Numerical investigation of the effects of size segregation on pulverized coal combustion in a blast furnace

Accepted Manuscript Numerical investigation of the effects of size segregation on pulverized coal combustion in a blast furnace

Dongling Wu, Ping Zhou, Hongjie Yan, Pengyu Shi, Chenn Q. Zhou PII: DOI: Reference:

S0032-5910(18)30795-2 doi:10.1016/j.powtec.2018.09.067 PTEC 13741

To appear in:

Powder Technology

Received date: Revised date: Accepted date:

13 April 2018 21 September 2018 22 September 2018

Please cite this article as: Dongling Wu, Ping Zhou, Hongjie Yan, Pengyu Shi, Chenn Q. Zhou , Numerical investigation of the effects of size segregation on pulverized coal combustion in a blast furnace. Ptec (2018), doi:10.1016/j.powtec.2018.09.067

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Numerical investigation of the effects of size segregation on pulverized coal combustion in a blast furnace Dongling Wu1 , Ping Zhou1 , Hongjie Yan1 , Pengyu Shi1 , Chenn Q. Zhou1 ,2 1

2

School of Energy Science and Engineering, Central South University, Changsha 410083, China

Center for Innovation through Visualization and Simulation, Purdue University Northwest, 2200 169th Street, Hammond, IN 46323, USA

PT

Abstract: Pulverized coal combustion in a blast furnace raceway can play a critical role in blast furnace iron-making operations. This research study integrated the particle combustion and the

RI

raceway formation processes using a refurbished coupling method based on computational fluid

SC

dynamics (CFD). The CFD model was validated, and there was good agreement between the simulation results and the two sets of experimental data. The combustion behaviors of pulverized

NU

coal particles in an industrial blast furnace were also simulated over a wide size range. The simulation results showed that size segregation occurred along the vertical direction in the

MA

raceway. Along the coal plume, the effectiveness of the size segregation process can be significantly reduced by the turbulence generated owing to the existence of the lance, the

ED

expansion of the gas jet near the tuyere tip, and the gas recirculation at the front end of the raceway. The burnout rate of particles smaller than 60 µm was also shown to be sensitive to the

EP T

degree of weakening of the size segregation. The particle distribution in the coke bed indicated that the group of particles with diameters equal to 52.5 µm was associated with the largest proportion of unburnt char. This research extended the applicability of the existing numerical

AC C

methods and provided a better understanding of the pulverized coal injection (PCI) process from the viewpoint of size segregation, which is beneficial to the further optimization of PCI usage. Keywords: Pulverized coal combustion; Raceway formation; Size segregation; Blast furnace

Corresponding author: Ping Zhou Email: [email protected] Tel./Fax: +86–731–88879863

1

ACCEPTED MANUSCRIPT Highlights 

The process of coal particle combustion has been coupled with the raceway formation process. Size segregation characteristics and their evolution in the plume have been revealed.



The effects of size segregation on particle combustion performance have been investigated.

AC C

EP T

ED

MA

NU

SC

RI

PT



2

ACCEPTED MANUSCRIPT Nomenclature A

k kij

empirical constant, 4.0 specific surface area of coke particle, m2 /kg surface area of the particle, m2 empirical constant, 0.5 mass diffusion-limited rate constant, 5e-12 kinetics-limited rate pre-exponential factor, 450 drag coefficient 3 molar concentration of species i, mol/m diffusion coefficient of vapor in bulk, m2 /s 2 diffusion rate coefficient of oxygen, m /s mass transfer coefficient of species i, kg/(m·s) mean diameter of pulverized coal particles, µm diameter of coal particle, m diameter of coke particle, m particle diameter corresponding to the maximum unburnt char content, µm activation energy, J/kmol maximum relative error drag function mass transfer coefficient in the reaction of coke gasification, kg/(m2 ·s) chemical rate constant for jth heterogeneous reaction, m3 /(kg·s) 2 2 turbulence kinetic energy, m /s exchange coefficient for phase i and j

kj

rate constant of the reaction,1/s

mp

mass of particle, kg

mdi，0

original mass of a particle with a diameter of d i

mdi

mass of a particle with a diameter of d i after combustion

mdi _ unburnt _ char

unburnt char mass of particle group with a diameter of d i

M w ,i

molecular weight of species i, kg/mol

a Ap

B

C1 C2 CD Ci

D D0

km , j

M w, R M w, j

pox p

molecular weight of a reactant, kg/mol molecular weight of species j, kg/mol number of chemical species coal particle number partial pressure of oxidant species in the gas surrounding the combusting particle, Pa pressure, Pa

AC C

N np

SC

kf,j

NU

f

MA

E er _ max

ED

dub

EP T

d

dp dc

RI

PT

Di

pdi

percentage of unburnt char mass of group of particles with diameter d i

qp

heat source transferred from a particle, W

Re p

Reynolds number of particle

Rk

kinetic rate, s

R i ,r

net rate of production of species, kg/(m3 ·s)

Rj

T

overall reaction rate of the jth reaction during coke combustion, mol/(m3 s) Sherwood number Schmidt number temperature of continuous phase, K

Tp

particle temperature, K

Sh Sc

-1

3

ACCEPTED MANUSCRIPT Tm

average temperature of gas and coke, K

Ug

velocity of gas phase, m/s

Up

velocity of particle phase, m/s

vi' , r

stoichiometric coefficient for reactant i in the rth reaction

v' ,r

stoichiometric coefficient for a reactant in the rth reaction volatile matter of coal

Yd

mass fraction of particle

YP

mass fraction of any product species

Y

mass fraction of a particular reactant

Yi , s

vapor mass fraction at the surface

Yi ,

vapor mass fraction in the bulk gas

SC

Greek symbols

PT

stoichiometric coefficient for product j in rth reaction

VM

RI

v

" j ,r

1， 2  fixed _ carbon p

volatile yields

d p

char burnout rate of a particle with a diameter d i

p

NU

porosity of particle

MA

i

fraction of fixed carbon of a coal particle after combustion

volume fraction of particle phase

1  2

volatile yield factors

i

density of coal particle, kg/m size factor of coke particle

3

effectiveness factor of catalytic reaction

AC C



bulk density of coke, kg/m3

EP T

bc p

ED

g

surface strain tensor of particle phase porosity of coke bed gas dynamic viscosity, Pa·s



4

ACCEPTED MANUSCRIPT 1 Introduction The blast furnace is a crucial component in integrated steel mills. It uses blast gas and coke to provide heat and then converts iron ore into liquid iron in a huge vertical furnace. This conversion is an energy- and capital-intensive process. The iron ore and coke are charged from the top, and the hot blast is supplied by a blowpipe in the lower portion of the furnace. To save energy and

PT

reduce CO2 emissions, the technology of pulverized coal injection (PCI) through the blowpipe has been used to reduce coke consumption, and is now extensively used due to the accessibility and

RI

cheap production costs of coal [1]. Research efforts have been expended to understand the

SC

fundamental aspects of pulverized coal injection and to improve coal combustion performance. Owing to an extremely harsh operating environment and the difficulties in making direct

NU

measurements in a blast furnace, numerical simulations have been identified and proven as a powerful tool to provide fundamental understandings and guidelines for optimizing PCI in a blast

MA

furnace [2-14].

