gas compositions includingW. Yang et al. / Combustion

Table 1Definition of the parameters for characterization of bed comb

Parameters DefinitionFlame front speed (FFS) Distance betwe

Time consumed f

Sintering time (ST) Elapsed time untindicates a maxim

Duration time in combustion zone(DTCZ)

Retention time abture of char comb

Combustion zone thickness (CZT) Thickness of theture is greater thanbustion

Melting zone thickness (MZT) Thickness of theperature is higherore melting

Maximum temperature (MaxT) Maximum tempeat a given time

is defined as the highest temperature of the solid ma-terial at a given time. Sintering of the particles wouldrequire a high temperature, but this should be evalu-ated carefully with other parameters. The parametersrelated to the characteristics of the combustion zoneare significantly affected by the fuel characteristicsrelated to ignition, reactivity, and combustion time. Ifa fuel has low reactivity and long combustion time,CZT and DTCZ will be increased. However, the CZTand the DTCZ are also affected by the combustionconditions, such as air/fuel ratio, since the combus-tion time is directly influenced by the air supply.

3. Experimentsous origins. For measurement of particle diameter, theparticles are frozen by liquid nitrogen injection andthe mass fractions for each range of specific diameterFlame 145 (2006) 447463 451

Obtained by

(1)o pointspagation

Measurement and computation

flue gas temperature Measurement and computation

he initiation tempera- Computation

whose solid tempera-nitiation of char com-

Computation

where the solid tem-the initiation of iron

Computation

of the solid material ComputationO2, CO2, and CO are continuously measured. Massconservation of the gas is checked by measuring theinlet velocities and the volumetric gas flow rate at the

and F

t y y yMass, energy, and species conservation equationsfor each phase, as summarized in Table 4, weresolved. Submodels of solidgas reactions, gaseous re-actions, heat transfer, and geometrical changes of thesolid particles and bed structure are reflected in thecalculation of source terms and physical properties.

of the heat balance terms for the reference case arealso exhibited: solidgas convection, radiation, solidgas reaction (drying + char combustion/gasification+ limestone decomposition), and conduction, at y =0.11 m and y = 0.49 m. In the initial stage of thesolidgas reaction, solidgas convection plays an im-portant role in drying and heating the solid particles.452 W. Yang et al. / Combustion

Table 2Distribution of the pseudo particle sizes

Diameter (mm)+7.9 +5.0 +2.8

Mass fraction 0.0753 0.1151 0.2277

Fig. 4. Typical trends of the pressure difference and gas flowrate in the sintering pot [7].

outlet. Fig. 4 shows typical profiles of the pressuredifference and gas flow rate measured in this study.The initial pressure drop represents the amount ofcombustion air. After the initial value is set, the suc-tion pressure varies as coal combustion progresses,while the rotation speed of the fan is held constant.

4. Model study

A 1-D unsteady model was constructed to investi-gate the effects of phenomena in detail and to obtainthe values of quantitative parameters related to thecoal combustion characteristics in an iron ore sinter-ing bed. Table 3 outlines the scope of the model. Solidmaterial is modeled as homogeneous porous mediaand multiple phases, which enables the considerationof different fuel characteristics and various particlediameters. Therefore, governing equations can be de-scribed as a system of partial differential equationswith the general form

(2)(fv) + (v) = (eff

)

+ S.Details of the model were reported in a previouslypublished paper [7]. Simulation results show goodagreement with the measurement data for temperaturelame 145 (2006) 447463

Average+2.0 +1.0 1.00.1700 0.2417 0.1700 3.2 mm

distribution, flue gas composition, and flame propaga-tion speeds of the combustion zone [7].

