Zhao2016structure Charancteristics and Comvustibility of Carbonaceous PDF

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 37

Accepted Manuscript

Research Paper

Structure Characteristics and Combustibility of Carbonaceous Materials from


Blast Furnace Flue Dust

Di Zhao, Jianliang Zhang, Guangwei Wang, Alberto N. Conejo, Runsheng Xu,


Haiyang Wang, Jianbo Zhong

PII: S1359-4311(16)31366-7
DOI: http://dx.doi.org/10.1016/j.applthermaleng.2016.08.020
Reference: ATE 8819

To appear in: Applied Thermal Engineering

Received Date: 19 May 2016


Revised Date: 2 August 2016
Accepted Date: 2 August 2016

Please cite this article as: D. Zhao, J. Zhang, G. Wang, A.N. Conejo, R. Xu, H. Wang, J. Zhong, Structure
Characteristics and Combustibility of Carbonaceous Materials from Blast Furnace Flue Dust, Applied Thermal
Engineering (2016), doi: http://dx.doi.org/10.1016/j.applthermaleng.2016.08.020

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.
Structure Characteristics and Combustibility of

Carbonaceous Materials from Blast Furnace Flue

Dust

Di Zhaoa, Jianliang Zhanga, Guangwei Wang*a, Alberto N. Conejob, Runsheng Xua, Haiyang

Wanga, Jianbo Zhonga

a. School of Metallurgical and Ecological Engineering, University of Science and Technology

Beijing, 100083, China

b. Metallurgy Department, Morelia Technological Institute, Morelia, Mexico

Abstract: The structure characteristics and combustibility of carbonaceous materials from

gravitational dust and bag dust of hop pocket were investigated using laser particle size analyzer,

scanning electron microscopy, X-Ray Diffraction, polarization microscopy and

thermogravimetric analysis. Simultaneously, coal char and pyrolyzed coke were used as

comparison. The acid-washing process was performed to avoid the effects of inorganic matters

and beneficiate carbonaceous materials in dust. Three representative gas-solid reactivity models,

random pore model, volume model, and unreacted core model were applied to study kinetic

parameters. Results showed that carbonaceous materials in dust were mainly originated from

coke fines and those in bag dust of hop pocket presented a high reactivity, mainly attributed to its

more disordered crystalline structure and higher porosity. It was concluded from kinetic analysis

1
that volume model was the best model for simulating the combustion process. The activation

energies of bag dust of hop pocket, gravitational dust and coke calculated by this model were

118.6 kJ/mol, 141.7kJ/mol, 156.1kJ/mol, respectively, indicating carbon in bag dust of hop

pocket are easily reacted with oxygen and proving its high combustibility.

Keywords: Thermogravimetric; Blast furnace flue dust; Carbonaceous structure; Kinetic models;

Combustibility

1. Introduction

Blast furnace flue dust (BFD), a by-product of the ironmaking process, is one of the most

complex metallurgical residues. If not properly disposed of, it can cause water, atmospheric and

soil pollution, disrupt ecological cycles and pose environmental hazards. Additionally, sharply

shrinking in landfill capacity and stringent disposal restrictions increases the pressure of the iron

and steel making industry to develop recycling techniques for dust. Recent reviews show

significant efforts undertaken to reclaim dusts. Yao et al.[1] have presented the details of coal fly

ash and its various potential application in the soil amelioration, construction and ceramic

industry, catalysis, depth separation, zeolite synthesis, etc. Das et al.[2] have effectively

recovered carbon and iron by flotation and magnetic separation techniques. Yakovlev et al.[3]

have added 0.5 wt.% of metallurgical dust to improve physico-mechanical properties of plaster

materials, increasing compressive strength and softening coefficient. Amorim et al.[4] using H2

as a reducing gas have transformed BFD into reductants of organic contaminants and toxic

metals. Shen et al.[5] have prepared nanometer-sized iron oxide black pigment using a mixed

solution of ferrous and ferric sulfate leached from BFD. Robinson[6] has found that BFD could

be a solid reductant of preparing cold blend pellets. Francis[7] has converted a mixture of BFD

and blast furnace slag into magnetic glass-ceramics with good wear resistance and chemical

2
durability. These techniques have not been widely applied due to low productivity and recovery

rate of valuable elements, large space occupation and high operational costs.

Previous studies[8-10] have reported through chemical analysis that the major components

are iron oxides and coke fines. The content of Fe in BFD ranges from 20-40%, C content is

observed between 30-50% and the rest is basic oxides(CaO, MgO), thus BFD could be blended

into sintering materials. This internal circulation method[8,9-14] was classified into the direct

sintering and mini-pellet sintering process. However, fine particle size and high moisture of BFD

can cause poor permeability and reduce sintering quality [15] (just as Singh described), and the

accumulation of problematic elements(zinc, lead or alkaline metal) will block the upward pipes

and shorten the BF life. Besides, pyrometallurgical or hydrometallurgical processes[16-18] had

been developed to enrich or recover zinc and other alkali metals in dust, but their raw material

were those dusts consisting of high harmful elements. If not, high energy consumption, low

efficiency, long payback period of investment and many difficulties in equipment maintenance

would reduce profits. So some integrated steel plants[19] have recycled and injected BF dust

with pulverized coal(PC), which turned out a stable operation, but it has not been widely applied

due to lack of a mature and integrated technological process. However, many fundamental

studies of BFD have focused on calculating PC combustion rate not its combustion behaviors in

BF. Wu et al.[20] have identified different carbonaceous materials in BFD by petrographical

microanalysis. Gupta et al.[21] have studied three BFDs using chemical analysis. Sahajwalla et

al.[22] have measured the fraction of amorphous carbon for various coals by XRD. Machado et

al[23] have quantified the percentage of carbonaceous materials in BFD samples using the XRD

technique associated with chemical analysis and a simplified quantification method using only

three size fractions was further suggested. Jie et al.[24] have found a new, quite rapid and

effective method of quantitative determination by Raman spectroscopy. Mihaiescu[25] have

3
revealed the morphology of BFD using optical microscopy and determined its petrographic

composition. Previous investigations have paid little attention to exploring the correlation

between structure characteristics and its combustion reactivity, crucial to understand its

combustion behavior in the raceway region and further to promote tuyere injection technology of

recycling flue dust. Therefore, there is a strong need to do more studies on this subject.

