Upload
others
View
9
Download
0
Embed Size (px)
Citation preview
PULVERISED COAL INJECTION INTO BLAST FURNACE -
A PRACTICAL STUDY OF AN INTEGRATED IRON AND
STEEL WORKS
S. W. DU
I
PULVERISED COAL INJECTION INTO BLAST FURNACE -
A PRACTICAL STUDY OF AN INTEGRATED IRON AND
STEEL WORKS
A Thesis Submitted in Fulfillment of
the Requirement for the Degree of
Doctor of Philosophy
by
SHAN-WEN DU
(MChE)
Department of Chemical Engineering
The University of Newcastle, Australia
December, 2015
II
I hereby certify that this thesis is submitted in the form of a series of
published papers of which I am a joint author. I have included as part of the
thesis a written statement from each co-author; and endorsed by the Faculty
Assistant Dean, attesting to my contribution to the joint publications.
Signed: _______________________
( Shan-Wen Du)
III
To my family
IV
ACKNOWLEDGEMENTS
First and foremost, I would like to express my sincere gratitude to Professor John Lucas
for his enthusiastic supervision and guidance throughout the period of research. Without
his support and assistance, this work would not have been possible accomplished.
Great thanks go to Professor Terry Wall and Dr. Harold Rogers for their experienced
advice in coal combustion modelling and experiment, to Professor Jian-Long Yu and
Dr. Chatphol Meesri for their encouragements, and to Professor Ai-Bing Yu (New
South Wales University) and Dr, Yan-Song Shen for sharing their research works in
CFD modelling.
I would like to extend my appreciation to Professor Wei-Hsin Chen (National Cheng
Kung University in Taiwan) for his valuable suggestions to coal and biofuel combustion
tests, to Professor Chien-Hsiung Tsai (National Pingtung University of Science and
Technology) for his selfless help in CFD modelling, to Dr. Cheng-Peng Yeh for his
assistance with the prediction of raceway shape, and to Professor Wei-Kao Lu
(McMaster University) for his insightful lectures on blast furnace theory at China Steel
Corporation (CSC).
At CSC, continuous encouragement from Mr. Sing-Tsu Tsai is deeply appreciated. I
would like to thank Mr. Chung-Ken Ho (R&D, BF), Dr. Yung-Chang Ko (R&D, BF),
Mr. Ming-Tsai Hung (R&D, Cokmaking), Dr. Li-Heng Hsieh (R&D Sintering), Mr.
Chi-Sheng Chou (BF) and Mr. Che-Hsiung, Tung (BF) for their challenging comments
and constructive suggestions to this work.
Last but not least, I would like to acknowledge my wife Dr. Shao-Wen Su and my
children Won-Yu, Jia-Yu and Jei-Ruei for their support. They are always my motivation
to finish this thesis.
V
ABSTRACT
The economic benefits of pulverised coal injection (PCI) into blast furnace include a
reduction in the cost of hot metal, resulting primarily from decreased coke consumption
and an increase in hot metal production. Since coal is consumed directly, without going
through the cokemaking plant, PCI is also thought to be environmentally friendly.
Therefore, PCI has become a standard practice in many blast furnaces worldwide. To
improve the performance of PCI operation, a comprehensive understanding of
pulverised coal combustion behaviours are required. The goal of this thesis was to study
the coal burning characteristics in the regions of blowpipe, tuyere and raceway through
both numerical and experimental methodology.
From the validation of model, the calculation region and the application of calculated
results in practice, the PCI combustion model was developed through 4 phases in this
work: (1) validation of the coal combustion model by comparing its predictions with
experimental data; (2) investigation into the influence of operation conditions to coal
burnout in the regions of blowpipe and tuyere; (3) performance evaluation of coal blend
injection in terms of pressure loss caused by combustion within a simplified raceway
space; and (4) examination of combustion characteristics of oxy-coal injection
technology in the regions of blowpipe, tuyere and raceway, which is a porous space
featured by Eulerian-Eulerian multi-fluid approach.
In the first phase, the performance of coal devolatilisation models and kinetic
parameters were validated by comparing predicted gas temperature profiles with the
experimental results of Burgess et al. (1983). It is found that the kinetic parameters
proposed by Ubhayakar et al (1976) for the two competing devolatilisation model
permit a reasonable simulation of the measured results for blast furnace conditions.
The coal combustion in the regions of blowpipe and tuyere was modelled under the
conditions of CSC’s No3 blast furnace in the second phase. The influence of operation
conditions to coal burnout was comprehensively studied. It is found early ignition can
be achieved with higher coal burnout when the double lance is employed instead of the
VI
single one. Accordingly, the injection lance used at CSC was changed from the single to
the double air-cooled coaxial lance arrangement in 2002.
In the third phase, the calculation was extended to the raceway with simplified
configuration. The performance of coal blend injection was examined. As indicated in
the calculation results, a decrease in coal burnout is found with decreasing the coal
volatile content, while the pressure loss within the raceway can be abated due to less
volatile released to gas and moderate gas expansion in the combustion region. With
improved permeability, more hot blast air can be introduced into the blast furnace for
higher productivity. Consequently, the high volatile coal injection was replaced by the
coal blend (mixtures of high and low volatile coals) injection at CSC in 2003.
In the last phase, the Eulerian-Eulerian multi-fluid model was employed for the
prediction of raceway configuration with consideration of coke combustion in all coke
operation. Validation work against measured raceway shape and gas composition
distribution by Nogami et al. (2005) indicates that the model is acceptable for the
simulation. The calculation results show the oxy-coal lance injection enables to fulfil
two contradictory conditions at the same time: (1) to retard the coal combustion for
moderating the pressure loss in the upstream of coal plume; and (2) to enhance coal
combustion and reduce unburnt char generation in the downstream of coal plume.
Taking these advantages from the oxy-coal lance injection, blast furnaces can be
operated with more blast for higher productivity, or with higher PCI rate for lower fuel
cost, thereby achieving the goal of hot metal production with energy saving.
In this work, a drop tube furnace has been established and used to provide fundamental
insights on PCI coal combustion behaviours. The experiments were carried out in three
stages. In the first stage, the volatile release and the generation of char particle and tiny
aerosols in the region of coal plume were studied. Only the tested low volatile coal
(HGI=85) with larger size (100-200 mesh) exhibits fragmentation during heating. This
may encourage the use of low volatile coals in granular coal injection. Significant char
agglomeration is found for both tested high and low volatile coals with smaller size
VII
(200-325 mesh). It implies that excessive grinding may be avoided in PCI operation.
Considering the generation of tiny aerosols composed of soot particles and tar droplets,
it is mainly determined by the content of volatile matter and elemental oxygen.
In the second stage, a technology has been developed and employed at CSC to evaluate
the combustion efficiency of PCI coals. It is found that the coal burnout increases with
decreasing the fuel ratio (FC/VM), except for certain coals departing from the general
trend. It can be explained by the effect of maceral content to coal combustion. When the
coal size is smaller than 200 mesh, the burnout can not be improved further, resulting
from the agglomeration of fine particles. In the PCI operation at CSC, the coal quantity
passing through 200 mesh has been reduced from 80 to 60%.
The experiments for the last stage aim to gain a fundamental insight into the combustion
characteristics of pulverised biofuels under conditions pertinent to the raceway of blast
furnace. From the van Krevelen diagram, it is found that the rate of hydrogen release
from biomass fuels is faster than that of oxygen during the pre-treatment. An increase in
pretreating temperature almost linearly decreases the burnout of biofuels. As revealed in
the experimental results, the fuel properties, such as fuel ratio, burnout, and ignition
temperature, of biomass torrefied at 300 °C or pyrolysed between 400 and 500 °C, are
between a high-volatile bituminous coal and a low-volatile one. Therefore, the
pretreated biomass can partially replace the coals consumed for PCI and blends with
coals to keep reasonable burnout in raceways.
It is emphasised that, due to the objectives of this thesis, some results or
countermeasures obtained from the comprehensive experimental and numerical studies
have been taken into PCI operation at CSC. This seems as a limitation of this study, but
it may have a wide range of applications for the improvement of PCI operation.
,
VIII
LIST OF PUBLICATIONS INCLUDED AS PART OF THE THESIS
(1) Du, S. W., and Chen, W. H. (2006), Numerical prediction and practical
improvement of pulverised coal combustion in blast furnace, International
Communications in Heat and Mass Transfer, 2006, vol. 33, p. 327-334.
(2) Du, S. W., Chen, W. H. and Lucas, J. A. (2007), Performances of pulverised coal
injection in blowpipe and tuyere at various operational conditions, Energy
Conversion and Management, vol. 48, p. 2069-78.
(3) Du, S. W., Yeh, C. M., Yang, M. K. and Ho, C. K. (2004), Practice of high
productivity at No.3 blast furnace of China Steel Corporation", Conference of
Association for Iron and Steel Technology Proceedings (USA), p. 195-204.
(4) Du, S. W., Yeh, C. P., Chen, W. H., Tsai, C. H. and Lucas, J. A. (2015), Burning
characteristics of pulverized coal within blast furnace raceway at various injection
operations and ways of oxygen enrichment, Fuel, vol. 143, p. 98-106.
(5) Chen, W. H., Du, S. W. and Yang, T. H. (2007), Volatile release and particle
formation characteristics of injected pulverised coal in blast furnace, Energy
Conversion and Management, vol. 48, p. 2025-33.
(6) Du, S. W., Chen W. H. and Lucas J. A. (2010), Pulverised coal burnout in blast
furnace simulated by a drop tube furnace, Energy, vol. 35, p. 576-581.
(7) Du, S. W., Chen, W. H. and Lucas, A. J. (2014), Pretreatment of biomass by
torrefaction and carbonization for coal blend used in pulverized coal injection,
Bioresource Technology, vol. 161, p. 333-339.
IX
X
XI
XII
STATEMENT 4
Cheng-Peng Yeh, Wei-Hsin Chen, Chien-Hsiung Tsai and John Lucas attest that
Research Higher Degree candidate Shan-Wen Du contributed to the (1) construction of
calculation model; (2) case studies; and (3) writing to the paper entitled: Burning
characteristics of pulverized coal within blast furnace raceway at various injection
operations and ways of oxygen enrichment, published in Fuel (2015).
Cheng-Peng Yeh
Date:
Wei-Hsin Chen
Date:
Chien-Hsiung Tsai
Date:
John Lucas
Date:
Shan-Wen Du
Date:
Suzanne Ryan (Assistant Dean Research Training)
Date:
XIII
XIV
XV
XVI
LIST OF ADDITIONAL PUBLICATIONS
(1) Du, S. W., Ho, C. K., Tsai, S. T. and Yeh, C. M. (2001), Development of pulverised
coal injection lance at China Steel Corporation”, World Coal, vol. 10, p. 39-42.
(2) Chen, W. H., Du, S. W., Yang, H. H. and Wu, J. S. (2008), Formation
characteristics of aerosol particles from pulverised coal pyrolysis in
high-temperature environments, Journal of the Air & Waste Management
Association, vol. 58, p. 702-710.
(3) Chen, W. H., Du, S. W., Tsai, C. H. and Wang, Z. Y. (2011), Torrefaction of
biomasses in a drop tube furnace to evaluate their utility in blast furnaces”,
Bioresource Technology, vol. 111, p. 433-438.
(4) Yeh, C. P. Du, S. W., Tsai, C. H. and Yang, R. J. (2012), Numerical analysis of
flow and combustion behavior in tuyere and raceway of blast furnace Fueled with
pulverized coal and recycled top gas”, Energy, vol. 42, p. 233-240.
(5) Du, S. W., Ho, C. K. and Tung, C. H. (2013), Numerical Investigations into burning
characteristics of pulverised coal within the BF raceway at various injection lances”,
Proceedings of the Fifth Baosteel Biennial Academic Conference. Shanghai, China,
p. A300-A304.