Solid materials were considered as three solidphases of iron ore, coke, and limestone. Considerationof multiple solid phases makes it possible to investi-gate the effects of different fuel characteristics on thebed conditions. Other additives that were not involvedin the solidgas reactions are categorized as being partof the inert material. Each solid phase consists of solidcomponents such as moisture, char, iron ore, CaCO3,CaO, and inert material. Each of them has specificphysical properties such as density, specific heat, andthermal conductivity. O2, H2O, CO2, CO, H2, and N2were selected as the gas species. Devolatilization wasnot considered since the coal contains few volatiles(

andGeneral 1-D unsteadyConsideration of solid material Homogeneous porous media

Multiple solid phasesGoverning equations Mass, energy, and species conservation for each solid phase and gas phaseSubmodels

Heat transfer Conduction, convection (between solid and gas), radiation, heat exchangebetween solid phases

Solidgas reactions Drying, devolatilization, char combustion and gasificationGaseous reactions CO combustionGeometrical changes Particle shrinkage by char combustion/gasification

Generation of macroscopic internal poresBed height/porosity changes

Table 4Summary of governing equations

eff S

Mass Solid phase I 1

rsMs,I,rs

Gas phase 1 I rs Ms,I,rsEnergy Solid phase I hs,Is,I

fv,s,I ks,Is,I Cps,I

I hconv,gI As,I (Tg Ts,I ) + qconv,ss + qrad+rs yI Ms,I,rsHr

(rs

Ms,I,rs

)Cp,I Ts,I

Gas phase hggfv,gkggCpg

I hconv,gI As,I (Ts,I Tg)+rs (1 yI )Ms,I,rsHrs +

(rs

Ms,I,rs

)Cp,I Ts,I

Component k Solid phase I Ys,I,k

rsMs,I,k,rs

Gas species k Gas phase Yg,k

rs

I Ms,I,k,rs +

rg Mg,k,rg

Table 5Experimental and calculation cases

Case name Solid fuel U (m/s) Contents ofsolid fuel(mass%, wet)

Mean fueldiameter(mm)

Moisturecontents(mass%)

Experiment/calculation

Coke3.4 Coke 0.42 3.4 2.0 8.0 BothCoke3.8a Coke 0.42 3.8 2.0 8.0 BothCoke4.2 Coke 0.42 4.2 2.0 8.0 BothAirV0.26 Coke 0.26 3.8 2.0 7.0 BothAirV0.32 Coke 0.32 3.8 2.0 7.0 BothAirV0.42 Coke 0.42 3.8 2.0 7.0 BothAirV0.46 Coke 0.46 3.8 2.0 7.0 BothCoke100 Coke 0.42 3.7 2.0 7.0 CalculationAnth100b Anthracite 0.42 3.93 2.0 7.0 CalculationAnth1.2 Anthracite 0.42 3.93 1.2 7.0 CalculationAnth1.4 Anthracite 0.42 3.93 1.4 7.0 CalculationAnth1.6 Anthracite 0.42 3.93 1.6 7.0 CalculationAnth1.8 Anthracite 0.42 3.93 1.8 7.0 Calculation

a Coke3.8 is a reference case.b Anth100 is the same case as Anth2.0.

limestone composition, which mainly occurs in theinitial stage of char combustion. After limestone de-

value, indicating that the temperature of the solid ma-terial is higher than that of the gas flow. RadiationW. Yang et al. / Combustion

Table 3Scope of the modelcomposition is completed and the temperature of solidmaterial becomes higher than that of the combustiongas, the solidgas convection term falls to a negativeFlame 145 (2006) 447463 453and conduction between the solid material at an ob-jective cell and neighboring cells do not significantlyaffect the whole process of combustion propagation.

and F454 W. Yang et al. / Combustion

Fig. 5. Heat balance terms at y = 0.11 m and y = 0.49 m.

Comparing profiles of the heat balance terms at thepoints y = 0.49 m and y = 0.11 m, the peak valuesof solidgas convection and reaction at y = 0.49 mare found to be higher than the values at y = 0.11 m.These data show that the fresh air passes through athicker and hotter sintered zone where the reactionsare completed, and then contacts the fuel with higherair temperature. This causes the solid particles to beheated more rapidly and thereby accelerates the solidgas combustion at y = 0.11 m.