Thermogravimetric analysis(TGA) has been widely used to characterize the combustion

behavior of fuel and kinetics study[26,29-36] was also performed on that. Through this

thermodynamic analysis, the clarification of characteristics and mechanisms of fuel combustion

can be grasped. The non-isothermal method is often used in majority of studies, because only the

overall reactivity of dust in the isothermal one could be studied and in practical situation the

reaction temperature often changes, so kinetic parameters can be more accurate to evaluate the

reaction process. By this method, most workers have studied combustion process of coal

char[34-39,42,43], biomass char[30,31,39], soot and wastes[26,27,30]. To date, complete kinetic

research on carbonaceous materials in BFD have no found.

The aim of this work is to investigate on the structure features of carbonaceous materials in

two BFDs and their links with combustibility, meanwhile kinetics parameters were obtained by

fitting classical kinetic models. Firstly, acid-washing treatment for all samples was carried out to

extract carbonaceous materials and avoid the effect of metallic oxides on X-Ray Diffraction

(XRD) analysis as well as the combustion process. The structure characteristics were studied by

scanning electron microscopy (SEM) observation, XRD and petrographic analysis. The

combustion properties were investigated by a non-isothermal thermogravimetric analysis method.

Then three mathematical models including the random pore model(RPM)[32,33], volume

model(VM)[34], and the unreacted core model(URCM)[35,36] were employed to calculate

kinetic parameters, which explain combustive difference of carbonaceous materials.

4
2. Experimental

2.1 Sample preparation

Two BFDs, bag dust of hop pocket and gravitational dust (named as BD and GD,

respectively), original coal (CA) and lump coke (CK) were provided from one steel plant. The

samples were placed in the drying oven for 4 hours at 105°C, and lump coke was crushed with a

jaw crusher and sieved to <0.074mm particle size. To simulate that the carbonaceous materials in

dust underwent through a high temperature region, coal char(shorten as CC) and coke particles in

this work were prepared in a muffle furnace. The furnace was heated to 1300°C and kept for 25

min in an inert atmosphere of N2.

Some previous researches [2,9,10,13] have employed floatation to separate carbon from

other components in BFD. In this work, the acid-washing treatment for all samples was carried

out to remove the interference of inorganic matters and beneficiate carbon. Firstly, all specimens

were mixed with hydrochloric acid (6 mol/L) for 24 h at 50°C, then filtered and mixed with 40%

concentrated hydrofluoric acid for 36h at 60 C, finally washed with excess deionized water. This

process was repeated until the ash constituent of all samples was kept below 5%, which

approximately disregard any influence of the ash constituent. The proximate and ultimate

analysis of all samples before and after the acid-washing treatment is given in Table 1.

2.2 Sample characterization

Granulometric analysis was conducted with a LMS-30 scattering laser diffraction particle

size analyzer to observe the distribution of carbonaceous materials in the BFD. The micro-

morphology of dust was examined under a Quanta 250 Environmental SEM in a secondary

electron emission mode.

5
The structure of carbonaceous materials in dust, CC and CK was measured using a MAC

Science Diffractometer (M21XVHF22, MAC Science Co. Ltd. Japan) with Copper Kα radiation

over an angular range of 10-100 (20 /min). X-ray Diffraction (XRD) is known as a widely

established technique to determine the structural parameters of carbonaceous materials with good

reliability[21]. Previous studies have concluded that carbon microstructure had two types:

crystalline carbon and amorphous carbon, and carbonaceous materials have various proportions

of these two types. As crystalline carbon increases, more carbon layers are stacked, which is

represented by a structural parameter Lc (L002). The Lc values of carbonaceous samples can be

calculated using the classical Scherrer’s equation[37]:

0.89  
Lc 
 cos  (1)

where λ is the wavelength of the X-ray source (Cu Kα=0.154060 nm), β is the full width at half

maximum (FWHM) of the 002 reflection peak expressed in radians and is the center of 002 peak

expressed in degrees.

The two types of dusts are quantified by means of petrographical microanalysis. A DAS

microscope (Leica DMRP RXP), which is able to magnify samples to 500 diameters, is applied

in this work. The surface of sample is into divided 576 panes (24×24). There are more than 500

cross points to be measured. The counting-points method for the coal phase is conducted to

measure the fraction of area occupied by unconsumed coke fines(UCK) and unconsumed

pulverized coal(UPC) as well as oxides in BFD[38]. It’s well known that the structure of coke

particles can be divided into filament-like, granular mosaic, flowing, flake-like and blocky

textures, while UPC was observed in the form of undeformed and deformed coal. The area

percentage of coke and unconsumed pulverized coal can be calculated by Eqs.(2)-(4), where the

6
correctional parameter α, β, γ was respectively given as 0.9, 0.5, 0.1. The mass fraction of UCK

and UPC were obtained by carbon mass fraction in BFD and the corresponding mass ratio in

carbonaceous substance. The shape of particles in dusts can be considered as balls with the same

size and the area ratio can be taken as volume ratio. The gravity of carbonaceous materials in

dust had little difference. Based on this assumption, the percentage of volume could be equal to

the mass fraction. Thus the formula was given by Eq.(4)

UCK(%)    (1  2  3  4)    5 (2)

UPC(%)    (6  7) (3)

UCK(%)
 UCK(wt .%)  C(wt .%)  (4)
UCK(%)  UPC(%)

where 1: hemophilic silk carbon; 2: block structure; 3: flowing structure; 4: granular inlay

structure; 5: residue coke; 6: deformed coal; 7: undeformed coal.

2.3 Combustion tests

The combustive tests were carried out on HCT-3 thermogravimetric analyzer (manufactured

by Henven Scientific Instrument Factory, Beijing)[39] at atmospheric pressure and

measurements were carried out under the same condition using correct baselines after the

subtraction of predetermined ones without sample. In consideration of minimizing the effects of

mass transfer and avoiding heat transfer limitation, about 5mg of sample was placed in a crucible

with height of 1.5mm and diameter of 3 mm and all tests were performed under non-isothermal

conditions from room temperature to1200°C at three different heating rates: 2.5°C /min, 5°C

/min and 10°C /min. The compressed air was pumped at a gas flow rate of 60 ml/min. A

7
thermocouple was located close to the platinum to monitor the temperature. In this study, every

experiment was replicated three times.