A significant number of simulations have been performed for both model development and applications purposes. These simulations can be classified into three categories: 1) description of

ED

pulverized coal and coke combustion behaviors and prediction of the combustion efficiency of injected fuels, 2) prediction of the shape of the main combustion zone inside a blast furnace, i.e.,

EP T

the raceway, and 3) prediction of combustion efficiency of injected fuels. For the first research category, three-dimensional models were developed during the past decade

AC C

and were extensively used to simulate combustion behaviors. Shen et al. [4, 5] used an Eulerian– Lagrangian method to study pulverized coal combustion in two types of assumed raceway shapes. Yeh et al. [6] also developed a three-dimensional model to simulate coal particle combustion inside a raceway using numerical predictions. However, coke particle combustion in the region outside the raceway was not included in that simulated model. Additionally, some numerical models were developed to simulate the coke combustion behaviors by treating the region outside the raceway that was packed with coke particles as a coke bed and by describing it mathematically as a porous medium region. Shen et al. [9] extended the previous three-dimensional Lagrangian– Eulerian model to allow it to simulate coke particle combustion and multiphase flow in the tuyere– raceway–coke bed zone. Zhou et al. [11, 14] also established a three-dimensional CFD model with 5

ACCEPTED MANUSCRIPT the Eulerian–Eulerian method to simulate the gas–coal particle–coke particle flow and combustion in the tuyere–raceway–coke bed zone. In practical blast furnace operations, the coke particles are packed together in the lower part of the furnace, and hence the motion and combustion behavior of a single coke particle will be affected by the particles surrounding it. The model of porous media that is based on the continuum assumption is not capable of describing the interactions between single coke particles. The discrete element method (DEM) is the most extensively applied and

PT

promising method to simulate inter-particle interactions. Nogami et al. [15] developed a DEM

RI

model to describe the interactions between coke particles. Combining the DEM with a CFD model, the gas–coal particle–coke particle interaction was also simulated in an experimental-scale blast

SC

furnace.

NU

The raceway shapes used in different simulations can vary within a large range. Accordingly, gas– particle flow and particle combustion in different raceway shapes exhibit different characteristics

MA

[4-7, 9-11, 14, 15]. Therefore, the modeling of raceway formation has received increased attention [16-20]. It was found that the raceway formation is mainly attributed to the interaction between coke particles and blast gas [21-23], which is often modeled with the CFD–DEM method. This

ED

method has also been extensively employed by many researchers [24-26] to simulate other particle-related processes in a blast furnace. The pure CFD-based raceway formation model [6, 11,

EP T

14, 18, 19] was established by treating the gas and coke particle phases mathematically as interpenetrating continua. Both of the two continuous phases were then described with the

AC C

Eulerian approach, which has been alternatively referred to as a two-fluid method. In fact, except for the interaction between the coke particles and blast gas, the combustion behaviors of pulverized coal and coke also affect the raceway formation in the way of changing the inflowing gas momentum. [21]. Depending on whether the effects of combustion on the raceway formation were considered or not, the two-fluid method can be employed in two different ways, based on either one-way coupling or two-way coupling methods. The combustion effect on the raceway formation was considered only in the case of the two-way coupling method. The two-way coupling method integrated the two research categories mentioned above. Therefore, it is considered to be a more realistic method. Developed by Zhou et al. [11, 14], the two-way coupling method turns out to be very useful for industrial PCI optimization. 6

ACCEPTED MANUSCRIPT Although DEM is a promising tool in handling problems in many particulate flow systems [16– 20], it has a limited ability to address the complex chemical reactions in an industrial multiphase flow system, like an industrial blast furnace. This is because numerous particles are charged into the furnace, and the size of the coke particle is very small. The particle time-step in DEM then needs to be small enough so that reliable simulation results can be obtained. Therefore, a DEM simulation of an industrial particulate flow system can be extremely time-consuming. In contrast,

PT

based on the continuum assumption, CFD has turned out to be an effective and more practical

RI

method at present in the field of simulation of pulverized coal and coke combustion in an industrial blast furnace [13].

SC

In our previous CFD simulations [14], the Eulerian method was used to simulate the behaviors of

NU

pulverized coal particles with uniform sizes, but the feasibility of the two-way coupling method was limited when the combustion behaviors of different-sized coal partic les were simulated. An

MA

evident and interesting combustion phenomenon associated with the motion of coal particles of different sizes in the raceway is size segregation, which has been reported in some publications [3, 4]. In these simulations, coal particles were modeled with the Lagrangian method. Owing to the

ED

size segregation, particles located at different positions have different combustion characteristics in the raceway, and the subsequent unburnt coal particle distribution in the coke bed can be

EP T

affected. However, these characteristics had not been analyzed in-depth in the aforementioned simulations [3, 4], while total PC burnout rates in an assumed raceway, or at the end-point of the

AC C

assumed raceway, were used to evaluate the combustion performance. These burnout rates are not capable of revealing particle combustion behaviors related to size segregation. For this reason, further detailed analyses are needed. To more fully investigate the size segregation in a realistic raceway, in this research study, the two-way coupling method in our previous report [14] was refurbished with the coal particle phase tracked by the Lagrangian method. In addition, detailed characteristics of the size segregation were revealed in the study. Then, the effects that size segregation can have on particle combustion efficiency and the unburnt particle distribution in the coke bed were investigated. 2 CFD models 2.1 Geometrical model 7

ACCEPTED MANUSCRIPT The geometrical model of the blast furnace in this study was built based on a commercial blast furnace in China. The computational domain is delineated by the dashed line shown in Figure 1(a). The principal dimensions of the lower part of the blast furnace are listed in Figure 1(b) and Table 1. The three-dimensional computational domain containing one tuyere is shown in Figure 1(c). In the lower part of the furnace, plenty of coke particles are packed together. With the lateral injection of high-speed gas into the packed bed of particles, a raceway can be formed within the

PT

bed near the nozzle [24]. At this stage, the raceway shape is important and is decided by the flow

RI

behaviors of gas-coke particles in a packed bed [17]. At a specific gas inflow velocity, the shape of the raceway can be maintained with a limited number of particles near the moving raceway

SC

boundaries [15, 24]. For this reason, when simulating the combustion of pulverized coal at the

NU

second stage, the raceway can be regarded as a fixed cavity and the packed bed beyond the raceway as a porous medium.