The mechanism of combustion propagation in theiron ore sintering bed is somewhat different frompropagation in a waste bed, as investigated numeri-cally in a previous work [9]. In the case of the in-cinerator waste bed, radiation between the waste bedand hot gas/wall plays an important role in ignitionand combustion propagation, while convection is im-portant for heating and ignition of fuel particles in thecase of an iron ore sintering bed. The difference arisesfrom radiative heat input at the initial stage of theprocess in the waste bed. Furthermore, the radiativeheat input penetrates the bed better since the solid par-ticles inside the waste bed are much larger than thosein the sintering bed; solid particles simulated for thewaste bed were 1030 mm in diameter [9], while theparticles in the sintering bed were 13 mm.

Fig. 6 shows the calculation result of CZT andMZT compared with the thickness of the high-temperature region (>1000 K) for the reference case.The profile of the thickness of the high-temperatureregion includes the MZT profiles. The MZT and thethickness of the high-temperature region, which aredetermined only from the temperature, tend to in-crease as the process continues. On the other hand,CZT shows a relatively flat trend, accounting for theincreasing difference between CZT and MZT, thethickness of the high-temperature region. The gapsare due to convection cooling by inlet air in the upper

part of the high-temperature zone, which implies thatthe combustion heat accumulates during the process.Generally, the accumulation causes additional melt-lame 145 (2006) 447463

Fig. 6. Calculation results of CZT and MZT compared withthe thickness of the high-temperature zone (above 1000 K).

ing of iron ores, mainly in the lower part of the bed,thus decreasing bed permeability. This effect shouldbe reflected in the model for further numerical studyof the iron ore sintering process.

5.2. Effect of coke content

Fig. 7 shows the experimental results of temper-ature profiles and gas compositions for various cokecontents for a fixed air supply value. For higher cokecontent, the high-temperature zone is shown to bebroader and the peak temperature is higher, imply-ing that combustion is apparently better sustained forhigher fuel content in this condition. This is reflectedin the O2 and CO2 composition in the flue gas. Higherconcentration of CO2 and lower O2 concentration arefound for higher coke content, which shows that thecoke combustion rate increases for higher coke con-tent. Additionally, the width of the temperature profiletends to increase for higher coke content, which in-dicates that DTCZ at a given location is increasing.Fig. 8 shows the calculation results of DTCZ and peaktemperature for various coke contents. Increase of theDTCZ and MaxT for higher coke content is found.Additionally, the DTCZ and the MaxT are increasedfor the lower part of the bed. These increases are at-tributed to increased convection heat transfer from thegas phase and acceleration of coke combustion. Atthe same time, the coke combustion rate is limitedby oxygen diffusion at high temperature, where, forhigher coke content, the duration time is extended un-til the coke is completely consumed.

The results described above are supported by fur-ther calculation for the cases of various coke contentand by the model-based quantification parameterscombustion zone thickness (CZT) and maximum tem-

perature (MaxT) profilesas shown in Fig. 9. Nosignificant difference is found in the profiles of MaxTbetween 3.8 and 4.2% coke content, except for the

andW. Yang et al. / Combustion

(a)

(b)

Fig. 7. Test results of temperature profiles (a) and gas com-positions (b) for various coke content.

Fig. 8. Peak temperature and DTCZ for various coke content.

decreasing time of MaxT, which shows the comple-tion time of the coke combustion. However, the MaxTvalues during 2001100 s, where the MaxT profileshows a flat trend, are lowered by 50 K in the caseof 3.4% coke content. This difference can adverselyaffect the quality of the product, because high temper-ature should be maintained to guarantee high strength

of the sintered ores. Additionally, as coke contentincreases, the maximum temperature increases morerapidly and the completion temperature of melting isFlame 145 (2006) 447463 455

Fig. 9. Combustion zone thickness, melting zone thickness,and maximum temperature profiles for various coke content.

Fig. 10. FFS and sintering time for various coke con-tentexperimental and computed results.

reached earlier. This indicates that the temperature ishigher in the upper part of the bed. The coke combus-tion zone thickness and melting zone thickness alsoincrease for the higher coke content.