The combustive conversion(X) was calculated using the following equations:

m0  mt
X  (5)
m0  m

wherein m0 represents the sample mass at the start of reaction; mt is the sample mass at time of t;

m denotes the remaining mass at the end of reaction.

Some specific parameters in this work were used to evaluate the combustion process, such

as the initial combustive temperature (Ti), which was not a physical property of a fuel and

ascertained by the method of TG-DTG[39]; the burnout temperature (Tf), which was the

temperature of weight loss above 95%; the comprehensive combustive characteristic index S was

determined by the equation as follow:

(dX / dt ) max  (dX / dt ) mean


S (6)
Ti 2  T f

where (dX / dt ) max is the maximum combustive rate (mg/min); (dX / dt )mean represents the mean

value for combustive rate(mg/min).

2.4 Kinetic models

Kinetic analysis of combustion process is traditionally expected to produce an adequate

description of the gas-solid heterogeneous reaction process in the following form:

 k  Pg , T  f  X 
dX
(7)
dt

8
where k is the apparent reaction rate constant, which is a function of reaction temperature T and

partial pressure in the gas phase Pg; f  X  denotes a kinetic mechanism function in the

combustion reaction; t is time; X is the carbon conversion rate.

It is assumed that the partial pressure is kept constant during the process, the temperature

dependence of the apparent rate constant is introduced by replacing k Pg , T  with the Arrhenius

equation, which gives:

dX
 k0e  E RT f  X  (8)
dt

where k0 is pre-exponential factor; E is the activation energy; R is the universal gas constant.

Three well-known theoretical kinetic models, RPM, VM and the URCM, were applied to

simulate the combustion process of carbonaceous materials by calculating the function of carbon

conversion rate versus time or reaction rate versus conversion rate and further kinetic parameters

were obtained. Different mechanism functions are hypothesized in three models. Pore structure

and its evolution as the development of reaction is considered in the RPM model. When

chemical reaction is the controlled step, the combustion rate can be described as:

dX
 k0e  E / RT 1  X  1  ln 1  X  (9)
dt
where  denotes a parameter of particle structure, associated with the initial porosity  0 and pore

4L0 1   0 
length L, and the expression is   , wherein S0 is the initial surface area.
S 02

dX
 k0e  E / RT 1  X  (10)
dt

9
The URCM model assumes that the reaction occurs at the external surface of solid particle

and gradually moves inside, and there is always a shrinking core of non-reacted solid. The

reaction expression is given by:

dX
 k0e  E / RT 1  X 
23
(11)
dt

Under non-isothermal conditions, the temperature T is related to the heating rate β and time

t, and can be described as follows:

T  T0  t (12)

The final integrated expression is then given as follows by introducing Eq.(12) into Eq.(9):

 
X  1  exp    A0 
T  T0   exp   E    1  A  T  T0   exp   E    (13)
    
 4   RT   
1
   RT   

k0C n S 0 4L0k0C n
where A0  ; A1  .
1 0 S0

Similarly, Eq.(10) and Eq.(11) can be integrated with Eq.(12) to give Eqs.(14) and (15),

respectively:


X  1  exp   k0
T  T0   exp   E   (14)
 
   RT  

 k T  T0 
3
  E  (15)
X  1  1  0  exp   
 3  RT  

A previous study[40] concluded that it was necessary to employ at least three different

heating rates in TPR(temperature-programmed reaction technique) runs to reduce the fitting

error and obtain reliable activation energies. In this work kinetic parameters were obtained from

three TPR runs. The explicit equations of (9)-(11) describe the relations between conversion

rate, conversion and temperature under the control of chemical reaction. The nonlinear least-

10
squares fitting method is employed to calculate the kinetic parameters including E, k0 and 

from experimental data. In addition, the kinetic models would be further verified quantitatively

by comparing the experimental and calculated values in a nonlinear least squares algorithm,

given in the following expression[41]:

 X  X calc,i  N
N
DEV  X %  100  (16)
2
exp,i
i 1

where DEV  X % is relative error; X exp,i is experimental data; X calc,i is calculated by three

models; N is the number of test points.

3. Results and discussions

3.1 Fundamental characteristics of samples

Table 1 presents the proximate and ultimate analysis of BFD, CK and CA before and after

acid-washing treatment on dry basis. The ash content of BD-aw, GD-aw, CK-aw and CC-aw

(represent for samples treated with acid-washing) was relatively low with 4.01%, 5.10%, 2.03%,

1.37%, respectively, indicating that the demineralized samples can be basically considered as

their corresponding carbonaceous materials in this study. The particle size distribution of BFD

and CA before acid-washing treatment are given in Figure 1. It can be observed that the

predominant fraction was above 100μm for GD, whereas the main range of CA and BD were

about 10~100μm. Table 2 was obtained through the interpolation method for the analysis of size

distribution data, and is clearly seen that the average particle of GD, BD and CA was 170.43μm,

14.64μm, and 13.74μm. Approximately 86% of BD was below 38μm and practically no particles

above 90μm, which was very similar to the CA. On the contrary, there were no particles below

11
38μm in the GD. A classification based on particle size distribution can be applied to

differentiate the two dusts and enlarge the efficient utilization in commercial applications.

The morphology of dust particles after acid-washing treatment was characterized by SEM

technique; the images of BD-aw and GD-aw are presented in Figure 2. Consistent with the

particle size analysis, it can be found that large amounts of particles in BD-aw tend to be smaller,

irregular, sharped edge and fluffy; while particles in GD-aw are larger, porous and coarse, which

are typical features of coke fines. It can be initially estimated that carbonaceous materials of BD-

aw and GD-aw mainly originated from char fines and coke fines, respectively. In the BF, it’s

known that coke degradation in the BF stock line and upper shaft (lumpy and softening zones) is

generated by shattering and abrasion, and fines generation in this part are usually produced by

shear stress from the increased load of coke bed or the attrition caused by the upward gas flow,

while in the raceway produced by mechanical impact, thermal stress and solution loss reactions.

However, fines originated from pulverized coal injection(PCI) consist of unburned char and soot.