XVII
CONTENTS
ACKONWLEGEMENTS IV
ABSTRACT V
LIST OF PUBLICATIONS INCLUDED AS PART OF THE THESIS VIII
STATEMENT OF CONTRIBUTION OF OTHERS IX
LIST OF ADDITIONAL PUBLICATIONS XVI
LIST OF FIGURES XXV
LIST OF TABLES XXXII
CHAPTER 1 INTRODUCTION 1
1.1 Background 2
1.2 Auxiliary Fuel injection into blast furnace 5
1.3 Achievement of high coal injection rate 6
1.4 Utilisation of auxiliary fuels at CSC 8
1.5 CO2 emission at CSC 11
1.6 Objectives of the work 11
1.7 Thesis outline 12
CHAPTER 2 LITERATURE REVIEW 15
2.1 Coal combustion experiments under simulating PCI operation
conditions 16
2.1.1 Experiments using empty combustion rig 17
2.1.1.1 Effect of volatile matter content on combustion 17
2.1.1.2 Effect of coal size on combustion and granular coal
injection 19
2.1.1.3 Effect of hot blast conditions on combustion 19
2.1.1.4 Effect of injection rate on combustion 20
2.1.1.5 Effect of lance configurations on combustion 20
XVIII
2.1.1.6 Effect of co-injection on combustion 20
2.1.1.7 Effect of coal blend operation 21
2.1.1.8 Ignition and combustion of volatile matters 21
2.1.1.9 Ash fusion temperature 21
2.1.2 Experiments by coke-packed bed rigs and actual blast furnace 21
2.1.2.1 Effect of lance configuration on combustion 22
2.1.2.2 Effect of hot blast conditions on combustion 23
2.1.2.3 Movement of small coke within raceway 23
2.1.2.4 Effect of volatile content on combustion 24
2.1.2.5 Raceway control during operation at CSC 25
2.1.3 Coal combustion experiments by drop tube furnace 26
2.1.4 Summary of factors affecting coal combustion from
experiments 28
2.2 Modelling of pulverised coal combustion in blast furnace 29
2.2.1 Development of one-dimensional model 29
2.2.1.1 Model of Kuwabara et al. 30
2.2.1.2 Model of Burgess et al. 30
2.2.1.3 Model of He et al. 31
2.2.1.4 Model of Jamaluddin et al. 32
2.2.1.5 Model of Yamagata et al. 33
2.2.1.6 Model of Sato et al. 34
2.2.1.7 Summary of 1-D model 35
2.2.2 Development of two-dimensional model 44
2.2.2.1 Model of Aoki et al. 44
2.2.2.2 Model of Takeda and Lockwood 48
2.2.2.3 Model of Haywood et al. 51
2.2.2.4 Summary of 2-D model 53
2.2.3 Development of 3-D model 54
2.2.3.1 Model of Picard 54
XIX
2.2.3.2 Model of Guo et al. 56
2.2.3.3 models of Shen et al. 60
2.2.3.4 Model of by Gu et al. 68
2.2.3.5 Model of Nogami et al. 70
2.2.3.6 Summary of 3-D model 72
2.3 Sub-models for integrated calculation 73
2.3.1 Devolatilisation of coal 76
2.3.1.1 Single overall reaction model 77
2.3.1.2 Two competing reaction model 77
2.3.2 Char Oxidation 84
2.3.2.1 Field approach 84
2.3.2.2 Gibb Mode 87
2.3.3 Turbulence model 88
2.3.4 Gas combustion in turbulent flow field 89
2.3.4.1 Probability density function (PDF) of turbulence
chemistry 89
2.3.4.2 Eddy break up and eddy dissipation models 90
2.3.5 Lagrangian approach 91
2.3.6 Summary 93
2.4 Raceway shape 93
2.4.1 Observation of raceway 93
2.4.2 CSC’s raceway shape prediction model 96
2.4.2.1 Validation of raceway shape prediction model 99
2.4.2.2 Prediction of raceway shape in an operating blast
furnace 100
2.4.3 Summary of raceway shape prediction 103
2.5 Injection of biofuel into blast furnace 103
2.5.1 Combustion experiments and modelling 104
2.5.2 Summary of the biofuel injection 111
XX
2.6 Summary 111
2.7 Methodology 111
CHAPTER 3 NUMERICAL PREDICTION AND PRACTICAL
IMPROVEMENT OF PULVERIZED COAL
COMBUSTION IN BLAST FURNACE
113
Abstract 114
Nomenclature 115
Greek symbols 115
Subscripts 115
3.1 Introduction 116
3.2 Mathematical Formulation 117
3.2.1 Burning process of pulverized coal 117
3.2.2 Momentum and energy balance of a coal particle 117
3.2.3 Model of devolatilisation of coal particle 118
3.2.4 Turbulent combustion model 119
3.3 Results and discussion 120
3.3.1 Numerical validation and parameter selection 120
3.3.2 Impact of injection pattern 123
3.3.3 Practical improvement of blast furnace 125
3.4 Conclusions 126
CHAPTER 4 PERFORMANCES OF PULVERIZED COAL
INJECTION IN BLOWPIPE AND TUYERE AT
VARIOUS OPERATIONAL
127
Abstract 128
Nomenclature 129
Greek symbols 129
Subscripts 130
4.1 Introduction 131
XXI
4.2 Methodology 132
4.2.1 Gas-phase continuity and momentum equations 133
4.2.2 Coal particle momentum and energy equations 135
4.3 Results and discussion 138
4.3.1 Trajectories and residence times of coal particles 140
4.3.2 Injection pattern 140
4.3.3 Oxygen concentration and hot blast temperature 143
4.3.4 Hot blast temperature 144
4.3.5 Mass flow rate of carrier gas 144
4.3.6 Installation of ceramic sleeve 145
4.4 Conclusions 146
CHAPTER 5 PRACTICE OF HIGH PRODUCTIVITY AT NO
3 BLAST
FURNACE OF CHINA STEEL CORPORATION 148
Abstract 149
5.1 Introduction 150
5.2 Development of low flux sinter 150
5.3 Establishment of burden terrace 152
5.4 Development of one bit drilling method 154
5.5 Coal blend injection 155
5.5.1 Analysis of permeability of the furnace 155
5.5.2 Coal Combustion model within tuyere-raceway area 157
5.5.3 Calculation results and discussion 161
5.5.4 Plant trial of coal blend injection 162
5.6 Increase of hot metal production in No3 blast furnace 163
5.7 Conclusions 164
CHAPTER 6 BURNING CHARACTERISTICS OF PULVERIZED
COAL WITHIN BLAST FURNACE RACEWAY AT
VARIOUS INJECTION OPERATIONS AND WAYS OF
OXYGEN ENRICHMENT
165
XXII
Abstract 166
Nomenclature 167
Greek symbols 167
Subscripts 168
6.1 Introduction 169
6.2 Methodology 171
6.2.1 Gas-particle flow 171
6.2.1.1 Gas phase 171
6.2.1.2 Single particle in dispersed phase 172
6.2.1.3 Turbulence model 172
6.2.2 Turbulent combustion 173
6.2.3 Devolatilization of coal 174
6.2.4 Physical geometry and operating conditions 175
6.3 Results and discussion 177
6.3.1 Trajectories of coal particles 177
6.3.2 Oxygen consumption within the combustion region 179
6.3.3 Ignition and temperature distribution 181
6.3.4 Combustion efficiency of coal particles 184
6.3.5 Pressure loss 185
6.4 Conclusions 186
CHAPTER 7 VOLATILE RELEASE AND PARTICLE FORMATION
CHARACTERISTICS OF INJECTED PULVERIZED
COAL IN BLAST FURNACES
188
Abstract 189
7.1 Introduction 190
7.2 Experiment 192
7.3 Results and discussion 195
7.3.1 Devolatilisation extent 195
XXIII
7.3.2 Particle formation of the low-volatile bituminous coal 196
7.3.3 Particle formation of the high-volatile bituminous coal 199
7.3.4 Aerosol formation and reactivity 204
7.3.5 Reactivity of char and soot 204
7.4 Conclusions 207
CHAPTER 8 PULVERIZED COAL BURNOUT IN BLAST FURNACE
SIMULATED BY A DROP TUBE FURNACE 208
Abstract 209
8.1 Introduction 210
8.2 Experiments 212
8.2.1 Reaction system 212
8.2.2 Experimental procedure and conditions 214
8.3 Results and discussion 216
8.3.1 Combustion efficiency of individual coals 216
8.3.2 Influences of reaction temperature and particle size 219
8.3.3 Burnout of blended coals 221
8.4 Conclusions 223
CHAPTER 9 PRETREATMENT OF BIOMASS BY
TORREFACTION AND CARBONIZATION FOR COAL
BLEND USED IN PULVERIZED COAL INJECTION
224
Abstract 225
9.1 Introduction 226
9.2 Experimental 228
9.2.1 Materials and preparation 228
9.2.2 Burnout and ignition tests 230
9.3 Results and discussion 231
9.3.1 Proximate analysis and van Krevelen diagram 231
XXIV
9.3.2 Solid yield and energy yield 234
9.3.3 Ignition and burnout 238
9.4 Conclusions 241
CHAPTER 10 CONCLUSIONS AND RECOMMENDATIONS 243
10.1 Introduction 244
10.2 Achievements and conclusions 244
10.2.1 Modelling 244
10.2.2 Coal combustion experiments 247
10.3 Recommendations 249
10.3.1 Raceway control 249
10.3.2 Improvement of permeability by charging pattern 252
REFERENCES 254
XXV
LIST OF FIGURES
Figure 1.1 Schematic diagram of internal structure with five district zones
in a blast furnace. 2
Figure 1.2 Reduction reactions in blast furnace (Biswas, 1981). 3
Figure 1.3 Change in blast furnace operating conditions in Japan
(Ariyama et al, 2007). 6
Figure 1.4 Changes and problems of blast operation with high injection
rate. 9
Figure 2.1 Schematic of combustion rig attached with an empty
combustion chamber. 17
Figure 2.2 Influence of blast temperature on Q-factor. 18
Figure 2.3 Degree of combustion of the coals as a function of injection
rate. 20
Figure 2.4 Schematic of combustion rig attached with a coke bed. 22
Figure 2.5 Effect of lance arrangement on pulverised coal flow and
combustion efficiency by hot model (η (%): combustion at 300 and
600 mm from lance tip). 23
Figure 2.6 Change of coal burnout at the height of 700mm above tuyere
level with change of volatile content in coals. 24
Figure 2.7 Image of coal flow patterns at CSC. (a) single lance; (b) double
lance injection. 25
Figure 2.8 Image of blockage of tuyere by un-melted scab. 26
Figure 2.8 Schematic representation of a DTF. 27
Figure 2.9 Mechanism of pulverised coal combustion. 27
Figure 2.10 Critical operational factors on coal burnout. 28
Figure 2.11 Schematic view of combustion in front of a tuyere. 30
Figure 2.12 Schematic view of combustion with the injection of
pulverised coal. 31
Figure 2.13 Gas flow, entrainment and disentrainment patterns inside the
raceway. 32
Figure 2.14 Relationship between coal burnout and injection rate. 33
Figure 2.15 Modelling concept of pulverised coal flows in blow pipe. 34
Figure 2.16 Influence of injection lance on coal burnout along tuyere axis. 34
XXVI
Figure 2.17 Calculation domain in the region of blowpipe-tuyere. 45
Figure 2.18 Effect of coal volatile content on gas distribution in the
blowpipe. 46
Figure 2.19 Calculated particle trajectories of low volatile coal injection in
the blowpipe. 46
Figure 2.20 Calculated particle trajectories of high volatile coal injection
in the blowpipe. 47
Figure 2.21 Distributions of gas composition and gas temperature along
the centre line of tuyere. 47
Figure 2.22 Schematic representation of raceway structure used in the
simulation. 48
Figure 2.23 Contours of gas temperature and oxygen concentration in the
regions of blowpipe and raceway. 49
Figure 2.24 Lance design for coal injection. 50
Figure 2.25 Coal burnout comparisons for the various modifications of
lance. 51
Figure 2.26 Schematic of the experimental combustion rig. 52
Figure 2.27 Oxygen mass fraction (top) and gas temperature contours
(bottom) in the near injector region. low volatile coal (a) low volatile
coal; (b) high volatile coal injection. 52
Figure 2.28 3D meshed calculation domain. 55
Figure 2.29 Trajectories of coal particles within the tuyere and raceway. 55
Figure 2.30 Evolution of coal residence time within the raceway versus
particle diameter. 55
Figure 2.31 Main dimensions (in mm) of the coal combustion model (plan
view). 57
Figure 2.32 Gas velocity vectors in Y-Z plane (a) vector length to scale,
and (b) vector normalised showing recirculation zone. 57
Figure 2.33 Typical particle trajectories with colour scaled to particle size. 58
Figure 2.34 Gas species fraction isopleths in Y–Z plane. (a) Oxygen; (b)
volatile matter. 58
Figure 2.35 Calculated mass fraction of volatiles (a) and calculated mean
coal burnout (b) as a function of distance from the lance tip along the
centreline averaged for different particle sizes (in μm). 59
Figure 2.36 Comparison of burnout evolutions predicted by the previous
model (Case 1), and present model with char gasification reactions
(Case 2) and without char gasification reactions (Case 3). 62
XXVII
Figure 2.37 Burnout for Cases I, II and III: (a) along the centreline and (b)
at the exit. 62
Figure 2.38 Effect of blast temperature on burnouts at the distances of 300
mm and 925 mm from the lance tip, respectively. 63
Figure 2.39 Effect of oxygen enrichment on coal burnout 63
Figure 2.40 Effect of cooling gas type: (a) final burnout and (b) O2
distributions at the cross-plane of 550 mm from the lance tip. 64
Figure 2.41 Geometry of the model: (a), the whole model; (b), porosity
distribution (Zone 0: 1, Zone 1: 0.25, Zone 2: 0.5, Zone 3: 0.4); (c),
blowpipe and raceway; and (d), lance tip. The detailed dimensions are,
(1) for blowpipe, radius: 90 mm, and length: 800 mm; (2) for tuyere,
radius: 75/90 mm, and length: 135 mm; (3) for raceway, depth: 1600
mm, height: 1000 mm (925 + 75), and width: 710 mm; and (4) for
coke bed, depth: 3700 mm, height: 4500 mm, and width: 1000 mm. 66
Figure 2.42 Flow pattern of gas-particle flow: (a), vectors of gas phase in
the raceway; (b), streamlines of gas flow; (c), particle trajectories
coloured by particle mean size; and (d), particle trajectories coloured
by particle travelling time. 67
Figure 2.43 Combustion characteristics of coal along particle trajectories
in the coke bed. 67
Figure 2.44 Schematic of computational domain: (a) side view; (b) top
view. 69
Figure 2.45 Distributions of (a) gas velocity vectors and (b) gas
temperature (K) in the computational domain. 69
Figure 2.46 Coal burnout at the exit of the computational zone. 70
Figure 2.47 Schematic figure of hot model. 71
Figure 2.48 Comparison of calculated raceway shape with observation of
test. (a) All coke operation. (b) PCI operation. 71
Figure 2.49 Characteristics of raceway: (a) calculated raceway shape, and
(b) calculated gas velocity vectors. 72
Figure 2.50 Illustration of combustion phenomena of pulverised coal
(Ishii, 2000). 74
Figure 2.51 Framework of the CFD code and computational procedure of
the gas phase and solid (coal particle) phase (Du, et al., 2007). 75
Figure 2.52 Size effects on physical and temperature profile (Smoot and
Smith, 1985) 76
Figure 2.53 Schematic of the Blowpipe/Tuyere (combustion test section)
assembly of the pilot scale raceway hot model. 81
XXVIII
Figure 2.54 Gas temperature distributions for pulverised coal burning in a
reactor from experimental measurement and numerical predictions
using different devolatilisation models. 81
Figure 2.55 Comparison of the relationships between Y1 and Y2 in the
literature (Shen et al., 2008). 82
Figure 2.56 Rate-controlling regimes for char reactions (Smoot and
Smith, 1985). 85
Figure 2.57 Schematic illustration of raceway structure. 94
Figure 2.58 Representation of the movement of coke through the raceway. 95
Figure 2.59 Comparison of numerical and experimental results (Nogami