MA

This indicates that the PCI process in the lower part of a blast furnace can be separated into two processes. The first is the raceway formation process, and the second is the coal combustion process.

ED

Although the raceway formation process and the coal combustion process are simulated separately, the interactions between the two processes are important. Therefore, the two processes are

EP T

integrated with a coupling method. Details of the corresponding mathematical models for the different processes and the method used to couple the two processes are given in the following

AC C

sections.

2.2 Mathematical model of raceway formation process The coke bed is treated as a fluidized bed filled with coke particles during the raceway formation process. The combustion is not incorporated in this simulation. The transient three-dimensional approach is used to describe the multiphase flow in the fluidized bed, and the continuity equation of the gas and particle phase, and the conservation of momentum for a phase i (i = g (gas); i = p (particle)) can be expressed in accordance to Eqs. (1)–(3).

   g  gU g  t

     g  gU g   S gcombustion

8

(1)

ACCEPTED MANUSCRIPT    p  pU t

  p



p

 pU

S p

c o m b u s t i o n p

(2)

   i  iU i      i iUiUi    i p    i  i i g  S t 18i  g f S  ki jU i U ， i  ; jkij  j dp

(4)

 g and  p are the volume fractions of the gas and coke particle phases, respectively.

PT

where

(3)

RI

The mass source term of the gaseous continuity equation, S gcombustion , is equal to the total mass loss of both coal and coke particles due to the devolatilization and heterogenous combustion

SC

reactions and is obtained from the simulation of the particle combustion process. The variable

NU

S pcombustion is the mass loss of the coke particles due to the combustion reactions. It is relatively much less than the total mass of the packed coke particles in the lower part of the blast furnace.

MA

Furthermore, the consumed coke particles can be supplemented by those falling from the upper part of the blast furnace. Therefore, the effects of the mass loss on the motion of coke particles in

ED

this study are neglected. This means that S pcombustion  0 . Equivalently, kij is the exchange coefficient for the gas and coke particle phases. The momentum exchange f

between the two

EP T

phases is demonstrated by the O’Brien model [27] in which an empirical relationship determines the gas–coke particle interaction coefficient based on the particle terminal veloc ity measured in the fluidized and settling beds. The thermophysical properties of the gaseous phase used in Eqs (1)

AC C

and (3) are decided by the gas temperature, which is obtained by the combustion model given in the following section.

Additionally, the particle–particle interactions are described by the assumption that there are instantaneous binary collis ions between the coke particles and the fact that energy is dissipated owing to the inelastic nature of the collisions. 2.3 Mathematical model of the particle combustion process The mathematical model was established using CFD as based on the following assumptions. (1) The flows of pig iron and slag in the coke bed, and the break-up or coalescence of coke particles are not included 9

ACCEPTED MANUSCRIPT (2) Ash melting and aggregation are not considered 2.3.1 Gas–coal particle flow model The volume fraction of pulverized coal particles is approximately 2%. The Eulerian–Lagrangian method is used to simulate the gas–coal particle flow in the lower part of the blast furnace. For the gas phase, a 3-D, steady-state, realizable k–ε turbulence model was used. For the particle phase, since the volume fraction of pulverized coal in the blast was approximately 0.01%, turbulent

PT

dispersion effects were considered using a stochastic tracking approach. Governing equations for

RI

the gas–coal particle flow can be found in the literature [10].

When simulating the coal combustion process, the raceway shape has a constant profile

SC

determined by the simulation results of the raceway formation process. The coke bed that extends beyond the raceway, as mentioned in the geometry model, is treated as a porous medium. Gas flow

NU

resistances are calculated with Ergun’s equation [28].

MA

2.3.2 Coal combustion model

The combustion process for coal particles is summarized and simplified to a four-stage process, namely, moisture evaporation, devolatilization of raw coal, combustion of volatile matter, and the

ED

oxidation of the residual char. The reaction rate [29-34] for each stage is listed in Table 2. 2.3.3 Coke combustion model

EP T

For coke combustion, the C-O, Boudouard, and C-H2 O reactions were taken into account, but the water shift reaction (CO-H2 O) was neglected since the C-H2 O reaction dominated the higher

AC C

temperature region. The rates of the heterogeneous reactions R j ( j  1, 2, 3) were explored and determined to be first-order irreversible [35]. These reaction rates are calculated by the following equation, and the related parameters are listed in Table 3.

R j  k j Ci kj 

(5)

1 1



1

k f , j a  km, j bc kf ,j 

Di Sh  dp

Sh  1.5Re0.55 p

(6)

(7) (8)

where R j is the overall reaction rate of the jth reaction during coke combustion, k j is the rate 10

ACCEPTED MANUSCRIPT constant of the reaction, and k f , j and km , j are the mass transfer coefficient and chemical rate constant for the jth heterogeneous reaction, respectively. 2.4 Coupling of the raceway formation and coal combustion processes The coupling of the raceway formation and the coal combustion process was realized by means of data exchanging. Specifically, the distribution of the coke particle volume fraction obtained by the

PT

raceway formation model decides the geometric profile of the main combustion region (i.e., the raceway). On the other hand, the generated additional gaseous mass and the temperature rising

RI

owing to the coal and coke particle combustion can change the continuity equation and the

procedures of data exchange are given as follows.

SC

thermophysical properties of the gaseous phase used in the raceway formation model. The detailed

NU

After the combustion process in the initial raceway shape (obtained without consideration of the effects of the combustion reactions) was simulated, the particle mass loss and gaseous temperature

MA

were then stored at each cell of the computational domain. When the raceway formation process was simulated once again, the mass loss of coal and coke particles was treated as an additional source term of the continuity equation of the gas phase, and the thermophysical properties of the

ED

gaseous phase were recalculated based on the stored gaseous temperature. Then an updated coke

EP T

particle distribution was obtained, which is equivalent to an updated raceway shape. The simulation of the particle combustion process was re-carried out using the geometric model that includes the updated raceway. In this way, the simulation data was exchanged between the two

AC C

models. Based on the data exchange, the simulation of the particle combustion and raceway formation process continued in an alternated way. This alternated simulation process would not stop until the differences between the update and the previous results of the main dimens ions of the raceway shape were small enough. These procedures were realized on the platform of ANSYS–FLUENT with user-defined functions (UDFs). The detailed steps are shown in Figure 2. 3 Model validations Two sets of experimental data were used for validation: 1) a hot blast furnace model [15] and 2) a pilot-scale coal combustion test [36]. The operating conditions and coal properties in these experiments were used in CFD simulations. Figure 3(a) shows the comparisons between the CFD results and data from the hot blast furnace 11

ACCEPTED MANUSCRIPT model [15]. The molar fractions of species along the centerline of the blowpipe, as predicted by CFD, were in good agreement with the experimental data. The maximum relative error ( er _ max ) for each species was calculated. The maximum er _ max among all of the measured species was 3.86% for oxygen and 19.51% for hydrogen, respectively. The raceway depth obtained by the validation simulation was 0.375 m. Compared to the measured depth of 0.41 m, the relative error

PT

was 8.54%, which was acceptable from the engineering viewpoint. Figure 3(b) compares the CFD results with a set of pilot-scale coal combustion test data [36]. This

RI

combustion experiment was conducted in the blast furnace hot model by injecting pulverized coal

SC

into the hot air blast that flowed through a refractory-lined blowpipe and tuyere. The agreement between predicted and measured gas temperatures along the blowpipe and tuyere was within

NU

acceptable ranges. 4 Simulation conditions

MA

Table 4 lists the boundary and operating conditions of the raceway formation and the coal combustion process. The deadman zone is treated as impermeable and is not included in the

ED

computational domain, as shown in Figure 1. The boundary between the coke bed and the deadman zone in the following two processes is treated as a wall.