As with the increase of PCI rate the combustion efficiency of PC decreases, leading to an

outflow of char, soot and ashes from the raceway. This is attributed to the limited time and space

between the exit of the injection lance and the rear wall of the raceway.

The carbon origins can’t be accurately defined from SEM images due to the limitation of

microscopic area. Therefore, petrographic analysis of two dusts was performed in this study and

the results are given in Table 3. Contrary to the SEM inference, it can be observed from Table 3

that BD-aw and GD-aw both mainly consist of coke particles. The mass fraction of UCK in BD-

aw and GD-aw calculated by the equations of (2)-(4), is 88.67% and 92.91%, respectively,

suggesting that combustive efficiency of PCI in the BF was very high. In the tuyere, coal

particles will rapidly undergo the release of volatile matter, char combustion and gasification.

12
Coal devolatilization releases tar, which is considered as a source of forming soot particles by

direct dehydrogenation, and the primary soot particles diameters ranges from 25 to 60 nm.

Further down in the BF, the solution loss reaction promotes most of the soot and char particles

consumption. Therefore, coke fines may not be totally consumed, therefore the high amount of

coke fines in BFD.

3.2 Analysis of carbon microstructures in BFD

The characteristics of carbonaceous materials mainly depend on carbon microstructure,

which includes the microcrystalline units and spatial packing of graphite flake layer. The reduced

intensities of XRD profiles of BD-aw, GD-aw, CC-aw, CK-aw and CK are shown in Figure 3.

All samples contain a clear (002) and (100) band, but the 002 diffraction peak of CC presented a

broad hump due to its turbostratic structure between graphite and amorphous structure[41]. It

could be observed from CC-aw to CK-aw that the 002 peak became sharper and narrower and

the position of this peak was shifted to a larger 2-theta value, indicating that the microcrystalline

structure of carbonaceous materials was tended to be more ordered. Meanwhile, some structural

parameters, average stacking height Lc, average lateral size La, number of carbon atoms per

lamellae N and interlayer spacing d002, were calculated from XRD spectra and presented in Table

4. The Lc and d002 values were ranked from high to low in this order; CK-aw, GD-aw, BD-aw, CK

and CC-aw, and it was reflected that aromatic carbon lamellas were stacked higher in the spatial

arrangement. Also, it can be found that the La value of CK-aw and CC-aw was quite lower

because they were prepared in the condition of slow pyrolysis and their aromatic side chains had

enough time to be dropped and removed. In addition, graphite degree of the two BFDs was

situated between CK-aw and CK, indicating that the UCK of two BFDs were from coke fines

formed in the upper region of BF including lumpy and softening zones.

13
3.3 Thermogravimetric analysis

Conversion and conversion rate profiles (TG-DTG) of BD-aw, GD-aw, CC-aw and CK-aw

at different heating rates are shown in Figure 4. Representative combustion parameters of four

samples obtained from TG-DTG curves are given in Table 5.

The four samples after acid-washing treatment were dried in an oven and the weight loss of

water in the initial period was relatively small, thus all TG-DTG profiles showed in Fig.6 are

selected from 300°C. As a whole, the weight loss curves of BD-aw, GD-aw and CK-aw have

only one peak rate, but those of CC-aw have an obvious transformation, which prove the

petrographic results once more that BD-aw and GD-aw originate from coke fines. Besides, there

exists a little transition at 500-600°C in Figure 5(a) as the heating rate is high. This is very likely

due to a higher content of volatile matter in GD, so that the combustion of volatile matter and

fixed carbon is separated. The combustion process of carbon is generally divided into three

periods:(1)first period: a small increase in weight due to gas absorption; (2)second period: rapid

combustion of samples up to the largest weight loss; (3)third period: combustion was finished

and the weight was kept constant. It was also found that Ti and Tf increase gradually as the

increase of heating rates and TG profiles move to a high temperature area, which causes thermal

hysteresis. This is because combustion reactions of these samples are all endothermic, so more

time is needed to transfer heat to low internal temperature of samples under high heating rates.

Besides, original coal in this study is a mixture of bituminous and anthracite coal, and the

former, low rank coal, is initially reacted with oxygen during the combustion, thus there exists

another obvious combustive peak in the DTG curve of CC-aw. This phenomenon was also

reported in the study[42]. From Table 5, it can also be found that the values of Ti, Tf, maximum

and mean values of weight loss rate for four samples at the same heating condition are different.

The comprehensive characteristic index S was applied to evaluate the combustion property of

14
four carbonaceous materials, and a larger S value indicates better combustibility. The combustive

sequence of carbonaceous materials at the same heating rate could be described as follows: CC-

aw> CK-aw > BD-aw > GD-aw.

3.4 Correlation between combustion reactivity and characteristics of carbon

microstructures

The combustion of BFD after acid-washing consisted of devolatilization and combustion of

fixed carbon. On one hand, the combustion of fixed carbon could be heated and prompted by the

former process; on the other hand, pore structure in the surface of carbon is susceptible to tar and

pyrolysis gas when volatile matter is released. The fundamental process of FC combustion is that

oxygen atoms in the atmosphere diffuse and are absorbed to the surface of the porous structure,

then reaction is started, indicating carbon combustion is mainly affected by kinetic conditions,

microstructure and active sites in this process. Correlation between combustion reactivity and

carbon microstructure is described in Figure 5, and it is found that S values of all samples apart

from coke decrease with the increase of Lc. Take an example of two dusts, good combustibility is

mainly due to random carbon crystalline layers and broad interlayer space promoted that oxygen

atoms get absorbed in the micro pores and combined with active carbon atoms in all directions.

SEM micrographs shows that there are amounts of remaining macro pores and gap after volatile

matter in the surface of GD-aw is released, so less active carbon is in GD-aw; but granular inlay

is observed on the surface of BD-aw. The high content of volatile matter in BD-aw (shown in

Table 1) escapes during combustion and then it would have high porosity and a large reactive

surface, which could further explain why combustibility of BD-aw is better than GD-aw. The

combustion property of coke prepared in this work performs better than two dusts, which can be

attributed to significant amounts of active carbon sites. BFD went through series of complex

15
chemical and physical reactions in BF and most remaining coke fines in dusts were inactive,

even though graphite order of dusts was lower.