et al., 2004) of: (a) raceway shape; and (b) gas composition
distribution along central axial of tuyere. 100
Figure 2.60 Profile of CSC’s No3 blast furnace: (a) main dimensions
(unit: m); (b) calculation domain of a single tuyere. 101
Figure 2.61 Void fraction contours in combustion zone of 3D coke
packed furnace model: (a) top view; (b) side view. 102
Figure 2.62 The simplified calculation domain. Note that αg is the volume
fraction of gas inside the raceway. 103
Figure 2.63 Burnouts as a function of volatile matter of the injectants with
an air cooled lance and O/C = 2.0. Comparison is made with previous
results for PCI coals. 106
Figure 2.64 Differential pressure across the tuyere as a function of the
volatile matter of the injectants. 106
Figure 2.65 Schematic of the reaction system (1) cylinder; (2) carrier gas;
(3) secondary gas; (4) rotameter; (5) hopper; (6) preheater; (7) lance;
(8) DTF; (9) thermocouple; (10) ceramic tube; (11) heater; (12)
sampling probe; (13) cooling water; (14)cyclone; (15) residual solid
particles; (16) induced suction fan; (17) exhausted gas. 108
Figure 2.66 Distributions of burnout versus fuel ratio of raw and torrefied
biomasses as well as a HV coal. 109
Figure 2.67 Geometry and computational domain used in numerical
simulation. 110
Figure 2.68 Temperature profiles at an injection rate of 36 (kg solid fuel) /
(1000 Nm3 feed gas). 110
Figure 3.1 A schematic diagram of internal structure in a blast furnace. 118
Figure 3.2 A comparison of gas temperature distribution among
experimental measurement and two devolatilization models. 122
Figure 3.3 A schematic diagram of blowpipe and tuyere as well as their
sizes. 124
XXIX
Figure 3.4 Isothermal contours in blowpipe and tuyere under the
operations of (a) single lance and (b) double-lance injections. 125
Figure 4.1 A schematic diagram of internal structure in blast furnace. 133
Figure 4.2 Framework of the CFD code and computational procedure of
the gas phase and solid (coal particle) phase. 137
Figure 4.3 Gas temperature distributions for pulverize coal burning in a
reactor from experimental measurement and numerical predictions. 138
Figure 4.4 Trajectories and residence times of coal particles under the
operation of the base case. 140
Figure 4.5 Burning ratios of PC at various injection patterns. 142
Figure 4.6 Combustion situations of pulverized coal in (a) case 1 and (b)
case 3. 142
Figure 4.7 Burning ratios of pulverized coal at various oxygen
concentrations. 143
Figure 4.8 Burning ratios of pulverized coal at various hot blast
temperatures. 144
Figure 4.9 Burning ratios of pulverized coal at various mass flow rates of
carrier gas. 145
Figure 5.1 AE sensor system for measuring burden falling point. 153
Figure 5.2 Burden profile before (a)/ after (b) changing charging pattern. 154
Figure 5.3 Typical pressure distribution of No 3 blast furnace. 156
Figure 5.4 Physical geometry of combustion region. 157
Figure 5.5 Trajectories and residence time of coal particles in the
combustion region. 161
Figure 5.6 Oxygen concentration contour at cross section along
combustion region. 162
Figure 5.7 Pressure distribution along combustion region from lance exit. 162
Figure 6.1 Schematics of (a) physical sizes of computational domain as
well as the arrangements of (b) CSC’s double air-cooled lance and (c)
single and oxy-coal lance (α is the porosity within the raceway). 177
Figure 6.2 Distributions of coal particle trajectory and residence time
under (a) single lance, (b) double air-cooled lance, and (c) oxy-coal
lance injections. 178
Figure 6.3 Distributions of oxygen mole fraction under (a) single lance,
(b) double air-cooled lance, and (c) oxy-coal lance injections 180
Figure 6.4 Distributions of hydrogen mole fraction under (a) single lance,
(b) double air-cooled lance, and (c) oxy-coal lance injections. 181
XXX
Figure 6.5 Distributions of isothermal contours under (a) single lance, (b)
double air-cooled lance, and (c) oxy-coal lance injections. 183
Figure 6.6 Distributions of gas temperature along the centreline of tuyere
under single lance injection and oxy-coal lance injections at different
proportions of enriched oxygen. 183
Figure 6.7 A comparison of coal combustion efficiency among single
lance injection, double air-cooled lance injection, and oxy-coal lance
injections with different proportions of enriched oxygen. 185
Figure 6.8 A comparison of pressure loss among single lance injection,
double air-cooled lance injection, and oxy-coal lance injections with
different proportions of enriched oxygen. 186
Figure 7.1 Schematic diagram of pulverized coal injection and internal
structure of blast furnace around raceway. 192
Figure 7.2 Profiles of R-factor of two different coals at various reaction
temperatures. 196
Figure 7.3 Particle size distributions of coal F before and after
experiencing reactions with (a) larger feed particles and (b) smaller
feed particles 197
Figure 7.4 Peak locations of particle size distributions for coal F before
and after experiencing reactions. 198
Figure 7.5 SEM images of unburned chars of coal F at larger feed
particles (a-c) and smaller feed particles (d-f). 198
Figure 7.6 Particle size distributions of coal L before and after reactions
with (a) larger feed particles and (b) smaller feed particles. 200
Figure 7.7 Peak locations of particle size distributions for coal L before
and after experiencing reactions. 201
Figure 7.8 SEM images of unburned chars of coal L at larger feed
particles (a-c) and smaller feed particles (d-f). 201
Figure 7.9 SEM images of feed coal and unburned char particles shown in
cross-section. 203
Figure 7.10 Profits of soot and tar formations with respect to reaction
temperature. 205
Figure 7.11 Thermogravimetric analyses of the produced unburned chars
and soots at 1400oC. 205
Figure 8.1 Schematic of the reaction system. 213
Figure 8.2 Tests of experimental stability of four different coals under the
base experimental conditions. 215
Figure 8.3 Correlation between burnout and fuel ratio under the standard
combustion conditions. 218
XXXI
Figure 8.4 Distributions of burnout of Coal C, Coal D and Coal I at
various reaction temperatures. 219
Figure 8.5 Distributions of burnout of Coal B, Coal E and Coal I at
various particle sizes. 221
Figure 8.6 Distributions of burnout with respect to blending ratio for Coal
K individually blended with Coal A and Coal C. 222
Figure 9.1 (a) Volatile matter, (b) fixed carbon, and (c) fuel ratio values of
raw and pretreated biomass materials. 233
Figure 9.2 Atomic H/C versus O/C ratio (van Krevelen diagram) of raw
and pretreated biomass materials. 234
Figure 9.3 (a) HHVs and replacement factors of biomass materials based
on (b) Coal A and (c) Coal B. 237
Figure 9.4 (a) Solid yield and (b) energy yield of pretreated biomasses
materials. 238
Figure 9.5 Ignition temperatures of raw and pretreated biomass. 240
Figure 9.6 Burnout versus (a) pretreated temperature and (b) fuel ratio. 241
Figure 10.1 Zonal structures in a drill core. 250
Figure 10.2 Formation of cavity and bird’s nest in a pack bed. 251
Figure 10.3 The measured burden profiles and calculated descending rate
at No3 blast furnace. 253
XXXII
LIST OF TABLES
Table 1.1 Number of blast furnaces worldwide equipped with substitute
fuel injection (Schott, 2013). 6
Table 1.2 Average fuel rates of the blast furnaces in the EU 15 (Peters and
Bodo, 2009). 8
Table 1.3 General features of CSC’s blast furnaces. 10
Table 2.1 PCI combustion models and sub models used. 36
Table 2.2 Operational conditions selected by Burgess et al. (1983). 80
Table 2.3 Three sets of parameters used for predicting PC devolatilisation. 80
Table 2.4 Kinetics of single overall reaction and two competing reaction
models used in the PCI calculation models. 83
Table 2.5 Kinetics of char oxidation employed in the PCI combustion
models. 86
Table 2.6 Comparison of raceway observations in cold and hot models. 95
Table 2.7 Chemical reactions in coke–packed furnace model. 99
Table 2.8 Operating conditions and properties of coke–packed bed. 100
Table 2.9 Typical operating parameters of No3 BF for all coke operation 101
Table 2.10 Key properties of the bulk coal and charcoal samples. 105
Table 2.11 Key properties of the bulk coal and charcoal samples. 105
Table 2.12 Enhancement factor of higher heating value. 108
Table 2.13 Computational conditions for biofuel injection. 110
Table 3.1 Operational conditions selected by Burgess et al. (1983) 121
Table 3.2 Two sets of parameters used for predicting PC devolatilization. 122
Table 3.3 Operating conditions of PCI at CSC. 124
Table 4.1 Operating conditions (base case) of PCI at CSC. 139
Table 5.1 Main features of CSC’s No3 blast furnace. 150
Table 5.2 Typical energy consumption in the late period of the first
campaign. 152
Table 5.3 Reduction of SiO2 in sinter and slag volume. 152
Table 5.4 Variation of wall heat loss. 154
XXXIII
Table 5.5 Comparison between soaking bar tapping and one bit drilling. 155
Table 5.6 Parameters of devolatilisation kinetics.
160
Table 5.7 PCI Operation condition used in the calculation. 160
Table 5.8 Changes of pressure drop and permeability for coal blend
injection. 163
Table 5.9 Hot metal Production in CSC’s No 3 blast furnace. 163
Table 6.1 A list of fuel properties and operating conditions. 176
Table 7.1 Proximate and ultimate analyses of the investigated coals 195
Table 7.2 Summary of reaction physics of the two coals 206
Table 8.1 Proximate analyses (dry basis), fuel ratios and higher heating
values (HHV, dry basis) of the investigated coals. 216
Table 8.2 Maceral analyses of the investigated coals. 218
Table 9.1 Proximate, elemental, fiber, and calorific analyses of two coals
and raw biomass materials. 229
Table 9.2 Ash contents in pretreated biomass materials. 236
1
CHAPTER 1
INTRODUCTION
Pulverised coal injection technology is considered as an effective countermeasure to
reduce operation cost and CO2 emission of blast furnace. The injection rates achieved
by the blast furnaces worldwide are reviewed in this chapter. The development of
auxiliary fuel injection at China Steel Corporation is reported. The objectives and
outline of this work are briefly presented.
2
1.1 Background
Blast furnaces (BF) are currently a crucial and the most commonly employed facility in
ironmaking processes. It is also predicted that the blast furnace will remain the
successful process for hot metal production for the coming 30 years due to its
advantages in high productivity and heat utilisation (Geerds et al., 2011).
From the thermal point of view, the blast furnace works as a counter-current reactor in
which solids descend and gases ascend. Practically, iron oxides (i.e. sinter, pellet and
lump ore) and coke (reductant) are charged through the rotating chute with discrete
dumps to form alternating layers in the furnace. When descending, the charged iron
oxides are heated and reduced by ascending gases from tuyeres installed in the low zone
of blast furnace. Softening and melting of reduced iron and gangue materials begin
when the temperatures are high enough. Notably, the coke remains in solid state and
descents to the furnace hearth constructing a porous coke bed named deadman. The
generated liquid iron (hot metal) and slag trickle through the deadman to the hearth
bottom of the blast furnace. The hot metal and slag are cast into a main through when
the taphole is opened. For the generation of thermal energy and reducing gases required
for hot metal production, blast air heated to the temperature of 1100-1250°C is
introduced at a velocity around 180 m/s into the furnace through the tuyeres in the lower
zone of blast furnace. Consequently, a cavity call a raceway is formed in front of the
tuyere exit. It is found that the raceways of adjoining tuyeres are not connected with
each other (Nakamura et al., 1978).
According to the reaction characteristics, the entire blast furnace can be generally
divided into five individual zones (Figure 1.1) from the top downwards:
(1) lumpy zone for pre-reduction and reduction of iron oxides;
(2) cohesive zone for reduced iron and gangue materials fusing and melting;
(3) dripping zone for gas and liquid refining;
(4) raceway for carbon gasification and heat generation; and
(5) deadman zone for irrigated bed slag/metal refining.
3
Clearly, the five reaction zones are effectively integrated into a single shell in the blast
furnace ironmaking process. As a result, the blast furnace is still competitive for hot
metal production even though many new ironmaking technologies have been studied
and developed.
Figure 1.1 Schematic diagram of internal structure with five district zones in a
blast furnace.
Oxides: sinter, pellet and lump ore
Reductant: coke
Top gas: CO, CO2,
H2, N2, H2O
Blowpipe and
tuyere
Hot blast air
and PCI
Taphole
Lumpy zone with coke
and ferrous layers
Cohesive zone
Raceway
Dripping zone
Deadman
Hot metal and slag
4
An idealised blast furnace reduction process can be found in Figure 1.2 (Biswas, 1981).
Generally, the reduction of iron oxides can be catalogued into indirect reduction and
direct reduction, which occur at the upper and lower zones of a BF, respectively. In
indirect reduction process, iron oxides, including Fe2O3, Fe3O4 and FeO, are
exothermically reduced by reducing gases (CO and H2) into other iron oxides (Fe3O4
and FeO), resulting in CO2 and H2O as products. These normally occur at temperatures
below 850-900oC. In direct reduction at the low zone of blast furnace, iron and CO are
produced by carbon (from coke) reacting endothermically with iron oxides above
900oC. It is noted that hydrogen regeneration reaction (C + H2O = CO + H2) is less
endothermic and proceeds faster than the carbon monoxide regeneration loss reaction (C
+ CO2 = 2CO). For a blast furnace with all coke operation, partial replacement of coke
by hydrogen containing fuels may be an effective way for the improvement of blast
furnace performance. .