EP T

4.1 Boundary conditions of raceway formation simulation The diameters of the coke particles were set to 0.036 m according to the experimental results of Gupta et al.[37], where coke samples were obtained from the tuyere level of a blast furnace

AC C

through tuyere drilling. The initial coke volume fraction was 0.5. Except for the inlet, outlet, and the walls of the furnace, other boundaries were set as symmetrical boundaries. It needs to be pointed out that no coal particle was injected into the computational domain when simulating the raceway formation process. 4.2 Boundary conditions of particle combustion simulation The final coke distribution obtained from the simulation of the raceway formation process is shown in Figure 4(a). The volume fraction of coke particles in most of the regions of the blast furnace is equal to 0.6. The raceway shape was defined by the profile of the coke particle volume fraction of 0.6. Figure 4(b) and (c) show that the depth of the raceway is 21.7% of the radius of the 12

ACCEPTED MANUSCRIPT hearth. Moreover, the zone above the centerline is larger and occupies 83% of the volume of the raceway. The predicted raceway shape profile is in accordance with the experimental results [38], which also indicated that the region from the tuyere tip to the middle part of the raceway had a high-void fraction followed by a linear decrease of voidage toward the raceway boundary. The setting of the boundaries of the computational domain was the same as that in the raceway formation process.

PT

The porosity of the coke bed was isotropic and was empirically set to 0.4 [14]. Proximate and

RI

ultimate analyses of the pulverized coal are summarized in Table 5. The specific size distribution, which is measured by MASTERSIZER 2000, is represented by the Rosin–Rammler method and is

SC

shown in Figure 5.

NU

5. Results and discussion 5.1 Gas–particle flow characteristics

MA

5.1.1 General gas–particle flow pattern in the predicted raceway The gas flow velocity contour is presented in Figure 6(a). It can be determined that the gas jet gets

ED

separated by the carrier gas and most of the blast gas is recirculated in the raceway. The gas flow streamlines in the calculated raceway and surrounding coke bed are shown in Figure 6(b). It is

EP T

clear that the primary flow stream separates into two streams near the end of the raceway, forming two recirculation zones. The larger one is located above the tuyere centerline with the gas flowing in a counter-clockwise direction, and the other is below the centerline where gas flows in a

AC C

clockwise direction.

As coal particles are released from the coal lance tip, they disperse in the blowpipe and mix with the hot blast gas, thus forming the main coal plume, as shown in Figure 7(b). Along the tuyere’s axial direction, as the coal plume moves forward, it gradually expands with more small particles spreading out in a larger space. Moreover, most of these small particles follow the gas streamline and recirculate in the raceway for a long time. Particles with diameters larger than 120 µm are more likely to escape through the forehead of the raceway and continue to move forward along a straight path in the coke bed. The volume fraction of pulverized coal particles in the blowpipe is shown in Figure 7(b). The evolution of the volume fraction along the coal plume centerline is 13

ACCEPTED MANUSCRIPT plotted in Figure 7(c). It is obvious that the volume fraction experiences a sharp decline from 16.5% to 2% within a short distance of approximately 0.05 m once released into the blowpipe from the lance tip. When particles continue to move forward from the tuyere to the raceway, the volume fraction decreases abruptly at the tuyere tip, as shown in Figure 7(c). This is contributed by the sudden expansion of the coal plume once it enters into a larger space. From the trajectories of particles in Figure 7(a), it can also be observed that some large particles

PT

can even reach the boundary between the coke bed and the deadman zone. The relationship

RI

between the particle diameter and the percentage of coal particles that reach the coke bed– deadman zone boundary is shown in Figure 7(d). The result indicates that when the particle

SC

diameter is larger than 130 µm, the probability that a particle reaches the boundary is larger than

NU

50%.

The gas–particle flow pattern referred to above exhibits both similarities and differences with

MA

those in other simulation studies. For example, the two recirculation zones can also be obtained by CFD–DEM simulations [24, 25]. In contrast, for most of the CFD investigations, the symmetrical flow pattern about the centerline of the tuyere exists in a divergent tube-like raceway [4, 5], and

ED

only one large recirculation zone above the tuyere centerline was formed in a balloon-like raceway shape [2, 9, 12]. In our previous CFD research [14], there was one recirculation zone below the

EP T

centerline of the tuyere. The main reason for the nonexistence of the eddy above the centerline was the large voidage of the region near the tuyere tip. Diffusion of a particle phase near the wall

AC C

in a dilute flow was overpredicted owing to the existence of numerical diffusion upon the use of the Eulerian method for modeling coal particles [39]. This contributed to advanced coal particle combustion, which in turn led to the expansion of the raceway shape near the tuyere tip. 5.1.2 Size segregation in the coal plume Four cross-sections of the coal plume along the tuyere centerline are selected to examine the evolution of particle distribution. The first slice was the tuyere tip, and the remaining three were in the raceway and were located at X= -0.05 m, -0.5 m, and -0.8 m. The detailed particle size distributions at these cross-sections are shown in Figure 8. The positions of the cross-sections are depicted by the dashed lines in Figure 6(b). At the tuyere tip, as shown in Figure 8(a) and Figure 8(c), most of the coal particles are 14

ACCEPTED MANUSCRIPT concentrated below the centerline of the tuyere (Z= 0 m). Small particles flow at the outer surface of the plume, and large particles flow at the bottom, while the medium particles are concentrated in the center of the plume. This relationship between the particle position and diameter demonstrates a stratification distribution of the coal particles in the coal plume, namely their size segregation. This effect can also be demonstrated by the relationship between the average particle position and particle diameter, as shown in Figure 8(c). As the coal plume moves forward, more

PT

particles disperse into the upper and lower recirculation zones, thus leading to weakened size

RI

segregation in the downstream of the coal plume. This weakened downstream segregation can be illustrated by the increased standard deviations of particle positions in Figure 8(d).