3.5 Kinetic parameters

Relations of fractional conversion rate to conversion of samples simulated by the kinetic

models described previously are shown in Figure 6. It has been found that the conversion rate

increased with the increasing of conversion before the peak value of conversion rate. This could

be explained as follows: surface area and active sites would improve the conversion rate at the

initial stage and they kept unchanged after reaching a specific value. The apparent activation

energy of different materials changed, thus the reaction process of coal char (a mixture of

bituminous coal and anthracite coal) would be not accurately described by a simple and single-

stage kinetic model. Therefore, it was observed from Figure 6 that the simulation of CC-aw by

all models performed very poor. The corresponding kinetic parameters of BD-aw, GD-aw and

coke at different heating rate calculated by three models is shown in Table 6, and is noted that

fitting results by URCM for CC-aw were not shown due to very poor simulation. It is shown that

fitting profiles of two BFDs and CK-aw by URCM model was far from experimental ones and

the other two models can fit well with practical data (the highest R2 value). The activation energy

of BD-aw calculated by RPM and VM model was ranged from 118.6-122.8 kJ/mol, 141.7-146.2

kJ/mol for GD-aw, and 156.1-162.9 kJ/mol for CK-aw. Wu et al.[43] determined the activation

energy of char gasification reaction in the temperature range from 109.2-205.3 kJ/mol. and

reported an activation energy of 135.7-208.8 kJ/mol. Maryam et al.[44] analyzed the kinetics of

CO2 gasification of petroleum coke at 1173-1248 K and 0.1-2.4 MPa, and reported activation

energies, about 260±24 kJ/mol and 254±12 kJ/mol, for pressures of 0.1 and 1.4 MPa. Activation

energies reported in previous investigations are a little higher than those reported in this work,

16
which can be attributed to the sufficient pyrolysis (or different annealing conditions) and

intrinsic properties of coal. In addition, kinetic models performed not well under the condition of

high heating rate, indicating that thermal hysteresis at high heating rate should not be ignored.

The comparison between experimental and calculated data by the three models is presented

in Figure 7 and quantitative relative error is shown in Table 7. It can be observed that VM model

can fit best with experimental data and its relative error is the smallest, followed by RPM and

URCM. It’s well known that VM model assumed that the size of particles remained constant and

density varied evenly as the reaction went on the whole surface of particles; RPM model

assumed that particles contain amounts of uniform spherical pores and reactions occur on the

internal and not overlapping surface; meanwhile surface area varies with the increase of

conversion. From SEM images, it was confirmed that shapes of pore structure in coke carbon

atoms were not even, spherical and deformed, so the structure factor had little effect on reaction

and Eq.(9) was approximately equal to Eq.(10). These features can explain why VM and RPM

models give similar results.

4. Conclusions

Particle analysis can be used to classify the two types of dusts; From petrographic results,

carbonaceous materials in BD-aw and GD-aw came from above 95% of coke fines, suggesting

that char particles were almost consumed in the BF. The XRD analysis indicate graphite degree

of all specimens varied and their Lc values ranked as follows: CK-aw> GD-aw> BD-aw> CK>

CC-aw, indicating that coke fines in BD-aw and GD-aw originated from the lumpy and softening

zones in the BF. Therefore, it can be concluded that carbon structure in BFD exists large part of

aromatic structure and small part of amorphous one.

17
The combustion reactivity of BD was higher than that in GD, mainly attributed to lower

ordered crystalline structure of carbon and pyrolyzed coke performed better combustibility than

dusts due to its amounts of carbon active sites. Simultaneously, the kinetic parameters were

obtained by applying three models to simulate experimental data at different heating rates,

wherein VM and RPM fit well with a correlation coefficient of 0.9998, followed by URCM

models. The activation energy of BD-aw, GD-aw and CK-aw calculated by VM and RPM was

118.6-122.8 kJ/mol, 141.7-146.2 kJ/mol, 156.1-162.9 kJ/mol, respectively.

AUTHOR INFORMATION

Corresponding Author

*Telephone: 13811185467. E-mail: wgw676@163.com.

Notes

The authors declare no competing financial interest.

Acknowledgment

This work was supported by the National Natural Science Foundation of China and Baosteel
Group Co., LTD of Shanghai for the Key Joint Project (U1260202), the National Basic Research
Program of China (973 Program) (2012CB720401) and the Fundamental Research Funds for the
Central Universities (FRF-TP-15-063A1).

References

[1] Z. T. Yao, X. S. Ji, P. K. Sarker. A comprehensive review on the applications of coal fly ash,

Earth-Sci. Rev. 141(2015) 105-121.

[2] B. Das, S. Prakash, P. S. R. Reddy, An overview of utilization of slag and sludge from steel

industries, Resour. Conserv. Recy. 50(2007) 40-57.

18
[3] G. Yakovlev, V. Khozin, I. Polyanskikh, Utilization of blast furnace flue dust while

modifying gypsum binders with carbon nanostructures, in: The 9th International Conference

‘‘ENVIRONMENTAL ENGINEERING”, 2014, eISBN 978-609-457-640-9.

[4] C. C. Amorim, P. R. Dutra, M. D. Leão M, et al. Controlled reduction of steel waste to

produce active iron phases for environmental applications, Chem. Eng. J., 209(2012) 645-

651.

[5] L. Shen, Y. Qiao, Y. Guo, Preparation and formation mechanism of nano-iron oxide black

pigment from blast furnace flue dust, Ceram. Int., 39 (2013) 737-744.

[6] R. Robinson, High temperature properties of by-product cold bonded pellets containing blast

furnace flue dust, Thermochim. Acta., 432(2005) 112-123.

[7] A. A. Francis. Crystallization kinetics of magnetic glass–ceramics prepared by the

processing of waste materials, Mater. Res. Bull., 41(2006) 1146-1154.

[8] N. AE. l-Hussiny, M. E. H. Shalabi, Effect of recycling blast furnace flue dust as pellets on

the sintering performance, Sci. Sinter., 42(2010) 269-281.

[9] A. Yehia, F. H. El-Rahiem, Recovery and utilization of iron and carbon values from blast

furnace flue dust, Miner. Process. Extr. M., 2013.