Figure 1.2 Reduction reactions in blast furnace (Biswas, 1981).
5
1.2 Auxiliary fuel injection into blast furnace
In the blast furnace ironmaking process, coke plays a particularly important role for
stable operation, because it provides mechanical support for the descending materials
and ensures permeability for the ascending gas within the cohesive zone, and for the hot
metal within the deadman zone. Besides, it reacts with hot blast air in the raceway to
generate energy and reducing gases for the hot metal production. Because the
metallurgical coals for cokemaking are expensive, the reduction of coke consumption is
always desirable for the blast furnace operation. With the developments in high
temperature stoves, big oxygen plants (providing large amount inexpensive oxygen),
measurements and operation models, the injection of auxiliary fuels, such as natural gas,
oil and pulverised coal, into the blast furnace as a substitute of coke in the raceway has
become an industrial reality. Apart from the economic benefit, the advantages below are
generally expected from the auxiliary fuel injection:
(1) to increase the productivity of the blast furnace, resulting from more oxygen being
added in the blast for maintaining the flame temperature of raceway;
(2) to smooth down the burden movement to assist in maintaining furnace stability; and
(3) to prolong the life span of coke battery since less coke has to be produced.
Due to ease of operation, natural gas followed by oil were popular injectants. Although
the adoption of the pulverised coal injection (PCI) technology was discouraged initially
primarily because there was no strong economic incentive before second oil crisis in
1978, some significant achievements on PCI operation have been reached. For instance,
Armco installed a commercialised PCI system at the Ashland, Kentucky plant in 1963
(Nolde et al., 1996), and the Shoudu Iron and Steel Corporation in China first achieved
a successful PCI practice in 1964 (Wei and Qi, 1983). After the crises, many companies
stopped injecting oil into blast furnaces and turned to PCI operation, because the coal
preparation and pneumatic transportation have become proven technologies by that
time. As shown in Table 1.1 (Schott, 2012), nearly half of all blast furnaces in the world
6
(47.7%) use PCI; while only 11.9% inject gas or oil, and 35.9 % remain all coke
operation because the blast furnaces are rather small or old, making cost effectiveness of
an application of substitute fuel injection questionable.
Table 1.1 Number of blast furnaces worldwide equipped with substitute fuel injection
(Schott, 2012).
PC
Oil Gas
All Coke Plastics
Oil with gas gas with PC
Africa 2 1 4 2
America 36 3 7 11 17 26
Asia 278 4 1 220
Australia 2 1 2
Europe 74 15 4 61 3 47 2
Total, % 47.7 4.1 11.9 35.9 0.2
1.3 Achievement of high coal injection rate
In the beginning of the adoption of PCI operation, it was predicted that the PCI rate
might be limited to 60 kilograms per tonne hot metal (Gudenaua and Kiesler, 1991) or
to about 15% of the fuel rate of blast furnace (Poos and Ponghis, 1990). Efforts towards
enhancing the PCI operation have been made by the blast furnaces worldwide for higher
coal injection rates.
As can be seen in Figure 1.3 (Ariyama et al., 2007), pulverised coal injection into blast
furnaces in Japan began in 1983, and the average pulverized coal injection rate achieved
by the Japanese blast furnaces has been gradually increased to 140 kg/tHM in 1999.
Since then, the coal injection rates were remained in a range of 120 to 135 kg/tHM.
Notably, a remarkable injection rate of 230 kg/tHM and monthly average of 218
kg/tHM have been attained by NKK’s Fukuyama No4 BF (Maki et al., 1996). In Korea,
POSCO’s Gwangyang No4 BF (inner volume: 5500 m
3) has reached a PCI rate of 200
7
kg/tHM with a low coke rate of 290 kg/tHM. This achievement has been set as the
operation target for the big blast furnaces (inner volume >4000 m3) in China (Zhang et
al., 2013). Besides, it should be noted that Baosteel No4 BF only spent one month after
blow in operation to reach a high PCI level of over 200 kg/tHM (Li et al., 2007).
In the 1990s, there were still 45 integrated works operated in whole Europe. Since then
blast furnace works have been shut down and partly been replaced by electric arc
furnaces (Peters and Lüngen, 2009). During this period, three European blast furnaces,
Thyssen Schwelgern No1 BF, Sidmar B BF and Hoogovens N
o7 BF, recorded high PCI
rates of 178, 197 and 199 kg/tHM respectively with low coke rate (Buss et al., 2000;
Peters et al., 1991). In 2008, the average PCI rate of the blast furnaces in the EU 15 is
123.9 kg/tHM, while extraordinary operation modes with PCI rate, as well as low coke
consumption, have been reached at some blast furnaces as indicated in Table 1.2 (Peters
and Lüngen, 2009). It is found that highest coal rate was realised at Tata Corus
(formerly Hoogoven) No6 with 235.1 kg/tHM as yearly average. The lowest coke rate
with 289.9 kg/tHM with a resulting total fuel rate of 516.9 kg/tHM was achieved at this
furnace. As shown in the Table, the lowest total reductant rate was achieved at Ruukki
blast furnace No1 (458.5 kg/t HM) in oil injecting operation mode.
Figure 1.3 Change in blast furnace operating conditions in Japan (Ariyama et al,
2007).
8
1.4 Utilisation of auxiliary fuels at CSC
China Steel Corporation (CSC) group is the only integrated and the largest steel
producer in Taiwan. It has 4 blast furnaces (No1 to 4) on its major production side in
Kaohsiung, and 2 blast furnaces (No5 and 6) in Taichung, with a designed annual hot
metal output of approximately 15 million metric tonnes. The general features of CSC’s
blast furnaces are listed in Table 1.3. In general, the hot metal production consumes the
most energy of an integrated steelwork (Babich et al., 2002). At CSC, more than 55% of
the entire energy is consumed by the blast furnace ironmaking process. Therefore many
technologies, such as auxiliary fuel injection and top gas recovery turbine (TRT), have
been applied for the reduction of energy consumption at the blast furnace process of
CSC.
Before the second oil crisis, oil was injected through tuyeres into CSC’s blast furnaces
as the substitute of coke. The injection rates were in a range of 50 to 60 kg/tHM.
Table 1.2 Average fuel rates of the blast furnaces in the EU 15 (Peters and Lüngen,
2009).
9
Concerned by the uncertain oil supply and substantial increase in prices after the crisis,
CSC introduced PCI technology to its blast furnaces in 1987 (Du et al., 2001).
To make the injected coals ignited earlier in the tuyere for higher combustibility, high
volatile coals (VM>35%) were solely injected through single lances into the blast
furnaces since CSC commenced its PCI operation. For the reduction of the unburnt char
generated in the raceway, double air-cooled coaxial lance was developed and applied at
CSC in 2001 (Du et al., 2001; Yeh et al., 2002; Du and Chen, 2006). To improve the
permeability in the lower zone of blast furnace, as well as to diversify the coal types for
the blast furnace injection, low volatile coals were blended into high ones in the PCI
operation of CSC in 2003 (Du et al., 2004). At present, the proximate volatile matter of
coal blend for injection operation at CSC’s blast furnaces is generally kept in the range
of 19 to 21%.
Apart from its economic benefit from the replacement of expensive coking coals by
cheaper thermal coals, PCI operation is also thought to be environmental friendly at
CSC since the injected coals are consumed directly, without going through the
coke-making plant. Therefore, high PCI rate is one of operation targets of CSC’s blast
furnaces. It should be emphasised that the blast furnace ironmaking is a complicated
process, so the replacement of coke with pulverised coal is not as simple as just
increasing the injection rates or improving the combustibility of coal injected, especially
when one considers the impact of coal combustion to the stability of raceway, fuel rates
and permeability of furnace. Figure 1.4 shows the impact of high coal injection rate on
blast furnace operation (Ishii, 2000). It suggests in high PCI rate, knowledge of the
details in the combustion region becomes more critical for high PCI rate. A brief
understanding on the coal combustion behaviours in the regions of blowpipe, tuyere and
raceway can lead to a more effective and safe operation with high PCI rate.
10
Table 1.3 General features of CSC’s blast furnaces.
BF No1 N
o2 N
o3 N
o4 N
o5 N
o6
Inner volume, m3 2624 3274 3606 3422 3274 3274
Hearth diameter, m 10.2 12 12.5 12.5 12 12
Cooling Stave Stave Stave Plate Stave Stave
Number of tapholes 2 2 4 4 3 3
Number of tuyeres 30 30 32 32 32 32
Blow-in/ campaign 2010
(4th
)
2006
(3rd
)
2009
(2nd
)
2014
(2nd
)
2010
(1st)
2013
(1st)
First campaign 1977 1982 1987 1996 2010 2013
Figure 1.4 Changes and problems of blast operation with high injection rate.
11
1.5 CO2 emission at CSC
The greenhouse gas of most relevance to the world steel industry is carbon dioxide. As
a matter of fact, CO2 liberated from blast furnaces approximately account for 3.5–5% of
total anthropogenic CO2 emissions (Wang et al., 2009). On average, 2.1 tonnes of CO2
are emitted for every tonne of crude steel produced at CSC. The utilisation of
alternatives to coal in PCI can abate the consumption of fossil fuels and, in the case of
biofuel or charcoal, mitigate CO2 emissions. It was reported by Babich et al. (2010) that
the injection of charcoal fines has been successfully practiced in some small charcoal
blast furnaces in Brazil with injection rates of 100 to 150 kg/tHM.
For the reduction of CO2 emission, a pilot plant for producing biofuel from local
agricultural wastes has been established in Malaysia by CSC. To successfully use
biofuel as partial replacement for pulverised coals through injection, it is required to
examine the fuel properties of biomass pretreated by torrefaction and carbonisation and
compare to those of high-volatile and low-volatile coals.
1.6 Objectives of the work
To provide brief insights into the coal burning characteristics in the lower zone of blast
furnace, one of objectives of this study is focused on the development of a 3-D CFD
based model to predict the coal combustion phenomena occurring in the regions of
blowpipe, tuyere and raceway, identification of the major operation parameters affecting
the coal combustion and the permeability of raceway, and application of the calculation
results to the PCI operation of CSC’s blast furnaces.
Although the combustion conditions within a drop tube furnace are less intense than
those in the region of blowpipe-tuyere-raceway, the drop tube furnace may provide a
better view at the combustion properties of coal without influences from gas flow
pattern, coal injection rate and char sampling. A drop tube furnace has been established
in this research to evaluate the combustion performance of PCI coals, coal blend and
12
biofuels in an environment with high heating rate (>104 K/s). Besides, the coal
combustion characteristics, such as volatile release, swelling and agglomeration, are
also examined using the furnace.
1.7 Thesis outline
The thesis begins with brief introductions to blast furnace ironmaking process, PCI rates
achieved in blast furnaces worldwide and the general features of PCI operation in the
blast furnaces of CSC. A review of the relevant literature is made in Chapter 2. The
development and application of the 3D CFD based coal combustion model are
presented in Chapters 3 to 6. In this work, the development of the model can be divided
into 4 phases: (1) validation of the coal combustion model by comparing its predictions
with experimental data; (2) investigation into the influence of operation conditions to
the coal burnout in the region of blowpipe and tuyere; (3) performance evaluation of
coal blend injection in terms of pressure loss due to coal combustion within a simplified
raceway space; and (4) examination of burning characteristics of pulverised coal with
different ways oxygen enrichment in the regions of blowpipe, tuyere and raceway,
which is a porous space featured by Eulerian-Eulerian multi-fluid approach.
The publications presented in this thesis are listed below:
Chapter 3
Du, S. W. and Chen, W. H. (2006), Numerical prediction and practical improvement of
pulverized coal combustion in blast furnace, International Communications in Heat and
Mass Transfer, vol. 33, p. 327-334.
Chapter 4
Du, S. W., Chen, W. H. and Lucas, A. J. (2007), Performances of pulverized coal
injection in blowpipe and tuyere at various operational conditions, Energy Conversion
and Management, vol. 48, p. 2969-78.
13
Chapter 5
Du, S. W., Yeh, C. M., Yang, M. K. and Ho C. K. (2004), Practice of high productivity
at No.3 blast furnace of China Steel Corporation, Proceedings of AISTech Conference,
Tennessee, USA, p. 195-204.
Chapter 6
Du, S. W., Yeh, C. P., Chen, W. H., Tsai, C. H. and Lucas, J. A. (2015), Burning
characteristics of pulverized coal within blast furnace raceway at various injection
operations and ways of oxygen enrichment, Fuel, vol. 143, p. 98-106.
By using the drop tube furnace established at CSC, the research is designed to (1) study
the volatile release and particle formation within the coal plume, in which the oxygen is
insufficient; (2) establish a method to evaluate the coal combustion efficiency for the
selection of PCI coal at CSC; and (3) examine the combustion of bio fuels pretreated by
torrefaction and carbonisation for their utilisation in PCI operation. The experimental
results and findings are presented in Chapters 7 to 9 by the publications below:
Chapter 7
Chen, W. H., Du, S. W. and Yang, T. H. (2007), Volatile release and particle formation
characteristics of injected pulverized coal in blast furnace, 2007, Energy Conversion
and Management, vol. 48, p. 2025-33.
Chapter 8
Du, S. W., Chen, W. H. and Lucas, A. J. (2010), Pulverized coal burnout in blast
furnace simulated by a drop tube furnace, Energy, vol. 35, p. 576-581.
Chapter 9
Du, S. W., Chen, W. H. and Lucas, A. J. (2014), Pretreatment of biomass by
torrefaction and carbonization for coal blend used in pulverized coal injection,
Bioresource Technology, vol. 161, p. 333-339.
14
Finally, Chapter 10 summaries the main conclusions from this work, followed by
recommendations for future work on the subject.