SC

Figure 8(c) and (d) show that the size segregation in the coal plume exhibits different

NU

characteristics along the tuyere centerline. Once particles enter into the raceway and move forward for a distance of 0.05 m, the variation in the average position of the particles is small. However,

MA

the degree of size segregation is weakened significantly, as demonstrated by the increased standard deviation. As particles continue to flow in the raceway, the coal plume expands gradually, making all the sized particles move far away from the centreline. Near the front end of the

ED

raceway, Figure 8(c) and (d) indicates that the effectiveness of the size segregation process was mainly owing to the enhanced dispersion of the particles which were smaller than 60 µm. The

EP T

evolution of the size segregation along the tuyere centerline can result in changes in the conditions of particle combustion and thus contribute to different particle combustion behaviors.

AC C

5.1.3 Mechanism of size segregation Size segregation in the coal plume is different from that in a dense gas–particle system. The size segregation in a dense particulate system was mainly caused by shear (small particles) or by imbalance contact forces (large particles) [40]. In a dilute flow, the size segregation was mainly attributed to different motions based on the effects of fluid-induced forces. The particle behavior suspended in fluid flow could be measured by the particle Stokes numbers, which can be calculated with the use of Eq. (9).

 p d p2ug Stk  18 g L

(9)

The Stokes number was mainly decided by the particle diameter and gas velocity. In the blowpipe, 15

ACCEPTED MANUSCRIPT the particle Stokes number varied from 0.07 to 153. For small particles, small Stokes numbers indicate a short response time to the gas flow. For this reason, these fine particles can be easily carried by the gas and recirculate in the upper part of the raceway. However, for larger particles with Stk values being much greater than unity, the particles become more unresponsive to the gas flow. Therefore, they are more likely to maintain the original flow direction when the flow

PT

moves forward. Additionally, a larger gravitational force makes these large particles tend to flow downward and away from the tuyere centerline. Different flow motions of different sized particles

RI

in the vertical direction lead to the appearance of the size segregation in the raceway.

SC

5.2 Particle combustion characteristics

5.2.1 Evolution of particle char burnout in the blowpipe and the raceway

NU

Distributions of gas species, such as CO2, O2, and CO, are shown in Figure 9(a)–(c) to give a comprehensive picture of the PCI combustion process. Most oxygen is consumed along the coal

MA

plume owing to the combustion of coal particles, especially those located at the outer surface of the plume. As a result, a high-temperature zone appears at the outer surface of the plume, as

ED

shown in Figure 9(d). The temperature in the center of the plume is lower, and an oxygen-deficient zone exists there. These characteristics confirm the existence of the low-burnout zone inside the

EP T

plume. However, at the front-end of the raceway, the gas temperature increases, thus indicating the occurrence of an intense combustion reaction. Furthermore, the significant decrease in CO2 and the increase in the CO concentrations near the end of the raceway indicate the existence of a high

AC C

conversion rate from CO2 to CO.

Owing to the existence of the size segregation in the coal plume, particles located at different positions have different accessibility to oxygen, thus leading to different char combustion characteristics. These characteristics can be qualitatively presented by the char burnout at three typical locations. They are the top, centerline, and bottom parts of the coal plume. It is evident from Figure 10(a) that in the upstream of the plume, small and large coal particles at the top and bottom of the plume start to combust earlier than the centered particles in an oxygen-enriched environment. This leads to the delayed combustion of medium-size particles in the center of the plume, followed by the formation of a low-burnout zone. Herein, the char burnout was equivalent 16

ACCEPTED MANUSCRIPT to the mass exchange from the coal particle to the gas intended for char combustion and was proportional to the reaction rate. The total burnout rate of a particle group with a specific diameter was extensively used in many published studies [4, 10, 12]. However, because the particle burnout was calculated based on ash balance, it cannot describe the performance of carbon combustion in a straightforward way. The

 d , was used and was calculated based i

PT

particle char burnout rate of a particle with a diameter di ,

on Eq. (10). It is the ratio of the weight loss owing to char combustion to the original mass of

RI

fixed carbon of a coal particle:

d  md   fixed _ carbon (md ,0  md _a,0  md _vol,0 ) 100% i

i

i

i

(10)

SC

i

The mass percentage of unburnt char of different sized particles was then calculated as follows:

mdi _ unburnt _ char n

100%

NU

pdi  f (d i ) 

m

(11)

di _ unburnt _ char

MA

i

mdi _ unburnt _ char  mdi，0   fixed _ carbon，0  (1  di )

(13)

ED

f (dub )= max pdi

(12)

where d ub is the particle diameter at which the maximum mass percentage appeared.

EP T

Coal particle combustion behaviors are not only affected by the size segregation in the coal plume but also by variations of the coal plume itself. Images of flame recently captured in the front of the

AC C

tuyere peephole from several industrial blast furnaces proved that the variation of coal plume had a strong influence on the coal combustion in the raceway as well as the activity of the hearth [41]. Therefore, more emphasis should be laid on the particle combustion behaviors along the plume centerline rather than on the tuyere centerline, which was examined in many previous PCI simulations to demonstrate the combustion performance in the raceway [4-7, 9, 14, 15]. The performance of particle char burnout along the plume centerline is shown in Figure 10(b). In the blowpipe, the burnout increased as the coal plume moved forward for a distance of approximately 0.01 m from the lance tip, and then it decreased as the plume kept moving forward to the tuyere tip. This indicated that the size segregation weakened once coal particles were released from the lance tip and entered the blowpipe. However, the segregation did not weaken 17

ACCEPTED MANUSCRIPT more until the particles entered the raceway. The weakening of the size segregation in the blowpipe was a result of the enhanced particle dispersion in the vertical direction, which was significantly affected by the turbulence downstream of the lance tip generated owing to the existence of the lance itself [42]. Therefore, the phenomenon in accordance to which the size segregation does not weaken further also suggests that the turbulence becomes ineffective on particle dispersion. The findings have been applied recently to optimize the lance configuration by

PT

further undermining the size segregation in the blowpipe [42].

RI

Once coal particles enter the raceway, the char burnout increases significantly near the tuyere tip. This was also attributed to the weakening of the size segregation due to the enhanced dispersion of

SC

all the partic les with different sizes, as shown in Figure 11(a). However, the significant increase

NU

was immediately followed by a noticeable decline. This was because the oxygen transferred to the plume center declined owing to the additional oxygen consumption by the outer particles, as

MA

indicated in Figure 9(b) and (d). As coal particles continue to move forward, the burnout increases gradually. Near the end of the raceway, the burnout is maximized. Similarly, the burnout increase here was also a result of weakened size segregation. However, the weakening of the segregation

ED

observed herein was mainly caused by the strengthened dispersion of particles that were smaller

EP T

than 60 µm, as shown in Figure 8(d).