Coke content can also influence the propagationof the combustion zone. Fig. 10 shows calculationand experimental results of FFS and sintering time forvarious coke contents. As coke content increases, thepropagation speed of the combustion zone decreases.The sintering time for the case of 4.2% coke con-tent is approximately 100 s longer than that of thecase of 3.4% coke content. These results are physi-cally reasonable because higher coke content reflectsa lower air supply per unit mass of fuel, which resultsin lower combustion propagation speed. The calcula-tion results show good agreement with the experimen-tal results, thus implying that the model adequatelydescribes combustion characteristics for various cokecontents.

5.3. Effect of air supplyThe air supply rate is represented by the averageair velocity, U , which is set at 0.260.46 m/s. Ba-sically, the amount of combustion air is determinedby the initial pressure difference of the suction fan,

and F456 W. Yang et al. / Combustion

(a)

(b)

Fig. 11. Test results of temperature profiles (a) and gas com-positions (b) for various air velocities.

which is set just after the end of ignition. The amountof air supply varies with time because the particle sizeand porosity of the bed are changing, as shown inFig. 4. The curves of the air inlet velocities for variousinitial pressure differences are estimated and fitted bypolynomials for model calculations.

Fig. 11 shows experimental results for temperatureprofile and gas composition for various air velocities.The temperature profile and gas composition are dra-matically affected by the air supply, as expected. Forlower air velocities, the temperature rise at a givenpoint is significantly delayed, which implies that thecombustion zone propagates more slowly through thesintering bed. From the results of flue gas composi-tion, an increase of CO2 concentration can be foundin the case of higher air velocity, and coke combustionis completed later for the lower air supply. This indi-cates that the fuel consumption rate is increased whenthe combustion propagation becomes faster at higherair velocity. Fig. 12 shows simulation and measure-ment results of the FFS and sintering times for variousair inlet velocities. The simulation results are in close

agreement with the measurement results, demonstrat-ing that the coke combustion rate can be well esti-mated by this numerical model. FFS is proportionallylame 145 (2006) 447463

Fig. 12. FFS and sintering time for various air veloci-tiesexperimental and computed results.

increased along with the air inlet velocity, while thesintering time shows an opposite trend. These para-meters are directly related to the productivity of thesintered ore. Specifically, a high rate of air supply ac-celerates FFS, and then sintering time can be reduced;i.e., the time for the whole process can be saved, andthus productivity can be improved. However, a highrate of air supply requires more fan power and greatercapacities of the flue gas treatment system, which re-quires careful P setting of the suction fan.

Fig. 13a shows the calculated DTCZs at threepoints for various air velocities. The profiles of theDTCZs have different trends for each location withinthe bed. At y = 0.49 m, which represents the upperpart of the bed, the DTCZ values are rapidly de-creased in the velocity range of 0.20.4 m/s, while theprofiles for y = 0.30 m and y = 0.11 m are less dra-matically changed. When the air supply is small, cokeparticles in the upper part of the bed required a longertime to burn out. The lower part of the bed is not sig-nificantly affected by the change in air supply. DTCZprofiles show a decreasing trend as the air supplyis increased, which indicates that the coke combus-tion rate at a given point is accelerated with a largeramount of air. When the average air velocity is over1.0 m/s, the DTCZ value for y = 0.49 m is rapidlydecreased, which means that the convection heat lossbetween the solid particles and the cold air is greaterthan the heat input from the combustion front. AtU = 1.06 m/s, coke combustion cannot be sustainedany longer, and the flame becomes extinct. Fig. 13bshows the FFSs for air velocities. The relationshipbetween combustion propagation and air supply isknown to be proportional when the combustion is lim-ited by oxygen (Regime 1). However, if more air issupplied, combustion propagation is limited by con-vection cooling (Regime 2). In that regime, FFS tendsto be constant or decreased even if the air supply is

increased. Over a limited air supply, combustion can-not be sustained and is extinguished when the con-vection cooling exceeds the amount of combustion

andW. Yang et al. / Combustion

(a)

(b)

Fig. 13. Computed results of DTCZs (a) and FFSs (b) forvarious air velocities.