15
CHAPTER 2
LITERATURE REVIEW
In this chapter, a comprehensive review of the previous experimental and numerical
studies on the pulverised coal combustion in the blast furnace is presented. An overview
of factors affecting coal combustion is made. It also discusses the application of sub
models in the integrated coal combustion models. The CSC’s internal research works,
including the validation of devolatilisation models and the prediction of raceway shape
are briefly described in this chapter. The application of biofuel in the PCI operation is
outlined. Finally, the methodology of the work is presented.
16
2.1 Coal combustion experiments under simulating PCI operation conditions
Combustion process of injected coal in the blowpipe-tuyere-raceway region can be
characterised by (Steiler et al., 1996):
(1) high heating rate (>104 K/s) of injected coal due to very high blast temperature
(1050-1250oC);
(2) short residence time for injected coal in the combustion area (<20 ms) caused by
high blast velocity (>160m/s); and
(3) insufficient mixing of injected coal particles with hot blast gas (Du et al., 2007;
Goto et al., 2002).
It is predicted that as PCI rate is increased, more amount of unburnt char is generated
and accumulated in the blast furnace due to incomplete conversion of injected coal,
leading to dirtying of the deadman and finally decrease in the furnace productivity and
increase in the fuel rate. Therefore, considerable attention has been focussed on the
improvement of coal combustion efficiency for the PCI operation.
For studying the factors influencing coal combustion during PCI operation, many coal
combustion experiments have been carried out by using combustion rigs with empty
combustion chambers (Guo et al., 2011; Mathieson et al., 2005; Du et al., 2001; Picard
et al., 2000; Kim et al., 1996; Babich et al., 1996; Steeghs et al., 1996; Ueno et al.,
1993; Suzuki et al., 1984; Burgess et al., 1983; Bortz and Flament, 1983; Narita et al.,
1982;), or, in some cases, with coke bed reactors (Nogami et al., 2004; Ariyama et al.,
1994; Sato et al., 1994; Yamagata et al., 1992; Yamaguchi et al., 1992). Practically, the
operation conditions of the rigs, such as blast air (temperature, oxygen level and
velocity), lance configuration and coal flow rate, can be set as closely as possible to
those of actual blast furnaces. Notably, some operation difficulties, including pulverised
coal preparation, char sampling, temperature measurement and keeping a constant coal
injection rate, may be encountered with the combustion rigs (Du et al., 2010). Despite
the inherent difference between the realistic combustion environment of the raceway
17
Figure 2.1 Schematic of combustion rig attached with an empty combustion chamber.
and the drop tube furnace, the drop tube furnace is considered as an effective device
when one attempts to evaluate the combustion performance of PCI coals in an
environment with high heating rate (Du et al., 2010). The sections below will
summarise the findings of the coal combustion experiments.
2.1.1 Experiments using empty combustion rig
Figure 2.1 shows a combustion rig connected with an empty combustion chamber
(simulated as the raceway) developed by BlueScope Steel Research (Mathieson et al.,
2005). Due to operational ease in comparison with the coke bed reactors, empty
combustion rigs have been preferred for practising pulverised coal combustion
experiments. A summary of the effects of operation parameters and coal properties to
the coal combustion obtained from those experimental works is presented as followed:
2.1.1.1 Effect of volatile matter content on combustion
(1) The burnout of injected coal was strongly dependent on the coal rank. High volatile
coals were burnt preferentially to low volatile coals (Vamvuka et al., 1996; Steeghs
et al., 1996; Ueno et al., 1993; Malgarini, 1991; Suzuki et al., 1984; Wakimoto et
al., 1983; Burgess et al., 1983; Bortz and Flament, 1983). However, high volatile
18
matter content may not be sufficient to characterise coal combustion efficiency in
the blast furnace operation, because the formation of tar and soot generated from the
released volatile (Chen et al., 2008; Bortz and Flament, 1983) and the pressure loss
due to the combustion within the raceway (Du et al., 2004; Du et al., 2015) should
be considered as the operation challenges with high coal injection or high blast air
flow rate.
(2) Some low volatile coals showed higher combustion efficiency than that expected
from their volatile levels (Mathieson et al., 2005). It was explained by the formation
of fragments from chars, leading to the increase of surface area for combustion (Li
et al., 2014).
(3) Under nitrogen atmosphere, the final volatile yield of injected coal was higher than
the proximate analysis (Q factor >1), and the yield increased with increasing the
blast temperature as indicated in Figure2.2. Moreover, the coal particle started to
swell significantly when the weight loss of coal was higher than 20%. With 51%
weight loss, the diameter of coal could be 1.3 times higher than the initial size
(Ueno et al., 1992). The Q factor is the ratio of the measured volatile content in high
heating rates to the proximate volatile matter.
Q factor: ratio of
final volatile yield
to that by
proximate analysis
Figure 2.2 Influence of blast temperature on Q-factor.
19
2.1.1.2 Effect of coal size on combustion and granular coal injection
(1) The burnout of injected coal increased with decreasing the coal particle size
(Wakimoto et al., 1983; Bortz et al., 1983; Narita et al., 1982). Alternatively, it was
found in the research work by Du et al. (2010) that the coal combustion could not
be improved further as a result of the agglomeration effect when the feed particle
size was smaller than 75 μm.
(2) To get higher combustion efficiency, many blast furnaces inject pulverised coal
(80% < 75 m) (Hutny et al., 1991). On the other hand, British Steel injects
granular coal (100% < 5mm, and 95% < 2mm) into the blast furnace (Gathergood
and Jukes, 1996). A sampling probe has been developed by Guo et al. (2012) to
take granular coal particles from the raceway of an operating blast furnace (Lai-Wu
Steel in China). Owing to rapid release of moisture and volatile matters, significant
cracking of injected granular coal was observed through microscope structure
analysis. Consequently, coal combustion was boosted with the increase of reaction
surface area by coal cracking. Practically, No1 blast furnace of Lai-Wu Steel
adopted granular coal injection (GCI) in practice in 2005, and an average injection
rate of 168 kg/tHM was achieved in 2012 (Zhu and Xu, 2014). However, injecting
granular coal into blast furnace has not yet been widely practised by the
international steel producers.
2.1.1.3 Effect of hot blast conditions on combustion
(1) It was reported (Guo et al., 2011; Du et al., 2001; Ueno et al., 1992) the increase in
oxygen content and temperature of blast were the effective countermeasures for
improving the coal combustion efficiency.
(2) Increasing blast pressure up to 3 kg/cm2 promoted combustion efficiency markedly
(Wakimoto et al., 1983).
20
2.1.1.4 Effect of injection rate on combustion
As indicated in Figure2.3 (Vamvuka et al., 1996), the coal burnout was decreased with
the increase of coal injection rate (Vamvuka et al., 1996; McCarthy et al., 1986; Suzuki
et al., 1984). This implies the increase of unburnt char generated in the raceway is
unavoidable when one attempts to promote the PCI rate to a certain amount.
2.1.1.5 Effect of lance configurations on combustion
The double lance was found to have the highest combustion efficiency followed by
oxy-coal lance and single one (Du et al., 2001).
2.1.1.6 Effect of co-injection on combustion
The coal combustion was improved when MgO was added in amounts of 3-4% to the
PCI coal (Guo et al., 2011). On the other hand, it was reported by Vamvuka et al. (1996)
that an increase of additive amount of shredded light fractions (from scrap automobiles)
resulted in a decrease in the coal burnout. Besides, by injecting coke oven gas to an
outer peripheral area of the pulverised coal plume, the combustion efficiency of coal
was improved significantly (Suzuki et al., 1990).
Figure 2.3 Degree of combustion of the coals as a function of injection rate.
21
2.1.1.7 Effect of coal blend operation
Adding a high volatile coal to a low volatile coal, a linear relationship of combustion
efficiency was reported by Yu (1999). In some experiments of POSCO, the combustion
efficiency of coal blend is even higher than that of high volatile coal (Kim et al., 1996).
Despite the increase in the generation of unburnt char, the pressure loss caused by the
coal combustion in the lower zone of blast furnace could be abated when a low volatile
coal was blended into a high one (Du et al., 2004).
2.1.1.8 Ignition and combustion of volatile matters
(1) The ignition of injected coal within a tuyere was observed by using an endoscope
(Picard et al., 2000). It was found inflammation never occurred when an 11% VM
content coal was injected; at 23% VM, a flame was not present at every trial; at 30%
VM, a flame attached to the lance was always observed.
(2) During the coal combustion, there were two zones observed in terms of temperature
in the flame (Wakimoto et al., 1983). The first zone (higher heating rate) close to the
tuyere was for volatile combustion, and second zone was for char combustion. It is
similar with the calculated temperature profile in a combustion chamber (Du and
Chen, 2006).
2.1.1.9 Ash fusion temperature
From the amount of slag/ash deposits in blowpipe, it suggested that high ash fusibility
(low ash fusion temperature) was a desirable characteristic of coals for PCI (Scaife,
1983; Ostrowski, 1983).
2.1.2 Experiments by coke-packed bed rigs and actual blast furnace
For simulating the raceway of blast furnace, the coke-packed rigs have been operated
under blast furnace operation conditions. Figure 2.4 shows a coke-packed combustion
rig developed by NKK (Ariyama et al., 1994).
22
The main findings obtained from the rigs or actual furnaces are summarised below.
2.1.2.1 Effect of lance configuration on combustion
(1) From the images taken by a high speed camera in the blowpipe, the pulverised coal
combustion was not uniform across the blowpipe with operating a single lance. The
injected coal was likely to move as a group of particles, and formed the luminous
flame around it. This suggested that the hydrocarbons from coal pyrolysis were
decomposed to generate soot (Sato et al., 1994).
(2) Compared with single lance operation, double lance operation revealed a relatively
smooth change of brightness measured by a high speed camera, and the average
brightness level was higher than that given by the single lance operation.
Consequently, higher combustion efficiency could be reached by using the eccentric
double lance (Sato et al., 1998) as shown in Figure 2.5. In practice, NKK has been
using the eccentric double-lance arrangement at its Fukuyama No4 BF, and an
injection rate of 218 kg/tHM was announced by the furnace (Maki et al., 1996).
Figure 2.4 Schematic of combustion rig attached with a coke bed.
23
(3) An oxy-coal lance, with a coal flow in the inner pipe and an oxygen flow in the
annulus of the coaxial lance, was tested in a single tuyere at an actual blast furnace
(Wikström et al., 1996). The extension of the mixing and the combustion zones
were measured by a thermovision camera. The combustion of coal was found to be
improved markedly by using the coaxial lance as compared with single lance
operation.
2.1.2.2 Effect of hot blast conditions on combustion
The increase of blast temperature and oxygen content in hot blast gas were effective
countermeasures for improving the coal burnout in the raceway, and an optimum
oxygen enrichment of 4% for hot blast gas was suggested by Yamaguchi et al. (1992).
2.1.2.3 Movement of small coke within raceway
Upon inspection of the raceway coke shell (bird nests) taken from the rig, it was
concluded by Nogami et al. (2004) that the movement of small coke particles contracted
by gasification almost determined the formation of raceway shell. Small particles
delivered by gas flow to the raceway boundary did not return inside raceway space and
Figure 2.5 Effect of lance arrangement on pulverised coal flow and combustion
efficiency by hot model (η (%): combustion at 300 and 600 mm from lance tip).
24
enter into the packed beds. Then, most of them lost the momentum by collision with
coke in packed beds and accumulated on the surface domain of the packed beds with
decreasing their mass by solution-loss reaction.
2.1.2.4 Effect of volatile content on combustion
Measurements by a sideways tuyere probe in the raceway of an actual furnace show the
combustibility of injected coal was improved with the increase in volatile matter
content (Takeda et al., 1990). This is similar to the experimental results given by the
empty combustion rigs.
From the combustion experiments with use of a coke-packed hot model (Yamagata et
al., 1992), the combustion efficiency of coal decreased with the decrease of volatile
matter in tuyere level. On the other hand, under 200kg/tHM injection rate, the coal
burnouts measured at the height of 700mm above tuyere were at high level of over 95%
for all tested coals with different volatile contents (18.9-44.9%) as indicated in Figure
2.6. The reaction of char with CO2 (carbon solution loss reaction: C + CO2 = 2CO) was
thought to be predominant after the char particles leave the raceway.
Figure 2.6 Change of coal burnout at the height of 700mm above tuyere level with
change of volatile content in coals.
25
2.1.2.5 Raceway control during operation at CSC
As indicated in Figure 2.7a taken by the tuyere monitoring system, the dispersion of
coal plume to hot blast gas within the raceway was poor with the single lance (Du et al.,
2001). Obviously, the contact area between coal particles and hot blast gas could be
increased when the double lance is employed (Figure 2.7b). To prevent the coal plume
from being forced backwards into the bustle pipe, leading to possibility of explosion,
coal injection should be stopped at the tuyere blocked by un-melted scab as shown in
Figure 2.8 (Ho and Du, 2008).
Figure 2.7 Image of coal flow patterns at CSC. (a) single lance; (b) double lance
injection.
(a)
(b)
26
2.1.3 Coal combustion experiments by drop tube furnace
As a matter of fact, the PCI coal combustion experiments by the combustion rig
with/without coke bed are manpower and cost consuming process. Alternatively, a drop
tube furnace (Figure 2.8) can be used to simulate the situation of fuel particles suddenly
exposed to a high-temperature environment with heating rates ranged from 104 to10
5
K/sec.
The experimental results given by the furnace are indicated below:
(1) Through optical pyrometer observation (Figure 2.9), two-step combustion of single
coal particle was found; a first and very fast step of volatile combustion followed by
a much slower step of char combustion (de Lassat de Pressigny et al., 1990).
Devolatilisation experiments performed with coals with different volatile levels at
1200oC in a nitrogen atmosphere showed the ultimate volatile released was close to
1.7 times higher than that given by ASTM standard method (de Lassat de Pressigny
et al., 1990).