5.2.2 Overall pulverized coal combustion performance Firstly, as an essential indicator of the overall combustion performance and from the viewpoint of

AC C

total mass loss, the total burnout rates of pulverized coal in the predicted raceway and the blowpipe-raceway-coke bed zone were calculated according to the ash balance equation, Eq.(14). These were equal to 59.62% and 70.70% respectively. Total burnout rate =

1  ma ,0 / ma 1  ma ,0

100%

(14)

where ma ,0 is the ash content of the original coal, and ma is the ash content of the collected burnt residual. It is the total weight loss of the coal due to evaporation, devolatilization, and char reaction. Although most of the pulverized coal particles combust in the raceway, the difference between the two burnout rates indicates that a small amount of pulverized coal particles also burns in the coke 18

ACCEPTED MANUSCRIPT bed. As previously demonstrated, the particle char burnout rate was used in this study. The curve plotted in Figure 12 shows the variation of the particle char burnout rate with the particle diameter. The steep slope indicates that burnout rates of particles with a diameter smaller than 60 µm are sensitive to the particle size. This can also be proved by the unburnt char distribution at different slices, as shown in Figure 11(b). For particles with diameters in the range of 60 µm and 80 µm, the

PT

slope starts to become flat. This is consistent with the slight increases of the char burnout rate of

RI

these particles as the plume moves forward, as indicated in Figure 11(a). As the particle diameter continues to increase, the char burnout rate keeps decreasing. For a group of particles with a

SC

diameter larger than 120 µm, the char burnout rate drops to approximately 2%. The combustion

NU

performances of these large particles are limited by the residence time owing to increased inertia, which also makes them move in straight directions along the boundary of the deadman zone, as

MA

shown in Figure 7(a).

The relationship between the mass percentages of unburnt char and particle diameter is also plotted in Figure 12. Compared to the original size distribution in Figure 5, it is evident that the

ED

size distribution of the particles is modified in accordance with their carbon content. Compared to the unburnt particle distribution at different slices in Figure 11(b), the results in Figure 12 exhibit a

EP T

similar distribution to that at the plane x= -0.8. This indicates that the recirculation of small particles at the end of the raceway plays a significant role in improving the overall coal

AC C

combustion performance. As the particle diameter increases, the mass percentage of unburnt char increases at first and then decreases. The maximum percentage appears when the particle diameter,

dub , is 52.5 µm. This means that the unburnt char in the coke bed is mostly contributed by particles around 52.5 µm. Additionally, it can be inferred that the degree of size segregation in the coal plume can change as a function of the coal properties, the original size distribution, as well as the operating conditions. With the pulverized coal combustion performance analyzed from the relationship between the burnout rate and the size segregation ins ide the plume, the effects that the PCI operating conditions have on the burnout rate of the different sized particle can be better understood, and more useful suggestions can then be offered on the optimization of the operation of the PCI 19

ACCEPTED MANUSCRIPT process.

6 Conclusions This research coupled the particle combustion process with the raceway formation process based on CFD. Good agreements between the CFD results and two sets of experimental data were obtained. The CFD simulation results in a commercial blast furnace were analyzed by focusing on

PT

the gas–particle flow characteristics in the predicted raceway, the phenomenon of size segregation in the coal plume, and its effects on the combustion performance. The following are some

RI

important conclusions which can be used to guide the optimization of the PCI operation.

SC

(1) The method used to couple the raceway formation process and the pulverized coal combustion process was proved useful for predicting the raceway shape and for examining the pulverized coal

NU

injection process in the lower part of an industrial blast furnace.

(2) Suspended flowing particles with different sizes exhibited size segregation in the vertical

MA

direction. The weakening of the size segregation along the coal plume was mainly attributed to the turbulence generated by the existence of the lance, main gas jet expansion in the raceway near the tuyere tip, and the gas recirculation in the upper part of the raceway.

ED

(3) Combustion conditions of the coal particle could be changed owing to the weakening of the size segregation, thus leading to different particle combustion performances. The weakening of the

EP T

size segregation near the tuyere tip contributed to a slight increase in the char burnout rate for all sized particles. Near the end of the raceway, the weakening of the segregation arising from gas

AC C

recirculation significantly improved the burnout rate of the particles that were smaller than 60 µm. (4) Within the investigated s ize ranges, particles with a diameter of 52.5 µm yielded the largest percentage contribution of unburnt char in the coke bed. Table 1 Geometry dimension of the lower part of blast furnace Parameter Volume of blast furnace/m3 Tuyere diameter/mm Tuyere number Lance diameter/mm Distance from lance tip to tuyere tip/mm Lance insertion angle/°

Value 1800 120 26 14 250 10

Table 2 Reaction models for coal particle combustion Reaction

M odel

Reaction rate 20

ACCEPTED MANUSCRIPT

Devolatilization of coal Coaldry = VM + Char

Two-competing model[38]

Eddy dissipation model[39]

Char combustion Char+O 2=CO 2

Kinetic/diffusion rate model[40]

1VM1  (1  1 )Char1 1  VM [daf] Coaldry =  , 2  2VM 2  (1   2 )Char2  2  1.251  0.921

 '  Y     vi ,r M w,i A min   '  ， k   v,r M w,     R i ,r  min  YP     R i ,r  vi' ,r M w,i AB  N P  k "  v M j j ,r w, j     0.75 dm p   DR T  T  / 2    Ap pox 0 k , D0  C1  p  dt D0  Rk dp

Rk  C2e(E/RTP )

SC

Volatile matter combustion VM +O 2=CO 2+H 2O+N 2

 Yi , s  Yi ,  1/3   d p (2.0  0.6 Re1/2 1  p Sc ) D  p ln   dt 1  Yi ,s  

dm p

PT

Convection/diffus ion model[35-37]

RI

M oisture evaporation Coal=Coaldry +H 2O(g)

Table 3 Coke particle combustion model

Parameters in reaction rate

NU

Reactions

km,1  6.53 10 (a / bc ) Tm exp(22140 / Tm ) ,1  1 5

C+O 2=CO 2

C+CO 2=2CO

2 0.5

 d  a k   p  p m,i (0.04  0.238 p ) Di  ; 2   p 

km,3  13.4  Tm exp(17310 / Tm ) ,3  1

ED

C+H 2O=CO+H2

3( cos   1)

MA

km,2  8.31109 exp(30190 / Tm ) ,2 

Table 4 Boundary conditions and operating conditions

AC C

EP T

Parameter Blast temperature/K Blast pressure/kPa Blast rate/Nm3min-1 Oxygen enrichment rate/wt.% PCI rate/t h-1 Carrier gas Carrier gas rate/ Nm3h-1 Carrier gas pressure/ kPa Coke particle density/ kgm-3

Value 1408 371 3800 3.5 28 air 800 900 900

Table 5 Parameters of pulverized coal analysis Parameters Percentage M oisture 2.61% Volatile matter 13.75% Fixed carbon 68.49% Ash 15.14% C 80.20% H 3.71% O 4.49% N 3.22%

Acknowledgments 21

ACCEPTED MANUSCRIPT This work was supported by the Natural Science Foundation of China (Grant No. 61573383), the Project of Science and Technology of Hunan Province (Grant No. 2015WK3005), and by the China Scholarship Council. The authors gratefully acknowledge the engineers and technicians from Valin Xiangtan Iron & Steel Corporation for providing operational data and suggestions for

AC C

EP T

ED

MA

NU

SC

RI

PT

this study.