heat (Regime 3). Previous simulation results [9] fora fixed bed of wood cubes in a radiative environ-ment show these regimes clearly. However, the FFScurve shown in Fig. 13b does not show the character-istics of Regime 2, which are caused by the differenceof particle diameter between the fixed bed (typically20 mm in diameter) and the iron ore sintering bed(typically 23 mm in diameter). Previous research [9]shows that the range of air supply for Regime 2 be-comes broader when the particle size of the fuel isincreased. When the particle size is increased, the to-tal surface area is decreased. The convection coolingeffect should then be decreased, which means thatcombustion can be better maintained for a higher airsupply. In the case of an iron ore sintering bed, therange of Regime 2 does not exist due to the small par-ticle size.

Fig. 14 shows the heat balance terms at y = 0.49 mand y = 0.30 m under extinction conditions of U =1.06 m/s. The level of combustion heat at y = 0.49 mis not high compared with the convective heat trans-fer, which means that the combustion heat is not suf-

ficient to propagate the combustion process down-ward. In the extinction conditions caused by excessiveair, the gas temperature is not high due to convec-Flame 145 (2006) 447463 457

Fig. 14. Heat balance terms at y = 0.49 m and y = 0.30 munder extinction conditions (U = 1.06 m/s).

tion cooling and also because the convective heat in-put from the gas flow is not sufficient to heat fuelparticles up to a temperature of initiating char com-bustion. Consequently, combustion heat generated aty = 0.30 m is dramatically decreased, as shown inFig. 14. From these results, it can be seen that igni-tion conditions and combustion characteristics of theupper part of the bed (that is, the initial stage of theprocess after ignition) affect the whole process sig-nificantly, especially for the features related to self-sustaining combustion.

Fig. 15 shows the details of the phenomena de-scribed above, using CZT and MaxT. While the air ve-locity affects the flame front speed and sintering timesignificantly, combustion zone thickness and meltingzone thickness do not show considerable differencesfor various air suction rates, except that the times forthe profiles to fall to zero reflect the different prop-agation speeds of the combustion front and sinteringtime. These results show that a greater air supply canimprove productivity without compromising the qual-ity of the sintered ores. However, an increased airsupply rate results in a higher amount of combustiongas and thereby requires a larger flue gas treatmentsystem.

From the case studies related with air/fuel ratio, itcan be seen that the newly defined quantitative para-meters can effectively describe the combustion char-acteristics in a solid fuel bed as well as the effects ofthe major operating parameters.

6. Fuel substitution: coal for coke

The study described in this section focuses on ob-taining basic results in the process of setting adequate

operating conditions for using anthracite as a substi-tution fuel. The anthracite used in this study is fromChina and is employed in commercial iron ore sinter-

and F

s, anda basic analysis for the characterization of the fuel,including an elemental analysis, proximate analysis,and calorimetric analysis. The anthracite employedin this study contains 5% volatiles. Coke containsmore fixed carbon and less ash content than an-thracite, which makes the heating value of coke higherthan that of anthracite. The ratio of fixed carbon in thecoke and in anthracite corresponds with the ratio ofthe heating values of coke and of anthracite.

A thermogravimetric analysis was carried out toinvestigate the difference of reactivity of coke andanthracite. Test apparatus was built in house for a rel-atively large sample, typically on the order of 1 g.The heating rate was set to 10 C/min from roomtemperature to 1000 C and was maintained constantthereafter. Nitrogen was used as a carrier gas for con-stant temperature under 1000 C, while air was used

the initial mass fraction of anthracite was set to 3.93%based on the content of fixed carbon. The differenceof internal pores of the two fuels is reflected in thesolid property model for density, specific heat, andsurface area. The initial values of the volume fractionof internal pores are set to 0.1 for the anthracite and0.45 for the coke. These values are derived from themeasurement values of the apparent densities for thecoke and the anthracite.