(2) The performance of a combustion rig and a drop tube furnace was compared by Li et
al (2014). It was found the measured burnouts from both the drop tube furnace and
the rig (Mathieson et al., 2005) produced approximately linear trends as a function
Figure 2.8 Image of blockage of tuyere by un-melted scab.
Lance
Un-melted
scab
Lance Coal
plumes
27
of coal volatile matter. In addition, the coal burnout in drop tube furnace was more
sensitive to the coal volatile matter than that in the rig, resulting from differences in
operation, temperature, residence time, and heating rate between the drop tube
furnace and the PCI rig.
Figure 2.8 Schematic representation of a DTF.
Figure 2.9 Mechanism of pulverised coal combustion.
28
2.1.4 Summary of factors affecting coal combustion from experiments
Based on the experimental results, some key findings for the coal combustion in blast
furnace can be drawn:
(1) The factors, which inference the coal burnout significantly, can be summarised into
three categories as shown in Figure 2.10:
(2) From the temperature profile, the combustion of injected coal in the raceway can be
characterised into two stages: (1) a rapid rise in gas temperature caused by the
combustion of volatile matters emitted from the injected coal, and (2) a slow process
of char reactions.
(3) Due to very short residence time for coal particles in the raceway, complete
combustion of the injected coal in the raceway is quite unlikely in the blast furnace
operation. Notably, because char gasification proceeds much faster than that of coke
(Iwanaga, 1991; de Lassat de Pressigny et al., 1990), the permeability of gas and
liquid in the furnace may not be significantly affected by the generation of unburnt
Blast conditions
1. Temperature
2. Oxygen enrichment
3. Pressure
Coal burnout
Coal properties
1. Volatile content
2. Particle size
3. Coal blend
4. Cracking
Injecting facilities
1. Configuration of
lance (mixing of coal
with blast or oxygen)
2. Secondary fuel
injection
3. Injecting additives
Figure 2.10 Critical operational factors on coal burnout.
29
char as long as the consumption rate of unburnt char exceeds its accumulation rate in
the furnace.
(4) With rapid heating environment at blast furnace tuyere, the reactions in the raceway
include fast volatile release and combustion followed by slow char reactions.
Moreover, the final (or ultimate) volatile yield of injected coal in the raceway is
higher than that by proximate analysis. It suggests the conversion of injected coal in
the raceway may come largely from the coal devolatilisation. Furthermore, the
raceway combustion behaviours are more likely to be dominated by the release and
combustion of volatile matter rather than by the gasification of char and coke.
2.2 Modelling of pulverised coal combustion in blast furnace
As indicated above, experimental combustion rigs and drop tube furnaces have been
established and applied for the investigation of pulverised coal combustion behaviour in
the blast furnace. It should be noted the experimental studies mainly focused on the
evaluation of operation parameters to coal burnout. Alternatively, numerical models,
including Computational Fluid Dynamics (CFD), are able to analyse the physical fluid
system of raceway more cost effectively and rapidly than the experimental procedures.
From the perspective of mathematical modelling, the pulverised coal combustion
process, involving turbulent multiphase flow coupled with momentum, mass and heat
transfer, and various homogenous and heterogeneous chemical reactions, can be studied
comprehensively.
2.2.1 Development of one-dimensional model
In the 1980s, one-dimensional (1-D) mathematical models were developed to study the
combustion behaviours in raceway for all coke operation, or for PCI operation.
Common features of the 1-D models include: (1) the injected coal particles were
dispersed uniformly with the hot blast gas; and (2) the raceway was assumed as
30
cylindrical and non-spreading jet, and (3) gas leaved from the main flow zone through
the roof of the cylindrical path. A brief discussion on the key models is presented
below.
2.2.1.1 Model of Kuwabara et al.
Based on the equations of motion and continuity of gas and coke, a 1-D kinetic model
was developed to describe the behaviour of coke combustion in front of a tuyere area
without injecting auxiliary fuel (Kuwabara et al., 1981). According to the observation
using an endoscope by Greuel et al (1974), the raceway was assumed to comprise a
cylindrical jetting space which had the same diameter as the tuyere exit as shown in
Figure 2.11. With the increase of oxygen level from 21% to 23%, early ignition of coke
in the raceway was found, resulting in shifting the peak levels of gas temperature and
concentration of CO2 towards the tuyere exit.
2.2.1.2 Model of Burgess et al.
A plug flow model was developed to simulate the fuel-lean combustion of pulverised
coal in the blowpipe (Burgess et al, 1983). The coal particle velocity was calculated by
Tuyere
Figure 2.11 Schematic view of combustion in front of a tuyere.
31
the equation of motion of the particle. From the sensitivity tests of the model, blast
temperature and the volatile content of injected coal, as compared with blast velocity,
coal size, and oxygen content in blast, showed a more effect on coal burnout. Due to the
assumption of instantaneous mixing of coal particles with hot blast gas, the calculated
gas temperatures in the blowpipe were significantly higher than the measured ones.
2.2.1.3 Model of He et al.
Competitive combustion of coke and pulverised coal in the raceway was considered in
the 1-D model developed by He et al. (1986). As indicated in Figure 2.12, the raceway
was modelled as a cylindrical jet surrounded by coke bed. Heterogeneous reactions of
coke and pulverised coal with O2, H2O and CO2 were taken into account. The model
results showed that (1) an increase in blast temperature resulted in increases in the
amount of released volatile and gas temperature; (2) higher burnout could be given by
the coal with higher volatile content; (3) when pulverised coal was injected, the peak
temperature of the raceway was slightly lower than that of all coke operation; and (4)
the peak temperature of gas in the raceway moved closer to the tuyere exit when
operating higher PCI rate.
Figure 2.12 Schematic view of combustion with the injection of pulverised coal.
32
2.2.1.4 Model of Jamaluddin et al.
In the model developed by Jamaluddin et al. (1986), the flow in the region of
blowpipe-tuyere was divided into three regions, the potential core (coal flow), the main
flow (blast), and mixing layer in which the injected coal particles and the blast were
mixed uniformly and the heating and pyrolysis of the injected coal particles took place.
The area of mixing layer grew in cross-section. The particle velocity was calculated
from the equation of motion of the particle. The char was treated as pure carbon and its
combustion proceeded in two steps, the first one char was oxidised by O2, CO2, H2O, O
and OH to produce CO and H2, and second constituting the oxidation of the primary
combustion products to CO2 and H2O. The raceway combustion was treated as a 1-D
process. The model assumed that falling cokes with recirculated combustion gas entered
the raceway from the roof of raceway as shown in Figure 2.13. Notably, the model is
the first published model to consider raceway with recirculated gas.
The calculated results showed the predicted gas temperature and concentration were
sensitive to (1) combustion of gases recirculated from the upper part of the raceway
Figure 2.13 Gas flow, entrainment and disentrainment patterns inside the raceway.
33
cavity, (2) combustion or gasification of the coke particles falling into the raceway, and
(3) reactions at the coke bed forming the raceway boundary.
2.2.1.5 Model of Yamagata et al.
A calculation model was developed by Yamagata et al. (1994) to examine the effects of
the rate of PCI up to 300 kg/tHM, coal volatile content and particle size on combustion
in the region of blowpipe-raceway. Both gas and solid phase were modelled as one
dimensional flow with complete mixing in the blowpipe and raceway cavity. This model
consisted of the equations of continuity and motion for gas and coal particles. Collision
of fine particles with packed bed was also considered. The heterogeneous reactions of O2,
CO2 and H2O with char and coke particles were controlled by mass transfer in the gas
film and chemical reactions. As can be seen in Figure 2.14, the coal burnout was
decreased with an increase in injection rate, leading to more unburnt char generated in
the raceway.
Figure 2.14 Relationship between coal burnout and injection rate.
34
2.2.1.6 Model of Sato et al.
Sato et al. (1996) introduced a dispersion angle of coal plume in the blowpipe as
indicated in Figure 2.15. In the model, the dispersion angle increased with an increase in
lance diameter. Furthermore, the dispersion of multiple lances in the blowpipe was
simulated by that of single one with larger diameter. In Figure 2.16, the performance of
double lance and single lance injection in terms of coal burnout were compared.
Obviously, the double lance operation achieved higher coal burnout than that given by
the single one due to better dispersion of coal particles with hot blast gas.
Figure 2.15 Modelling concept of pulverised coal flows in blow pipe.
Figure 2.16 Influence of injection lance on coal burnout along tuyere axis.
35
2.2.1.7 Summary of 1-D model
Table 2.1 shows a summary of key 1-D models and the sub-models applied. A brief
summary of the calculation results obtained from the 1-D model is shown below.
(1) The calculated peak levels of gas temperature and concentration of CO2 are shifted
towards the tuyere exit with an increase of PCI rate. Practically, for the prevention of
tuyere from burn out, the lance tip should be moved towards the tuyere exit when the
coal injection rate is significantly increased.
(2) Double lance is more effective than single one in mixing injected coal particles with
hot blast gas. Consequently, the coal burnout can be improved.
It is noted that many important operational features, especially relating to the flow
spatial properties of the process, are not considered in the 1-D models. For example, in
the 1-D model, the injected coal particles were assumed to be dispersed uniformly, or
dispersed with a growth angle in the hot blast gas. In fact, the assumptions are quite
different from the observations given by the tuyere monitoring system (as shown in
Figure 2.7). Practically, accurate information about the dispersion of injected coal is
required for the PCI operation, because it is critical not only to control the
devolatilisation of injected coal particles in the raceway, but also to protect tuyeres from
failure due to PCI abrasion. Therefore the practical application of the 1-D models in PCI
operation may be limited.
36
Table 2.1 PCI combustion models and sub models used.
Author(s) Kuwabara et al. (1981) Burgess et al. (1983)
Flows patterns 1-D flow 1-D flow
Coal particles are dispersed
uniformly with the blast gas
Calculation
domain
Blowpipe, tuyere and raceway
(cylindrical jetting space)
Cylindrical combustion
chamber
Devolatilisation - Two competing reaction model
Volatile Reaction - Instantaneous combustion
Char Reaction - C + 1/2 O2 = CO (combustion
heat to solid phase)
CO + 1/2 O2 =CO2
(combustion heat to gas phase)
Coke Reaction C+ O2 CO2
C+CO2 2CO
C +H2OH2 + CO
Coke temperature is assumed
to be 0.8 time of gas
temperature
-
Raceway shape As a cylindrical jetting space
with the same diameter of
tuyere and 1.5meters in length
Voidage: 0.6
Diameter of coke: 20 mm
-
Sensitivity
Analysis
Blast conditions (O2, moisture,
temperature)
Blast conditions, coal type,
coal size, injection rate.
Validation Measured data by Inatani et
al. (1973)
Measured temperatures from
combustion rig
Remarks Without injecting auxiliary
fuel
Raceway not included
Author(s) He et al. (1986) Jamaluddin et al. (1986)
37
Table 2.1 contd.
Flows patterns 1-D flow
Injected coal dispersed
uniformly with the blast air
Quasi-two-dimensional flow
Modelled as a combination of
plug flow regions, viz., coal
flow, blast, and mixing layer
which is quantified through a
jet-spread angle
Calculation
domain
Blowpipe, tuyere and raceway
(cylindrical jetting space)
Coke and injected coal
dispersed uniformly with the
blast air
Modelled as a one dimensional
jet with recirculating gas and
coke particles entrained at the
regions of 0.2 – 0.73 depth of
raceway roof
Devolatilisation First order reaction
Two-competing reaction model
Volatile reaction Instantaneous combustion Instantaneous combustion
Char reaction C + 1/2 O2 CO
C + CO2 2CO
C + H2O H2 + CO
C + 1/2 O2 CO
C + CO2 2CO
C + H2O CO + H2
C + O CO
C + OH CO + 1/2 H2
Coke reaction in
raceway
C + 1/2 O2 CO
C + CO2 2CO
C + H2O H2 + CO
C + 1/2 O2 CO
C + CO2 2CO
C + H2O CO + H2
Raceway shape Cylindrical jet with spreading
gas through coke wall
Coke diameter: 20mm
Proposed as a pipe with
non-smooth wall (coke bed)
Coke diameter: 30 mm
Sensitivity
Analysis
Blast conditions, injection rate,
coal properties,
Blast conditions, injection rate,
soot conc., coal properties
Validation Measured data by Ariyama et
al. (1994)
Measured data by Inatani et al.
(1973)
Remarks The model comprises warm-up
zone and burning zone
The first model considering
recirculating gas entering the
raceway
38
Table 2.1 contd.
Author(s) Yamagata et al. ( 1992 ) Sato et al. (1996)
Flow patterns 1-D flow
Injected coal dispersed
uniformly with the blast air
Quasi-two-dimensional flow
Coal plume is dispersed along
an angle determined by the
lance diameter
Calculation
domain
Blowpipe tuyere and raceway
(cylindrical jetting space)
Coke and injected coal
dispersed uniformly with the
blast air
Blowpipe and coke bed
(cylindrical space)
Injected coal dispersed into a
packed bed (the void fraction of
the raceway is linearly
predictable)
Devolatilisation Overall reaction model:
DV/dt = k(Vo-V)0.5
K=4.0 105 exp(-17500/RT)
Single overall reaction model
Volatile reaction - n≦2m+2
CmHn + m/2O2 = mCO + n/2H2
n>2m+2
CmH2m+2 +m/2O2 = mCO +
(m+1)H2
Char reaction C + 1/2 O2 CO
C + CO2 2CO
C + H2O CO + H2
C + 1/2 O2 CO
C + CO2 2CO
C + H2O CO + H2
Coke reaction in
raceway
C + 1/2 O2 CO
C + CO2 2CO
C + H2O CO + H2
C + 1/2 O2 CO
C + CO2 2CO
C + H2O CO + H2
Raceway shape 480mm in depth Cylindrical jet
Sensitivity
Analysis
Injection rate, volatile content Multiple lance and oxy-coal
lance injection
Validation Measured burnout and
gas composition from a coke
bed
Measured dada by Fukuyama
No5 blast furnace
Remarks The first 1D model available
for the evaluation of double
lance
The first 1D model available
for the evaluation of multiple
lance and oxy-coal lance
39
Table 2.1 contd.