22

ACCEPTED MANUSCRIPT References: [1] Ishii K. Advanced pulverized coal injection technology and blast furnace operation: Oxford, 2000. [2] Liao JH, Yu AB, Shen YS. Modelling the inject ion of upgraded brown coals in an iron making b last furnace. Powder Technol. 314(2017)550-556. https://doi.org/10.1016/j.powtec.2016.11.005. [3] Du SW, Yeh CP, Chen WH, Tsai CH, Lucas JA. Burn ing characteristics of pulverized coal within blast furnace raceway at various injection operations and ways of oxygen enrich ment. Fuel.

PT

98(2015)-106. https://doi.org/10.1016/j.fuel.2014.11.038. [4] Shen YS, Guo BY, Yu A B, Zulli P. A three-dimensional numerical study of the co mbustion of coal

RI

blends in blast furnace. Fuel. 88(2009)255-263. https://doi.org/10.1016/j.fuel.2008.08.013.

combustion

in

b last

furnace.

ISIJ

Int.

48(2008)777-786.

NU

https://doi.org/10.2355/isijinternational.48.777.

SC

[5] Shen YS, Guo BY, Yu AB, Maldonado D, Austin P, Zulli P. Th ree -dimensional modelling of coal

[6] Yeh CP, Du SW, Tsai CH, Yang RJ. Nu merical analysis of flow and combustion behavior in tuyere

MA

and raceway of b last furnace fueled with pulverized coal and recycled top gas. Energy. 42(2012)233-240. https://doi.org/10.1016/j.energy.2012.03.065. [7] Zhang SF, Bai CG, Wen LY, Qiu GB, Lu XW. Gas-particle flo w and co mbustion characteristics of

ED

pulverized coal injection in blast furnace raceway. J Iron Steel Res Int. 17(2010)8 -12. https://doi.org/10.1016/S1006-706X(10)60175-6.

pulverized

EP T

[8] Wijayanta AT, Alam MS, Nakaso K, Fukai J, Kunito mo K, Shimizu M. Nu merical study on biochar

injection

in

blast

furnace.

ISIJ

Int.

54(2014)1521-1529.

https://doi.org/10.2355/isijinternational.54.1521.

AC C

[9] Shen YS, Yu AB, Austin P, Zulli P. Modelling in -furnace phenomena of pulverized coal in jection in iron making b last furnace: Effect of co ke bed porosities. Minerals Engineering. 33(2012)54 -65. https://doi.org/10.1016/j.mineng.2011.10.014. [10] Li YQ, Zhang XH, Zhang JY, Zhou JM, Yan HJ. Nu merical simu lation and optimization of pulverized coal injection with enriched o xygen into blast furnace. Appl Therm Eng. 67(2014)72-79. https://doi.org/10.1016/j.applthermaleng.2014.02.062. [11] Okosun T, St reet SJ, Zhao J, Wu B, Zhou CQ. Influence of conveyance methods for pulverised coal

inject ion

in

a

blast

furnace.

Ironmaking

&

Steelmaking.

44(2017)513-525.

https://doi.org/10.1080/03019233.2016.1217116. [12] Shen YS, Guo BY, Yu AB, Austin PR, Zu lli P. Three-dimensional modelling of in-furnace 23

ACCEPTED MANUSCRIPT coal/coke

combustion

in

a

blast

furnace.

Fuel.

90(2011)728-738.

https://doi.org/10.1016/j.fuel.2010.08.030. [13] Wang HT, Zhao W, Chu MS, Feng C, Liu ZG, Tang J. Current status and development trends of innovative blast furnace ironmaking technologies aimed to environmental harmony and operation intellectualization.

J

Iron

Steel

Res

Int.

24(2017)751-769.

https://doi.org/10.1016/S1006-706X(17)30115-2.

PT

[14] Zhou CQ. CFD modeling for high rate pulverized coal injection (PCI) to blast furnaces. USA : Final report of AISI/DOE Technology Roadmap Program for the Steel Industry; 2008.

RI

[15] Nogami H, Yamaoka H, Takatani K. Raceway design for the innovative blast furnace. ISIJ Int.

SC

44(2004)2150-2158. https://doi.org/10.2355/isijinternational.44.2150.

[16] Ueda S, Natsui S, Nogami H, Yag i J, Ariyama T. Recent progress and future perspective on modeling

of

blast

furnace.

ISIJ

Int.

50(2010)914-923.

NU

mathematical

https://doi.org/10.2355/isijinternational.50.914.

MA

[17] Sarkar S, Gupta GS, Kitamura SY. Predict ion o f raceway shape and size. ISIJ Int. 47(2007)1738-1744. https://doi.org/10.2355/isijinternational.47.1738. [18] Mondal SS, So m SK, Dash SK. Nu merical pred ictions on the influences of the air blast velocity,

ED

initial bed porosity and bed height on the shape and size of raceway zone in a blast furnace. J Phys D Appl Phys. 38(2005)1301-1307. https://doi.org/10.1088/0022-3727/38/8/ 030.

EP T

[19] Rangarajan D, Shio zawa T, Shen YS, Cu rtis JS, Yu AB. In fluence of operating parameters on raceway properties in a model blast furnace using a two-fluid model. Ind Eng Chem Res.

AC C

53(2014)4983-4990. DOI: 10.1021/ie301936r. [20] Miao Z, Zhou ZY, Yu AB, Shen YS. CFD-DEM simu lation of raceway formation in an iron making

b last

furnace.

Po wder

Technol.

314(2017)542-549.

https://doi.org/10.1016/j.powtec.2016.11.039. [21] Guo J, Cheng SS, Zhao HB, Pan HW, Du PY, Teng ZJ. A mechanism model fo r raceway formation and variation in a blast furnace. Metall Mater Trans B. 44(2013)487-494. https://doi.org/10.1007/s11663-013-9833-9. [22] Rajneesh S, Gupta GS. Importance of frictional forces on the formation o f cavity in a packed bed under

cross

flo w

of

gas.

https://doi.org/10.1016/S0032-5910(03)00136-0. 24

Powder

Technol.