[10] U. Leimalm, M. Lundgren, L. S. Ö kvist, Off-gas dust in an experimental blast furnace part

1: characterization of flue dust, sludge and shaft fines, ISIJ. Int., 50(2010) 1560-1569.

[11] A. Masutaro Method of preparing metal containing pellets from blast furnace dust and

converter dust, U.S. Patent 3,652,260, 1972-3-28.

19
[12] M. Fröhling, O. Rentz, A case study on r-aw material blending for the recycling of ferrous

wastes in a blast furnace, J. Clean. Prod. 18(2010) 161-173.

[13] C. Lanzerstorfer, M. Kröppl, Air classification of blast furnace dust collected in a fabric

filter for recycling to the sinter process, Resources, Conservation and Recycling, 86(2014)

132-137.

[14] C. Liu, Z..Xie, F. Sun, System dynamics analysis on characteristics of iron-flow in sintering

process, Appl. Therm. Eng., 82(2015) 206-211.

[15] P. K. Singh, A. L. Kumar, R. K. Rai, Recycling of environmentally hazardous wastage of

integrated steel plants, in: international conference on integrated waste management and

green energy engineering (‘ICIWMGEE.’), 2013.

[16] M. Zhang, W. S. Li, W. Y. Wang, Recovery of potassium chloride from blast furnace flue

dust, RSC. Advances, 5(2015) 84901-84909.

[17] J. M. Steer, A. J. Griffiths. Investigation of carboxylic acids and non-aqueous solvents for

the selective leaching of zinc from blast furnace dust slurry, Hydrometallurgy, 140(2013)

34-41.

[18] M. K. Jha, V. Kumar, R. J. Singh, Review of hydrometallurgical recovery of zinc from

industrial wastes, Resources, conservation and recycling, 33(2001) 1-22.

[19] L. S. Ökvist, Co-injection of basic fluxes or bf flue dust with PC into a BF charged with

100% pellets, Doctoral thesis, Luleå University of Technology, 2004.

[20] K. Wu, R. Ding, Q. Han, Research on unconsumed fine coke and pulverized coal of BF dust

under different PCI rates in BF at Capital Steel Co., ISIJ. Int., 50(2010) 390-395.

20
[21] S. Gupta, V. Sahajwalla, P. Chaubal, Carbon structure of coke at high temperatures and its

influence on coke fines in blast furnace dust, Metall. Mater. Trans. B., 36(2015) 385-394.

[22] L. Lu, V. Sahajwalla, C. Kong, Quantitative X-ray diffraction analysis and its application to

various coals, Carbon, 39(2001) 1821-1833.

[23] A. S. Machado, A. S. Mexias, A. C. F. Vilela, E. Osorio, Study of coal, char and coke fines

structures and their proportions in the off-gas blast furnace samples by X-ray diffraction,

Fuel. 114(2013) 224-228.

[24] J. Yu, L. Sun, J. Xiang, New method of quantitative determination of the carbon source in

blast furnace flue dust, Energ. Fuel. 28(2014) 7235-7242.

[25] D. C. Mihaiescu, G. Predeanu, C. Panaitescu, Characterization of some blast furnace waste

dusts, UPB Sci. Bull, 76(2014) 227-234.

[26] C.C. Ma, J.B. Gao, L. Zhong, S.K. Xing, Experimental investigation of the oxidation

behavior and thermal kinetics of diesel particulate matter with non-thermal plasma, Appl.

Therm. Eng. 99 (2016) 1110–1118.

[27] G.C. Liu, Y.F. Liao, S.D. Guo, X.Q. Ma, C.C. Zeng, J. Wu, Thermal behavior and kinetics

of municipal solid waste during pyrolysis and combustion process, Appl. Therm. Eng. 98

(2016) 400-408.

[28] J. Giro-Paloma, R. del Valle-Zermeño, A. I. Fernández, Thermogravimetric study of a phase

change slurry: effect of variable conditions, Appl. Therm. Eng., 107(2016) 329-338.

21
[29] P. Sarkar, A. Mukherjee, S. G. Sahu, Evaluation of combustion characteristics in

thermogravimetric analyzer and drop tube furnace for Indian coal blends, Appl. Therm. Eng.,

60(2013) 145-151.

[30] Z. Cai, X. Ma, S. Fang, Thermogravimetric analysis of the co-combustion of eucalyptus

residues and paper mill sludge, Appl. Therm. Eng., 106(2016) 938-943.

[31] C. Jaroenkhasemmeesuk, N. Tippay-awong. Thermal degradation kinetics of s-awdust under

intermediate heating rates, Appl. Therm. Eng., 103(2016) 170-176.

[32] S.K. Bhatia, D.D. Perlmutter, A random pore model for fluid-solid reactions: І isothermal

kinetic control, AIChE. J. 26(1980) 379-385.

[33] S.K. Bhatia, D.D. Perlmutter, A random pore model for fluid-solid reactions: ІІ Diffusion

and transport effects, AIChE. J. 27(1981) 247-254.

[34] J. Ochoa, M.C. Cassanello, P.R. Bonelli, CO2 gasification of Argentinean coal chars: a

kinetic characterization, Fuel. Process. Technol. 74(2001) 161-176.

[35] S. Kasaoka, Y. Sakata, S. Kayano, Kinetic evaluation of reactivity of various coal chars for

gasification with carbon dioxide in comparison with stream, Int. Chem. Eng. 25(1984) 160-

175.

[36] J.Y. Shang, E.W. Eduardo, Kinetic and FTIR studies of the sodium catalyzed steam

gasification of coal char, Fuel, 63(1984) 1640-1609.

[37] S. Ergun, VH. Tiensuu, Interpretation of the intensities of X-rays scattered by coals, Fuel,

38(1959) 64-78.

22
[38] Ministry of Metallurgical Industry of China. GB/T 8899-1998. Determination of maceral

group composition and minerals in coal, Ministry of Metallurgical Industry of China:

Beijing, China, 1998.

[39] G. Wang, J. Zhang, X. Hou, Study on CO2 gasification properties and kinetics of biomass

chars and anthracite char, Bioresource. technol. 177(2015) 66-73.

[40] K. Miura, P.L. Silveston, Analysis of gas-solid reactions by use of a temperature-

programmed reaction technique, Energ. Fuel. 3(1989) 243-249.