Fig. 18 shows the calculation results of the heatbalance terms for Anth100. The difference of fuel re-activity is well shown in the profile of the solidgasreaction term, compared with the profile in Fig. 5.Both at y = 0.49 m and y = 0.11 m, the peak valuesof the solidgas reaction heat in the case of Anth100are lower but the reaction time is longer. These fea-tures reflect the difference of reactivity between coke458 W. Yang et al. / Combustion

Fig. 15. Combustion zone thickness, melting zone thicknes

Table 6Results of elemental and proximate analysis for coke andanthracite

Coke AnthraciteElemental analysis (%)

C 87.86 86.04H 0.63 1.34O 9.42 11.85N 1.55 0.56S 0.54 0.21

Proximate analysis (%, dry base)Volatile 2.26 4.99Fixed carbon 83.75 79.69Ash 13.99 15.32

Heating value (kJ/kg) 26352 24999

ing processes in Korea. Table 6 shows the results offor constant temperature of 1000 C. Flow rate of thecarrier gases was set to 10 l/min. Fig. 16 presentsthe results of the thermogravimetric analysis for cokelame 145 (2006) 447463

maximum temperature profiles for various air velocities.

and anthracite in the char oxidation environment. Thethermogravimetric curves show that the mass reduc-tion of coke by char combustion is faster than thatof anthracite. From the differential thermogravimetriccurves for the two kinds of fuel, a meaningful differ-ence of char reactivity can be found. The reactivityof anthracite is lower than that of coke, which can af-fect the combustion conditions in the sintering bed.The difference of reactivity is attributed to the differ-ence of true surface area as well as intrinsic surfacereactivity [19]. As shown in Fig. 17, coke has manyinternal pores, while the pores of the anthracite usedin this study were not distinctively found. Examina-tion of the quantitative relationship between the poresizes and the true surface area is beyond the scope ofthis study. The influence of the different fuel charac-teristics on the combustion conditions in the sinteringbed was investigated in this model study. In the test,and anthracite. In Anth100, compared with Coke100,the combustion propagation is retarded due to slowgeneration of combustion heat. As shown in Table 7,

andFig. 16. Results of thermogravimetric analysis for coke and anthracite.

(a)

(b)

Fig. 17. SEM photograph of the anthracite (a) and coke (b).W. Yang et al. / CombustionFFS and sintering time represent the difference infuel, which can significantly affect the productivity.Fig. 19a shows the temperature profiles in the upperFlame 145 (2006) 447463 459part (y = 0.49 m) and the lower part (y = 0.11 m)of the bed, and a delay of temperature rise at y =0.11 m in case of Anth100 appears. Another feature

(a)(b)

Fig. 19. Calculation results for coke and anthracite:460 W. Yang et al. / Combustion and Flame 145 (2006) 447463

Fig. 18. Heat balance terms at y = 0.49 m and y = 0.11 mfor coke and anthracite.

for Anth100 shown in the MaxT profile in Fig. 19a islower MaxT in the time range of 100400 s. This im-plies that the maximum temperature of the region nearthe top of the bed is considerably low. This indicatesthe possibility of not obtaining sufficiently high tem-perature for the melting of the iron ores near the top of

Table 7FFS and sintering time for Coke100 and Anth100

Case name FFS (cm/min) Sintering time (s)Coke100 2.98 1210Anth100 2.76 1296

the bed, in which case the quality of the sintered oreswould likely be deteriorated. Meanwhile, the lowercombustion rate of the anthracite results in a differ-ent CZT profile, as shown in Fig. 19b. CZT valuesfor Anth100 are lower than those for Coke100 duringthe early stage of the process (t = 100300 s), sincethe effect of convection heat loss by air is increasedand the combustion rate is lower. During the laterstage of the process, however, the combustion zonebecomes broader. However, MZT values for Coke100are larger than those for Anth100. This appears tobe related to the convective cooling effect; a greateramount of air is introduced when anthracite combus-tion is completed, because the completion time is de-layed compared to the case of Coke100. Moreover,the melting zone does not appear during the earlier(a) temperature profiles, (b) CZT and MZTs.

andW. Yang et al. / Combustion

Fig. 20. Heat balance terms at y = 0.49 m and y = 0.11 mfor Anth1.2.

stage of the process for Anth100, suggesting a pos-sibility of deficiency in forming sintered ores in theupper part of the bed. Considering that the valuesof MaxT for Anth100 are not larger than those forCoke100, thermal concentration appears to be unfa-vorable for Anth100. In conclusion, the combustioncharacteristics of anthracite as a substitution fuel arenot as favorable as coke for the sintering process.