134(2003)72-85.

ACCEPTED MANUSCRIPT [23] Gupta G, Rudolph V. Co mparison of b last furnace raceway size with theory. ISIJ Int. 46(2006)195-201. https://doi.org/10.2355/isijinternational.46.195. [24] Hilton JE, Cleary PW. Raceway format ion in laterally gas -driven particle beds. Chem Eng Sci. 80(2012)306-316. https://doi.org/10.1016/j.ces.2012.06.044. [25] Natsui S, Ueda S, Fan ZY, Andersson N, Kano J, Inoue R, et al. Charact eristics of solid flow and stress distribution including asymmetric phenomena in blast furnace analyzed by discrete element

PT

method. ISIJ Int. 50(2010)207-214. https://doi.org/10.2355/isijinternational.50.207. [26] Hou Q, E D, Kuang S, Li Z, Yu A B. DEM -based virtual experimental blast furnace: A state

model.

Po wder

314(2017)557-566.

SC

https://doi.org/10.1016/j.powtec.2016.12.017.

Technol.

RI

quasi-steady

[27] Syamlal M, O’Brien TJ. Co mputer simu lation of bubbles in a fluidized bed. AIChE Sy mp.

NU

85(1989)22-31. https://doi.org/10.1016/S0032-5910(98)00017-5.

[28] Ergun S. Fluid Flow through packed columns. Chem Eng Prog. 48(1952)89-94.

MA

[29] M iller R, Harstad K, Bellan J. Evaluation of equilibriu m and non -equilibriu m evaporation models for many-droplet gas-liquid flo w simu lations. International Journal of Mult iphase Flow. 24(1998)1025-1055. https://doi.org/10.1016/S0301-9322(98)00028-7.

ED

[30] Ranz WE, Marshall WR. Evaporation from Drops .1. Chem Eng Prog. 48(1952)141-146. [31] Ranz WE, Marshall WR. Evaporation from Drops .2. Chem Eng Prog. 48(1952)173-180.

EP T

[32] Ubhayakar SK, Stickler DB, Von Rosenberg CW, Gannon RE. Rap id devolatilization of pulverized coal in hot combustion gases. Sy mposium (International) on Co mbustion.

AC C

16(1977)427-436. https://doi.org/10.1016/S0082-0784(77)80342-1. [33] MagnussenBF H. On mathemat ical models of turbulent co mbustion with special emphasis on soot formation and combustion. 16th Sy mp (Int’l) on Co mbustion The Combustion Institute. 1976. https://doi.org/10.1016/S0082-0784(77)80366-4. [34] Field MA. Rate of co mbustion of size -graded fract ions of char fro m a lo w-rank coal between 1 200

K

and

2

000

K.

Co mbustion

and

Flame.

13(1969)237-252.

https://doi.org/10.1016/0010-2180(69)90002-9. [35] Omori Y. Blast furnace phenomena and modelling. The Iron and Steel Institute of Japan, Elsevier, London. 1987:255. [36] Burgess J, Jamaluddin A, McCarthy M, Mathieson J, No mura M, Truelove J, et al. Pulverised coal 25

ACCEPTED MANUSCRIPT ignition and combustion in the blast furnace tuyere zone. Joint Symposium of ISIJ and AIMM, Tokyo, Japan (1983)129-141. [37] Gupta S, Ye ZZ, Kim BC, Kanniala R, Kerkkonen O, Sahajwalla V. Coke graphitization and degradation

across

the

tuyere

region

in

a

b last

furnace.

Fuel.

113(2013)30-37.

https://doi.org/10.1016/j.fuproc.2013.05.074. [38] Nozawa K, Kamijo T, Sh imizu M. A quantitative description of void distributions in blast furnace Tetsu-to-Hagané.

81(1995)882-887.

PT

raceway.

https://doi.org/10.2355/tetsutohagane1955.81.9_882.

RI

[39] Vegendla SNP, Heynderickx GJ, Marin GB. Co mparison of Eu lerian –Lagrangian and Eulerian–

SC

Eu lerian method for dilute gas –solid flow with side inlet. Co mputers & Chemical Engineering. 35(2011)1192-1199. https://doi.org/10.1016/j.compchemeng.2010.09.001.

solids.

Journal

of

NU

[40] Savage S, Lun C. Part icle size segregation in inclined chute flow of dry cohesionless granular Fluid

1889(1988)311-335.

MA

https://doi.org/10.1017/S002211208800103X.

Mechanics.

[41] Zhou DD, Cheng SS, Zhang RX, Li Y, Chen T. Uniformity and Activit y of Blast Furnace Hearth by Monitoring Flame

Temperature of

Raceway Zone. ISIJ Int. 57(2017)1509 -1516.

ED

https://doi.org/10.2355/isijinternational.ISIJINT-2017-091. [42] Murai R, Kashihara Y, Murao A, Sato M. Convergent-divergent injection lance for the

EP T

enhancement of combustion efficiency of pulverized coal at blast furnace. Tetsu-to-Hagane.

AC C

103(2017)609-616. https://doi.org/10.2355/tetsutohagane.TETSU-2017-054.

26

ACCEPTED MANUSCRIPT Highlights  The process of coal particle combustion has been coupled with the raceway formation process. 

Size segregation characteristics and their evolution in the plume have been revealed.



The effects of size segregation on particle combustion performance have been investigated.

PT

Fig. 1. Geometry and porosity of the computational domain. (a) Schematic of blast furnace; (b) Dimensions of lower part of blast furnace; (c) Computational domain in 3D Fig. 2. Procedures of integrating PC combustion and raceway formation models based on ANSYS-FLUENT Fig. 3. Comparison of simulation results with measured data. (a) Hot model experiment [15];(b) Pilot-scale test[40]

AC C

EP T

ED

MA

NU

SC

RI

Fig. 4. Shape of raceway. (a) Coke fraction; (b) Side view; (c) Top view Fig. 5. Size distribution of the PC with Rosin-Rammler distribution Fig. 6. Gas flow in the raceway and coke bed (a) Gas velocity contour; (b) Gas flow streamline Fig. 7. Gas-particle flow in the raceway and coke bed. (a) Particle trajectory; (b) Particle volume concentration in the blowpipe; (c) Particle volume concentration along the centreline of coal plume; (d) Percentage of coal particle reaching the boundary of deadman zone Fig. 9. Gas species and temperature distribution. (a) CO2; (b) O 2; (c) CO; (d) Gas temperature Fig. 10. Combustion characteristics in the coal plume. (a) Contour of the residual char burnout; (b) Evolution of char burnout along the plume centerline Fig. 11. Evolution of particle combustion performance in the raceway. (a) Char burnout rate of different sized coal particle; (b) Unburnt particle distribution Fig. 12. Variations of particle burnout rate and unburnt char percentage with particle diameter

27

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12