[41] B. Manoj, A.G. Kunjomana, Study of stacking structure of amorphous carbon by X-ray

diffraction technique, Int. J. Electrochem. Sc. 7(2012) 3127-3134.

[42] Y.H. Xiang, Y. Wang, J.L. Zhang, Study on combustion kinetics of partial gasified coal char

by using distributed activation energy model, Journal of Combustion Science and

Technology. 9(2003) 567-570.

[43] S. Wu, J. Gu, X. Zhang, Variation of carbon crystalline structures and CO2 gasification

reactivity of Shenfu coal chars at elevated temperatures, Energ. Fuel. 22(2007) 199-206.

[44] M. Maryam, M.H. Josephine, Kinetic analysis of CO2 gasification of petroleum coke at high

pressures, Energ. Fuel. 25(2011) 4043-4048.

23
List of Figures
Figure 1. Particle size distribution of BF flue dust and raw coal before acid-washing treatment:
(a)GD; (b)BD; (c)CA.

Figure 2. SEM images of BD-aw and GD-aw under 500 and 5000 times magnification,

respectively.

Figure 3. XRD spectra of two BF dusts, CC-aw, CK-aw and CK.

Figure 4. Conversion and conversion rate of four samples at three different heating rates: (a)

GD-aw, (b) BD-aw, (c) CK-aw, (d) CC-aw.

Figure 5. (a) Correlation between S and Lc values of all samples; (b)a schematic illustration of

coke fines reacted with oxygen.

Figure 6. Fractional conversion rate to conversion simulated by three kinetic models: (a) GD-

aw, (b) BD-aw, (c) CK-aw, (d) CC-aw.

Figure 7. Experimental and calculated curves of weight loss to temperature at different heating

rates: GD-aw, (b) BD-aw, (c) CK-aw, (d) CC-aw.

24
Figure 1. Particle size distribution of BF flue dust and raw coal before acid-washing treatment:
(a)GD; (b)BD; (c)CA.

25
Figure 2. SEM images of BD-aw and GD-aw under 500 and 5000 times magnification,

respectively.

26
Figure 3. XRD spectra of two BF dusts, CC-aw, CK-aw and CK

27
-0.30
0.0 0.0 -0.3

DTG(mg/min)
-0.25

DTG(mg/min)
(a) -0.20 (b) -0.2
-0.15
0.2 0.2
-0.10 -0.1
-0.05
0.4 0.00 0.4 0.0

TG(%)
300 400 500 600 700 800 900 300 400 500 600 700 800 900
TG

Temperature(°C) Temperature(°C)

0.6 0.6

0.8 0.8
2.5°C/min 2.5°C/min
5°C/min 5°C/min
1.0 10°C/min 10°C/min
1.0

300 400 500 600 700 800 900 300 400 500 600 700 800 900
Temperature(°C) Temperature(°C)

-0.5
0.0 0.0 -0.4
-0.4
DTG(mg/min)

(c) (d)

DTG(mg/min)
-0.3
-0.3
0.2 0.2 -0.2
-0.2

-0.1 -0.1
0.4 0.4
TG(%)

TG(%)

0.0300 400 500 600 700 800 900 0.0


Temeprature(°C) 300 400 500 600 700 800 900
Temperature(°C)
0.6 0.6

0.8 0.8
2.5°C/min 2.5°C/min
5°C/min 5°C/min
10°C/min 10°C/min
1.0 1.0

300 400 500 600 700 800 900 300 400 500 600 700 800 900
Temperature(°C) Temperature(°C)

Figure 4. Conversion and conversion rate of four samples at three different heating rates: (a)

GD-aw, (b) BD-aw, (c) CK-aw, (d) CC-aw

28
Figure 5. (a) Correlation between S and Lc values of all samples; (b)a schematic illustration of

coke fines reacted with oxygen.

29
0.07
0.08
(a) RPM VM URCM
(b) RPM VM URCM
0.06 0.07

2.5°C/min 2.5°C/min
0.05 5°C/min 0.06 5°C/min
10°C/min 10°C/min
0.05
0.04
-1

-1
dX/dt,min

dX/dt,min
0.04
0.03
0.03

0.02
0.02

0.01
0.01

0.00 0.00
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
X X

0.08
(b) RPM VM URCM 0.10 (d) RPM VM URCM
0.07 2.5°C/min
2.5°C/min 5°C/min
0.06 5°C/min 0.08 10°C/min
10°C/min
0.05
-1
-1

0.06
dX/dt,min
dX/dt,min

0.04

0.03 0.04

0.02

0.02
0.01

0.00 0.00
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
X X

Figure 6. Fractional conversion rate to conversion simulated by three kinetic models: (a) GD-

aw, (b) BD-aw, (c) CK-aw, (d) CC-aw

30
0.0 0.0
(a) (b)
0.2 0.2

0.4 0.4
2.5°C/min
2.5°C/min
X

5 °C/min

X
5 °C/min
10 °C/min
0.6 10 °C/min 0.6
RPM
RPM
VM
VM
URCM
0.8 URCM 0.8

1.0 1.0

300 400 500 600 700 800 900 300 400 500 600 700 800 900
Temperature(°C) Temperature(°C)
0.0 0.0
(a) (b)
0.2 0.2

0.4 0.4
2.5°C/min
2.5°C/min

X
5 °C/min

X
5 °C/min
10 °C/min
0.6 10 °C/min 0.6
RPM
RPM
VM
VM
URCM
0.8 URCM
0.0 0.0 0.8

(c) 1.0

300 400 500 600 700 800 900


1.0

300 400 500


(d)
600 700 800 900
Temperature(°C) Temperature(°C)

0.2 0.2 0.0


(c) 0.0
(d)
0.2 0.2

0.4 0.4

X
2.5°C/min
0.6 5 °C/min 0.6 2.5°C/min
0.4 0.4 10 °C/min
RPM
VM
5°C/min
10°C/min
0.8 0.8 RPM
URCM
VM
X

URCM
X

1.0 1.0

2.5°C/min 400 500 600


Temperature(°C)
700 800 900 300 400 500 600
Temperature(°C)
700 800 900

0.6 5 °C/min 0.6 2.5°C/min


10 °C/min
5°C/min
RPM
10°C/min
0.8 VM 0.8 RPM
URCM
VM
URCM
1.0 1.0

400 500 600 700 800 900 300 400 500 600 700 800 900
Temperature(°C) Temperature(°C)

Figure 7.Experimental and calculated curves of weight loss to temperature at different heating

rates: GD-aw, (b) BD-aw, (c) CK-aw, (d) CC-aw.