However, the low reactivity of the anthracite canbe improved by decreasing the particle size and thusby increasing the volumetric surface area. In the charreaction model used in this study, the char combus-tion rate is increased proportionally with the apparentsurface area of the char particles. Fig. 20 shows theprofiles of the heat balance terms for Anth1.2. Theprofiles of solidgas reactions become similar to thoseof Coke100 (Coke3.8) in Fig. 5 while the profiles aty = 0.11 m are advanced, which means that combus-tion propagation becomes faster. This is also shown inFig. 21a, which shows that points where temperatureis rising are meaningfully advanced when smaller par-ticles of anthracite are used. Fig. 21a also shows theMaxT profile for various diameters of anthracite (1.2,1.6, and 2.0 mm) and its values in the time range of100400 s are increased. From these results, it can beseen that the productivity and quality of the sinteredores can be improved by using smaller particles of an-thracite. The thickness profiles (Fig. 21b) for variousdiameters of anthracite (1.2, 1.6, and 2.0 mm) alsoshow that the CZT can be decreased to a level corre-sponding to the use of coke by reducing the anthraciteparticle size. This trend is similarly observed whena highly reactive fuel is used, which implies that thedifference in fuel reactivity can be overcome by con-trolling the particle size of the fuel.

FFS and sintering time for the cases of variousparticle sizes are shown in Fig. 22. Sintering time is

shown to be shortened by approximately 100 s whenthe particle sizes of the anthracite are reduced from2.0 to 1.2 mm. This indicates the possibility of main-Flame 145 (2006) 447463 461

taining the same level of productivity when a portionof coke is replaced by anthracite by reducing the sizeof the particles.

7. Conclusion

A quantitative description of combustion charac-teristics in an iron ore sintering process is presented,based on sintering pot tests and a mathematical modelof solid bed combustion. Characterization of the bedcombustion is attempted through the introduction ofnewly defined parameters: duration time in the com-bustion zone (DTCZ), combustion zone thickness(CZT), melting zone thickness (MZT), and maximumtemperature (MaxT). Flame front speed (FFS) andsintering time (ST), well-known parameters for de-scribing the propagation of solid fuel combustion ina fixed bed, are also assessed. They can be obtainedby the mathematical model and sintering pot test.These parameters can represent the effects of the twofactors dominating solid bed combustion: operatingconditions relating to combustion and intrinsic fuelproperties. The effects of operating parameterscokecontents in the bed material and air supply rateon the combustion characteristics and the quantitativeparameters are investigated. It is found that fixed-bedcombustion can be well characterized by the quanti-fied parameters based on distributions of solid tem-perature and mass fraction of the char content in thebed. Effects of the intrinsic fuel properties are inves-tigated numerically for fuel substitution from coke toanthracite and various particle sizes of anthracite. Thefollowing conclusion can be derived from this study.

1. The dominating mode of heat transfer withinan iron ore sintering bed is convection between gasflow and solid material. Raw fuel is heated mainly byconvection in the initial stage, and char (or coke) com-bustion after the fuel is heated to the ignition temper-ature. The combustion rate of the fuel determines thecombustion propagation speed and the characteristicsof the combustion zone such as thickness and tem-perature. The effect of radiation is not as significantwith respect to the combustion process, in contrast toa fixed bed of wood cubes of large particle size ignitedby radiation.

2. Values of FFS and ST show the effect of airsupply on the whole combustion process. Because ofthe small particle sizes in the iron ore sintering bed,there was no region where the combustion propaga-tion speed was not affected by air supply. FFS profilesshow that combustion cannot be sustained due to ex-cessive convection cooling when air supply exceeds

a limiting point. In particular, the combustion condi-tions in the upper part of the bed (upstream of the gasflow) are very sensitive to the operating conditions,

and F

(a)(b)

Fig. 21. Calculation results for various diameters of anthracite: (a) temperature profiles, (b) CZT and MZTs.