Author(s) Aoki et al. (1993) Takeda and Lockwood (1997)
Flow patterns Axi-symmetric 2-D flow
Standard k- turbulent model
for gas phase
Lagrangian approach for the
trajectory of coal particles
Eulerian approach for coke
Axi-symmetric 2-D flow
Modified k- turbulent model
for gas phase
Lagrangian calculation for the
trajectory of coal particles
Calculation
domain
Blowpipe tuyere and raceway
determined by the force
balance between gas and solid
Blowpipe, tuyere and raceway,
including a jet zone and
transition zone
Devolatilisation Two competing reaction model Single overall reaction model
Q factor: 1.5
Volatile reaction Eddy dissipation model Eddy dissipation model
Char reaction C + O2 CO2
C + 1/2 O2 CO
C + CO2 2CO
C + H2O H2 + CO
C + 1/2 O2 CO
C + CO2 2CO
Coke reaction in
raceway
C + O2 CO2
C + 1/2 O2 CO
C + CO2 2CO
C + H2O H2 + CO
C + O2 CO2
C + CO2 2CO
C + H2O H2 + CO
Raceway shape Determined by force balance
among gravity force of coke,
gas-solid interaction force and
fractional force of coke at the
boundary of raceway
Including jet zone and
transition zone
Sensitivity
Analysis
Influence of coal size and
volatile content on the
dispersion of coal plume
Ways of oxygen enrichment,
lance configuration, coal size
Validation Raceway shape by cold model Raceway gas compositions
measured form an operating
furnace
Remarks Residence time of PC in the
combustion region: 12 ms
A high turbulent lance applied
at Kawasaki Steel
40
Table 2.1 contd.
Author(s) Haywood et al. (1994) Picard (2001)
Flow patterns Axi-symmetric 2-D flow
Standard k- turbulent model
for gas phase
Lagrangian approach for
calculating trajectory of coal
particles
3-D CFD based model
Standard k- turbulent model
for gas phase
Lagrangian approach for
calculating trajectory of coal
particles
Calculation
domain
A cylindrical combustion
chamber
Tuyere, empty raceway and
porosity media surrounding the
raceway
Devolatilisation Two competing reaction model Two competing reaction model
Volatile reaction Mass fraction/ PDF model Chemical reaction rate
(Arrhenius)
Char reaction C + 1/2 O2 CO
C + CO2 2CO
C + H2O H2 + CO
C + 1/2 O2 CO
Limited either by chemical
kinetic or diffusion of oxygen
Coke reaction in
raceway
Raceway shape Only blowpipe Measurements results from
blast furnaces
Sensitivity
Analysis
Injection rate and coal volatile
content
Particle size, coal VM, oxygen
enrichment, injection rate
Validation Gas composition measured
from a combustion rig
Remarks One of pioneers in PCI
combustion model by CFD
(1)Raceway shape featured by
tuyere coke sampling
(2)One of pioneers in 3D PCI
combustion model
41
Table 2.1 contd.
Author(s) Guo et al. (2005) Shen et al. (2008, 2009a,
2009b)
Flow patterns 3-D CFD based model
RNG k–e turbulence model for
gas phase
Lagrangian approach for
calculating trajectory of coal
particles
3-D CFD based model
Standard k- turbulent model
for gas phase
Lagrangian approach for
calculating trajectory of coal
particles
Calculation
domain
blowpipe and combustion
chamber
blowpipe and combustion
chamber
Devolatilisation Two competing reaction model Two competing reaction model
Stoichiometric parameters
given by the tests of Australian
coals with VM ranged from
12-40%
Volatile reaction eddy break up model eddy dissipation model
Char reaction ΦC + O2 → 2(Φ-1) CO + (2-Φ)
CO2
Φ is function of particle
temperature (Gibb, 1985)
ΦC + O2 → 2(Φ-1) CO + (2-Φ)
CO2
CO2 2CO
C + H2O H2 + CO
Coke reaction in
raceway
Raceway shape Combustion chamber Only blowpipe
Sensitivity
Analysis
Coal properties (size, VM and
flow rate), lance configuration,
cooling gas types for co-axial
lance
Coal properties (size, VM and
flow rate), oxygen level in blast
gas, lance configuration,
cooling gas types for co-axial
lance, coal blend
Validation Burnout given by the
combustion tests (Mathieson, et
al, 2005)
Measured burnout given by the
combustion tests (Mathieson, et
al, 2005)
Remarks (1) evaluation of oxy-coal
lance
(2) fine particles dispersed
more widely
42
Table 2.1 contd.
Author(s) Shen et al. (2011) Gu et al. (2008)
Flow patterns 3-D CFD based model
Standard k–εturbulence model
for gas phase
Lagrangian approach for
calculating trajectory of coal
particles
3-D CFD based model
k--kp two-phase turbulence
model for gas and particle
phases
Calculation
domain
(1) Blowpipe, tuyere and
raceway; (2)deadman;
(3)dripping zone; and (4)
cohesive zone
Blowpipe and tuyere
Devolatilisation Two competing reaction model Two competing reaction model
Volatile reaction Eddy dissipation model Eddy break up model
Char reaction ΦC + O2 → 2(Φ-1) CO + (2-Φ)
CO2
C + CO2 2CO
C + H2O H2 + CO
C + O2 CO2
2C + O2 2CO
C + CO2 2CO
C + H2O H2 + CO
Coke reaction in
raceway
C + CO2 2CO
Raceway shape Designed in a shape of balloon Determined by CFD
calculation (Selvarasu et al.,
2006)
Sensitivity
Analysis
Coal properties (size, VM and
flow rate), oxygen level in blast
gas, lance configuration,
cooling gas types for co-axial
lance
Tuyere diameter
Validation Measured burnout given by the
combustion tests (Mathieson, et
al, 2005)
Remarks Recirculation of fine particles
(<70um) within raceway
Fine particles dispersed more
widely
43
Table 2.1 contd.
Author(s) Nogami et al. (2004)
Flow patterns 3-D CFD based model
Standard k–εturbulence model
for gas and solid phases
DEM for coke movement
Calculation
domain
Blowpipe, tuyere and raceway
Devolatilisation Two competing reaction model
Volatile reaction Eddy dissipation model
Char reaction C + O2 CO2
Coke reaction in
raceway
C + O2 CO2
C + CO2 2CO
C + H2O H2 + CO
Raceway shape Determined by DEM model
Sensitivity
Analysis
Oxygen enrichment
With/without PCI
Top gas recycling operation
Validation (1) Measured gas composition
along tuyere axis in the hot
model experiments
(2) Comparison of calculated
raceway shapes with the
observations in the hot model
experiments
Remarks (1) 2-D raceway shape observed
through the hot experimental
model
(2) No recirculating flow found
in the raceway
44
2.2.2 Development of two-dimensional model
By the 1990s, few two-dimensional (2-D) models came into place to assist the search
for optimising blast conditions and design improvements in coal injection and its
operation.
2.2.2.1 Model of Aoki et al.
To analyse the flow patterns, heat transfer and reactions of injected coal in the blowpipe,
tuyere and raceway, a 2-D mathematical model of pulverised coal combustion was
developed by Aoki et al (1993). Calculations of coal combustion in the blowpipe-tuyere
region were carried out in a cylindrical space described by a 2D axi-symmetric system
(Figure 2.17). In the coke bed region, the space was considered as a 2D x-y plane set on
the tuyere axis. The calculated results in the outlet of blowpipe-tuyere region were used
as the inlet conditions of coke bed region. The two-equation turbulence model, k-ε
model, was employed to calculate the turbulent dispersion of coal flow, heat transfer
and reactions. In both domains, the gas phase was treated as continuous phase, and the
motion of coal particles was solved by Lagrangian approach (Kuo, 1986). Coke
particles in the raceway were assumed as a quasi-fluid, and the two-phase flow of coke
and gas were calculated in the model. The effect of coal volatile content on the
combustion was analysed. The effect of volatile matter content of coal on the gas
temperature distribution in the blowpipe is shown Figure 2.18. Clearly, an increase in
coal volatile content caused the rise in gas temperature within the region of
blowpipe-tuyere. The particle trajectories of low volatile coal (VM: 19.5%) are shown
in Figure 2.19. It was found particles under 30 μm could not disperse towards the radial
direction, while some particles over 65 μm dispersed in close to the inner wall of the
tuyere. This was caused by the small radial diffusion of fine particles with a diameter
less than 30 μm in the blowpipe where the gas velocity was extremely high (Ishii,
2000).
45
Figure 2.20 shows the coal particles reached the inner wall of tuyere when the particle
size was bigger than 65 μm for the high volatile coal (VM: 44.6%) injection. It was
explained that when high volatile coals were injected, more volatile was released,
followed by early ignition and combustion of volatile in the region of blowpipe-tuyere.
As a result, the dispersion of coal particles in the region could be significantly enhanced
by the intensified turbulence. Notably, from the calculated particle trajectories, it was
found that the mean residence time of pulverised coal in the region of blowpipe-tuyere
was 5 ms, and it was 7 ms in raceway. Totally, the travelling time for coal particles
from the injection point of lance to the raceway boundary was around 12 ms.
Computational results of the gas composition and temperature distributions in the
raceway along the centre line of tuyere are presented in Figure 2.21 for both with and
without PCI operation. Peak positions of CO2 and gas temperature were moved towards
the tuyere nose with PCI operation. After reaching the peak level in the raceway, the gas
temperature decreased, resulting from the endothermic reaction of CO2 with carbon
from unburnt char and coke.
Figure 2.17 Calculation domain in the region of blowpipe-tuyere.
46
Figure 2.19 Calculated particle trajectories of low volatile coal injection in the
blowpipe.
Figure 2.18 Effect of coal volatile content on gas distribution in the blowpipe.
47
Figure 2.21 Distributions of gas composition and gas temperature along the
centre line of tuyere.
Figure 2.20 Calculated particle trajectories of high volatile coal injection in the
blowpipe.
48
2.2.2.2 Model of Takeda and Lockwood
Takeda and Lockwood (1997) have developed an integrated axi-symmetric 2D
mathematical model of pulverised coal combustion in the regions of blowpipe and
raceway. In this model, a cylinder was used to represent the region of blowpipe and
tuyere, and the raceway was modelled into two regions: (1) a jet core region with an
empty space, and (2) a transition region with the linear decrease in voidage to the
raceway boundary. The assumed configuration of calculation domains is shown in
Figure 2.22. A modified turbulence model (k-lm model) was adopted for predicting the
gas and particle flows, and the Lagrangian approach for particle phase with stochastic
fluctuation was used to calculate turbulent dispersion of coal plume. The volatile
combustion in gas phase was simulated by the eddy dissipation model (Magnussen and
Hjertager, 1976).
Figure 2.22 Schematic representation of raceway structure used in the simulation.
49
The predicted gas temperature and oxygen concentration for the injection of a high
volatile coal (38.7%) at a rate of 50 kg/tHM are presented in Figure 2.23. The gas
temperature increased slightly in the blowpipe region where the injected coal particles
were in the early stages of heating and devolatilisation. On the other hand, rapid
changes in gas temperature and oxygen concentration were found in the raceway. This
indicated that the coal combustion was intensified by the volatile combustion.
Figure 2.23 Contours of gas temperature and oxygen concentration in the regions of
blowpipe and raceway.
50
In Takeda’s research (1994), four types of injection lances shown in Figure 2.24 were
also evaluated. Swirl motion was imposed to the carrier gas and coal flow in Case D.
An outer diameter of the injection lance was enlarged from 0.02 to 0.04m in order to
enhance gas turbulence in the vicinity of the injection lance in Case E. Radial velocities
were added to initial gas and particle velocity at the exit of the injection lance in Case F.
Figure 2.25 compares the performance of the lances in terms of coal burnout. As
indicated in the comparison, remarkable improvement in coal burnout could be achieved
by Case E because an increase in the outer diameter of the injection lance gave better
mixing of the coal plume and hot blast gas through the intense turbulence downstream
of the injection lance.
Figure 2.24 Lance design for coal injection.
51
2.2.2.3 Model of Haywood et al.
In an attempt to promote the utilisation of Australian coals in PCI operation, a 2-D CFD
(Computational Fluid Dynamics) based model has been developed by Haywood et al.
(1994). The coal burning characteristics in the experimental combustion rig (Figure 2.26)
was investigated. In the model, the geometry of the rig was considered to be an
axi-symmetric system. The model consisted of several sub-models for describing
turbulent mixing, particle tracking, pyrolysis of coal, volatile combustion and char
combustion.
Figure 2.27a and 2.27b shows the oxygen mass fraction (top) and gas temperature
contours (bottom) in the near injector region with a low volatile coal (VM: 9.67%) and
a high volatile coal (VM: 39.0%) injection respectively. A comparison of the gas
temperature contours indicated that early ignition could be provided by injecting high
volatile coal, resulting in rapid decrease of oxygen at the rig axis. This implied the
Figure 2.25 Coal burnout comparisons for the various modifications of lance.
52
oxygen consumption rate in the centre region was faster than that of oxygen diffusion
from bulk gas. Therefore it was concluded that mixing will be a key factor for high PCI
rates.
(a) low volatile coal injection
(b) high volatile coal injection
Figure 2.27 Oxygen mass fraction (top) and gas temperature contours (bottom) in
the near injector region. low volatile coal (a) low volatile coal; (b) high volatile
coal injection.
Figure 2.26 Schematic of the experimental combustion rig.