31
List of Tables
Table 1. Proximate analysis and ultimate analysis of coal samples (%)

Table 2 Results from the analysis of particle size distribution

Table 3 The percentage of area occupied components of microstructure in BFD(%).

Table 4 Structural parameters of all samples calculated from the XRD spectra

Table 5 Representative combustion parameters of BD-aw, GD-aw, CK-aw and CC-aw at three

different heating rates.

Table 6 The calculated kinetic parameters of different samples

Table 7 The deviation between experimental and calculated data

32
Table 1. Proximate analysis and ultimate analysis of all samples (%)

Proximate analysis,% Ultimate analysis,%


Samples
a
FCd Ad Vd Cd Hd Oda Nd Sd
BD 10.9 81.64 7.46 11.67 0.69 5.43 0.18 0.39
GD 64 31.96 4.03 66.01 0.67 1.22 0.13 0.28
CK 85.5 12.83 1.67 83.82 0.59 0.65 0.34 0.74
CA 68.3 9.54 22.15 82.6 3.49 3.02 0.88 0.46
BD-aw 88.59 4.01 7.4 89.99 0.09 5.32 0.15 0.44
GD-aw 89.78 5.1 5.12 93.21 0.07 1.18 0.14 0.3
CK-aw 96.35 2.03 1.62 96.32 0.14 0.59 0.27 0.65
CC-aw 83.62 1.37 15.01 95.67 0.19 1.74 0.61 0.42

FC, fixed carbon; V, volatile matter; A, ash; d, dry basis. a, calculated by difference; aw, acid-washing
treatment

Table 2 Results from the analysis of particle size distribution

90- 180- 250- > Average


Samples <38μm 38-63μm 63-90μm
180μm 250μm 425μm 425μm particle
size
GD 0 2.12 7.76 45.07 27.3 17.75 0 170.43
BD 86.47 11.13 2.4 0 0 0 0 14..64
CA 83.74 12.69 3.57 0 0 0 0 13.74

Table 3 The percentage of area occupied components of microstructure in BFD(%).

Coke(%) UPC(%)
Samples ∑coke(%) ∑UPC(%) ∑coke(wt.%) ∑UPC(wt.%)
1 2 3 4 5 6 7
BD-aw - 9.77 0.97 60.46 14.29 1.95 - 65.51 0.98 88.67 1.33
GD-aw 0.65 15.69 1.31 71.89 7.52 1.31 - 81.34 0.26 92.91 0.30

where 1: hemophilic silk carbon; 2: block structure; 3: flowing structure; 4: granular inlay structure; 5: residue

coke; 6: deformed coal; 7: undeformed coal

33
Table 4 Structural parameters of all samples calculated from the XRD spectra

Samples θ(002) FWHM(002) θ(100) FWHM(100) Lc La d002 N


CK-aw 25.733 3.731 42.804 3.731 2.161 4.677 0.346 6.244
GD-aw 25.808 3.944 42.992 3.152 2.044 5.540 0.345 5.924
BD-aw 25.870 3.993 43.066 3.275 2.020 5.333 0.344 5.866
CK 26.464 4.573 43.396 2.330 1.765 7.505 0.337 5.244
CC-aw 24.947 5.134 43.129 4.488 1.568 3.893 0.357 4.394

Table 5 Representative combustion parameters of dusts, CK-aw and CC-aw at three different

heating rates.

(dG/d)vmax (dG/d)vmean 2 2
Samples β(°C/ min) Ti(°C) Tf(°C) S[mg /(min ·°C)]
(mg/min) (mg/min)
2.5 526.9 696.5 0.083 0.0064 2.74714E-12
GD-aw 5.0 539.8 723.3 0.143 0.0059 4.00317E-12
10 578.1 755.1 0.276 0.0060 6.56221E-12
2.5 567.7 683.9 0.094 0.0092 3.92361E-12
BD-aw 5.0 583.3 713.2 0.201 0.0066 5.46695E-12
10 598.2 743.6 0.328 0.0059 7.27267E-12
2.5 554.5 662.3 0.142 0.0066 4.60229E-12
CK-aw 5 574.0 694.6 0.229 0.0057 5.70364E-12
10 595.4 726.7 0.459 0.0067 1.19375E-11
2.5 536.7 685.1 0.155 0.015 1.17816E-11
CC-aw 5 553.2 691.0 0.341 0.031 4.99889E-11
10 570.5 727.6 0.470 0.061 1.21066E-10

34
Table 6 The calculated kinetic parameters of different samples

RPM VM URCM
Samples
E, E, E,
k0, min-1 ψ R2 k0, min-1 R2 k0, min-1 R2
kJ/mol kJ/mol kJ/mol
BD-aw 118.6 3.01E+04 8.13 0.9927 122.79 5.40E+04 0.9934 101.56 2.68E+03 0.9735
GD-aw 141.71 6.36E+05 8.25 0.9995 146.23 1.20E+05 0.9995 120.83 3.41E+04 0.9925
CK-aw 156.07 6.29E+06 5.6 0.9998 162.88 1.63E+06 0.9998 95.28 1.29E+03 0.9677
CC-aw 142.74 1.98E+06 147.8 0.9941 142.88 2.02E+06 0.994 80.63 2.66E+02 0.9263

Table 7 The deviation between experimental and calculated data

DEV(α)%
Samples
RPM VM URCM
GD-aw 3.84 3.71 5.72
BD-aw 1.16 1.02 3.30
CK-aw 0.74 0.57 7.63
CC-aw 3.59 3.60 10.86

35
Highlights
1. The combustion properties of carbonaceous materials in dust were investigated.
2. The mico-structures of carbon materials in dusts were exemplified distinctly.
3. The carbon activity differences were depended on their diverse mico-structures.
4. Kinetic parameters were obtained in the method of nonlinear least-squares fitting.
5. VM and RPM model for carbonaceous materials in dust and their pyrolytic coke.

36

You might also like