Fig. 22. FFS and sintering time for various particle sizes ofanthracite.

while those in the lower part of the bed are not sensi-tive due to heat accumulation.

3. Values of CZT, MZT, DTCZ, and MaxT re-flect the effects of parameters related to fuel contentand characteristics. Since the quantitative parametersshould be carefully controlled for the quality of the

4. The effect of fuel reactivity is seen in the propa-gation speed of the combustion front and the tempera-ture in the upper part of the bed, which were reflectedin the results of FFS, ST, and MaxT. The anthraciteshowed slower FFS and lower temperature in the up-per part of the sintering bed, which would adverselyaffect the productivity and quality of the sintered ores.CZT values in the lower part of the bed were signif-icantly increased due to the longer combustion timefor anthracite. Case studies for various particle sizesof anthracite, however, show that the combustion ratecan be improved by decreasing the particle size.

Acknowledgments

This study has been financially supported byPOSCO and an international collaborative researchprogram funded by the Korea Ministry of Science and462 W. Yang et al. / Combustionproduct, selection of fuel and its content is crucial toeffective operation. The approach presented here canprovide quantitative criteria.lame 145 (2006) 447463Technology. The authors also acknowledge supportfrom the Combustion Engineering Research Center(CERC).

andW. Yang et al. / Combustion

References

[1] I. Muchi, J. Higuchi, Testu-to-Hagane 56 (1970) 371381.

[2] R.W. Young, Ironmak. Steelmak. 7 (1977) 321328.[3] M.J. Cumming, J.A. Thurlby, Ironmak. Steelmak. 17

(1990) 245254.[4] F. Patisson, J.P. Bellot, D. Ablitzer, E. Marliere, C.

Dulcy, J.M. Steiler, Ironmak. Steelmak. 18 (1991) 8995.

[5] N. Nath, J. Da Silva, N. Chakraborti, Steel Res. 68(1997) 285292.

[6] M.V. Ramos, E. Kasai, J. Kano, T. Nakamura, ISIJInt. 40 (2000) 448454.

[7] W. Yang, C. Ryu, S. Choi, ISIJ Int. 44 (2004) 492499.[8] Y.R. Goh, R.G. Siddall, V. Nasserzaheh, R. Zakaria,

J. Swithenbank, D. Lawrence, N. Garrod, B. Jones,J. Inst. Energy 71 (1998) 110118.Flame 145 (2006) 447463 463

[9] D. Shin, S. Choi, Combust. Flame 121 (2002) 167180.[10] C. Ryu, D. Shin, S. Choi, J. Air Waste Manage. 52

(2002) 189197.[11] B. Peters, Combust. Flame 131 (2002) 132146.[12] M.L. Hobbs, P.T. Radulovic, L.D. Smoot, AIChE J. 38

(1992) 681702.[13] P.T. Radulovic, M.U. Ghani, L.D. Smoot, Fuel 74

(1995) 582594.[14] M.U. Ghani, T. Predrag, P.T. Radulovic, L.D. Smoot,

Fuel 75 (1996) 12131226.[15] C.D. Blasi, Chem. Eng. Sci. 55 (2000) 29312944.[16] A. Galgano, C.D. Blasi, Combust. Flame 139 (2004)

1627.[17] J. Cooper, W.L.H. Hallett, Chem. Eng. Sci. 55 (2000)

44514460.[18] J.C. Wurzenberger, S.W. Wallner, H. Raupenstrauch,

J.G. Khinast, AIChE J. 48 (2002) 23982411.[19] I.W. Smith, Proc. Combust. Inst. 19 (1982) 10451065.

Combustion characteristics in an iron ore sintering bed-evaluation of fuel substitutionIntroductionCharacterization of bed combustionExperimentsModel studyEffects of coke content and air supply rateReference caseEffect of coke contentEffect of air supply

Fuel substitution: coal for cokeConclusionAcknowledgmentsReferences