53
2.2.2.4 Summary of 2-D model
In the 2-D models, the gas phase is described by the transport equations of the
continuum phase, and the particle behaviours are calculated by Lagrangian approach, in
which the effects of the fluctuation of gas are considered on the particle dispersion. This
advantage allows one to precisely calculate the heating and devolatilisation of coal
particles, followed by volatile combustion. Table 2.1 shows a summary of key 2-D
models and the sub-models applied. The main findings obtained form the models are
indicated below.
(1) The dispersion of coal particle in the region of blowpipe-tuyere can be enhanced by
volatile release and combustion. Therefore, the extent of particle dispersion in radial
direction for high volatile coals is wider than that of low ones.
(2) The estimated resident time of coal particles in the regions of blowpipe, tuyere and
raceway is around 13 ms based on the 2-D calculation frame.
(3) As high volatile coals are injected, the gas temperature increases quickly in the
region of blowpipe-tuyere, while the oxygen concentration drastically decreases
within the coal plume.
(4) The burnout of injected coal in the region of blowpipe-tuyere-raceway is limited by
(a) short resident time of coal particles in the region, (b) mixing of coal particles
with hot blast gas, and (c) oxygen diffusion from bulk gas to the interior of coal
plume, and (d) coal properties, mainly volatile content and size.
Basically, the axi-symmetric 2D modes developed for the pulverised coal combustion in
the region of blowpipe-tuyere-raceway can generate results that are qualitatively useful
in practice (Chattopadhyay et al, 2010). However, with the axi-symmetric assumption,
some operation features can not be precisely analysed. For an example, all computations
are carried out with a lance located at the tuyere centre, resulting in underestimating the
influence of lance designs (such as inlet angle and configuration etc.) on the gas flow
pattern, trajectory of coal particle and coal combustion.
54
2.2.3 Development of 3-D model
To generate results which can be used for realistic operations, 3-D models by CFD
codes have been developed to simulate the PCI process under actual furnace conditions.
The models developed by different teams are described below.
2.2.3.1 Model of Picard
A 3D numerical model based on a CFD code was reported by Picard (2001).The
combustion domain of the model included tuyere, empty raceway (without coke) and a
porous media surrounding the raceway, as indicated in Figure 2.28. The raceway
dimensions were determined by the practical conditions of No5 blast furnace of
Dillingen works, and the porosity and deadman coke size were coming from probing
measurement performed in the Fos works. The standard k-ε model was applied for gas
turbulence, while the Lagrangian approach was used for simulating the dispersion of
coal particles in the gas. The release of volatile matters was modelled by two competing
devolatilisation model. In the model, the volatile matters were considered to be a
mixture of C3H8 and C6H6. In the model, the author considered radiative heat transfer
can be ignored in comparison with the convection heat transfer.
To understand the behaviours of coal particles in the region of tuyere-raceway, the
trajectories of coal particles were calculated in the model, as shown in Figure 2.29. It is
found that most of coal particles bumped into the raceway border and escaped there due
to high inertia. The particles average residence time in the raceway was 29 ms. Some of
particles which were smaller than 48 μm in diameter follow the recirculating gases in
the backward, resulting in longer residence time, as revealed in Figure 2.30. The effects
of injection rate and oxygen enrichment on coal burnout were examined in the model.
With 23% oxygen level in the hot blast gas, the coal burnout was decreased from 86.3
to 80.4% when the injection rate was increased from 170 to 230 kg/tHM. Besides, the
coal burnout could be promoted with higher oxygen level in the hot blast gas.
55
Figure 2.28 3D meshed calculation domain.
Figure 2.29 Trajectories of coal particles within the tuyere and raceway.
Figure 2.30 Evolution of coal residence time within the raceway versus particle
diameter.
56
2.2.3.2 Model of Guo et al.
In the research of Guo et al., (2005), the pulverised coal combustion behaviours in the
combustion rig (Figure 2.31) were simulated using CFD code. The empty combustion
chamber was adopted to simulate the raceway of blast furnace. In the model,
two-competing reaction model was chosen to simulate the de-volatilisation of coal, and
eddy breakup model was employed for the volatile combustion in gas phase. The model
assumed that the time scale of reaction was much shorter than that of turbulence and the
gas–gas reaction was controlled by turbulent diffusion.
The calculation results showed recirculation gas flow in the combustion chamber as can
be seen in Figure 2.32. The recirculation was resulted from the expansion design
between the tuyere and the combustion chamber. Figure 2.33 indicates typical coal
particle trajectories coloured by their sizes. The extent of radial dispersion of particles
exiting from a simple straight lance was very limited in the tuyere. On the other hand, a
particle segregation phenomenon could be found in the combustion chamber. Upon
exiting the injection lance, large particles with larger momentum maintained their initial
direction (the lance axial direction), while fine particle dispersed more widely. This is
quite different from the calculated particle trajectories given by Aoki et al. (1993) as
indicated in Figures 2.19 and 2.20. Figure 2.34 shows oxygen and volatile matter
distributions in the gas phase. The oxygen concentration in the coal plume region was
always low as the combustion of the volatile and char oxidation consume oxygen. Thus
the transport of oxygen primarily controlled the combustion rate.
57
Figure 2.31 Main dimensions (in mm) of the coal combustion model (plan
view).
Figure 2.32 Gas velocity vectors in Y-Z plane (a) vector length to scale, and (b)
vector normalised showing recirculation zone.
58
Figure 2.33 Typical particle trajectories with colour scaled to particle size.
Figure 2.34 Gas species fraction isopleths in Y–Z plane. (a) Oxygen; (b)
volatile matter.
59
Figure 2.35 shows the particle characteristics (devolatilisation and burnout) as a
function of axial distance from the lance tip along the centreline. At 0.2m, particles
below 30μm have nearly completed devolatilisation, whereas those above 80μm have
hardly started (Figure 2.35a). Up to 0.5m from the lance tip, the burnout levels were
significantly different for particles of different sizes (Figure 2.35b); however, beyond
0.5 m, there was little variation—at this distance, slow char oxidation was prevalent. In
other words, the coal burnout is mainly contributed by the release of volatile matter
from coal particles, and the char combustion is limited due to insufficient oxygen
around the particles.
(a) (b)
Figure 2.35 Calculated mass fraction of volatiles (a) and calculated mean coal
burnout (b) as a function of distance from the lance tip along the centreline
averaged for different particle sizes (in μm).
60
2.2.3.3 Models of Shen et al.
A series of PCI calculation models have been established by Shen et al. (2008, 2009a,
2009b) to analyse the combustion behaviours of injected coal in the empty tuyere
(Figure 2.31). Based on the measured Q factors of coals contenting 12-39 % volatile
matter (Ueno et al., 1993; Niksa et al., 1984; Maloney and Jenkins, 1984), a correlation
of stoichiometric parameters for two competing devolatilisation model was made (Shen
et al., 2008). In addition, char gasification reactions (C + CO2 = 2CO and C + H2O =
CO + H2) were considered in the model. Figure 2.36 compares the calculated burnouts
(with/ without considering char gasification) with those generated by Guo et al. (2005).
Case 1 and Case 2 were compared for downstream and upstream, respectively. In the
upstream, late ignition was found with the model. As for the downstream, the two
models predicted a similar burnout evolution beyond 0.6 m for the base condition. It is
found from Cases 2 and 3 that burnout curves were similar upstream, quite different
downstream. With the char gasification reactions, the downstream burnout level was
increased from 65 to 75%.
To evaluate the combustion performance of coal blend in the combustion rig, three
cases were simulated and compared by Shen et al. (2009a):
1. Case I: the blend of PC_A and PC_B, with the blend fraction 50% + 50%;
2. Case II: single coal case of PC_A (a larger but higher volatile(HV) coal); and
3. Case III: single coal case of PC_B (a fine but lower volatile (LV) coal)
Figure 2.37 compares the burnout evolutions of Cases I, II and III: (a) along the
centreline; and (b) at the exit. Along the centreline (Figure 2.37a), the final burnout of
Case I was closer to Case II, i.e., higher than the average value of the final burnouts of
two coals. Conversely, when the final burnout was calculated at the exit area (Figure
2.37b), the burnout of Case I was even slightly higher than both Cases II and III. That is,
the overall burnout of coal blend showed a non-additivity, i.e., synergistic effect. It was
concluded that the chemical interactions between two components in terms of particle
61
temperature and volatile content were responsible for the synergistic effect: the HV coal
releases more VM, helping form a higher gas temperature field, which then heated up
the LV coal and promoted its devolatilisation and combustion.
The effects of blast conditions on pulverised coal combustion were examined by Shen et
al. (2009b). It is found that the combustion efficiency of the injected coal could be
improved by increasing blast air temperature. However, as shown in Figure 2.38, the
burnout of the coal could not be increased further when the blast air temperature was
higher than 1200oC. Figure 2.39 shows the influence of oxygen enrichment on final
burnout. Clearly, the final burnout could be improved as more oxygen was added into
the blast.
The performance of coaxial lance with cooling gas of methane, oxygen and air flowing
through the annulus of the lance was compared by Shen et al. (2009c). Figure 2.40
shows the effect of cooling gas type on coal burnout. Note that the oxygen enrichment
in blast was kept with the same for all the three cases. It is shown that for the three types
of cooling gas, oxygen gave the highest burnout of 67%. In addition, a linear
relationship was found between the burnout and atomic O/C ratio in the gases delivered
to the tuyere. This was because, comparing the O2 distributions on a cross-section at the
distance of 550 mm from the lance tip (Figure 2.40b), by using methane as the cooling
gas, a larger amount of O2 was consumed when the methane was burning together with
the VM competitively. As a result, the VM combustion was slowed down, which
decreased the final burnout significantly. In addition, the amount of O2 available to the
subsequent char reactions was also reduced.
62
Figure 2.36 Comparison of burnout evolutions predicted by the previous
model (Case 1), and present model with char gasification reactions (Case 2)
and without char gasification reactions (Case 3).
Figure 2.37 Burnout for Cases I, II and III: (a) along the centreline and (b) at the
exit.
63
Figure 2.38 Effect of blast temperature on burnouts at the distances of 300 mm and
925 mm from the lance tip, respectively.
Figure 2.39 Effect of oxygen enrichment on coal burnout.
64
Figure 2.40 Effect of cooling gas type: (a) final burnout and (b) O2
distributions at the cross-plane of 550 mm from the lance tip.
65
For further study of the PCI operation in the regions of lance blowpipe tuyere raceway
and coke bed, a 3D mathematical model of the combustion of pulverised coal and coke
was developed by Shen et al. (2011) to simulate in-furnace phenomena of pulverised
coal injection in an ironmaking blast furnace. The model integrated not only pulverised
coal combustion model in the blowpipe-tuyere-raceway-coke bed (deadman zone) but
also coke combustion model in the coke bed. Figure 2.41 reveals the calculation domain
in the model.
Figure 2.42a shows the gas velocity vectors in the raceway cavity. It was found a
high-speed jet (220m/s) formed along the tuyere axis, which was similar to the cold
model observations by Inatani et al. (1976). After reaching the raceway boundary, the
gas flow started a large-scale recirculation above the main gas flow jet in the raceway. It
was probably because the porosity was set at a level of 0.25 for the deadman. In the
coke bed, as shown in Figure 2.42b, gas velocities decreased rapidly to <5 m/s within a
very short distance. It is shown in Figure 2.39c that, corresponding to the gas flow, the
coal particle trajectories inside the raceway had two different flow patterns: (i) an
inclined main coal plume located along the lower part of the raceway, where fine
particles were observed at the upper part of the plume initially and then left the main
coal plume before reaching the end of the raceway; and (ii) a large-scale recirculation of
the fine particles of up to 70 μm around the raceway centre. The coke bed also showed
two flow patterns of coal particles accordingly (Figure 2.42c): (i) the main coal plume
(relatively large particles of around 100 μm) penetrated into the deadman zone; (ii) the
recirculating fine particles exited mainly from the top of the raceway and then moved
upward into the dripping zone. Figure 2.42d shows the residence time of coal particles
along the main coal plume was around 10–50 ms before reaching the end of the raceway,
while the recirculating coal particles might be up to 0.9 s in the raceway. On the other
hand, compared with the raceway in the coke bed, the travelling time of the particles
penetrating the coke bed was quite long, around 1.0 s. The travelling time was even
longer in the deadman compared to the dripping zone. Moreover, Figure 2.43 shows the
66
coal burnout distribution along the particle trajectories in the coke bed. It is shown in
the simulation that the burnouts varied greatly in different zones of the coke bed, nearly
100% in the dripping zone, and around 75% in the deadman zone, in which the char
burnout could not be promoted further due to lack of O2 and CO2. Notably, the char in
the deadman could be consumed by the direct reduction of FeO in slag (Iwanaga, 1991).
Figure 2.41 Geometry of the model: (a), the whole model; (b), porosity distribution
(Zone 0: 1, Zone 1: 0.25, Zone 2: 0.5, Zone 3: 0.4); (c), blowpipe and raceway; and
(d), lance tip. The detailed dimensions are, (1) for blowpipe, radius: 90 mm, and
length: 800 mm; (2) for tuyere, radius: 75/90 mm, and length: 135 mm; (3) for
raceway, depth: 1600 mm, height: 1000 mm (925 + 75), and width: 710 mm; and
(4) for coke bed, depth: 3700 mm, height: 4500 mm, and width: 1000 mm.
67
Figure 2.42 Flow pattern of gas-particle flow: (a), vectors of gas phase in the
raceway; (b), streamlines of gas flow; (c), particle trajectories coloured by
particle mean size; and (d), particle trajectories coloured by particle travelling
time.
Figure 2.43 Combustion characteristics of coal along particle trajectories in the coke
bed.
68
2.2.3.4 Model of Gu et al.
Gu et al. (2008) has reported a three-dimensional multi-phase model using Eulerian
approach has been developed to simulate the coal heat transfer, devolatilisation and
combustion process inside a blast furnace tuyere. The calculation was based on an
Eulerian coordinate, while the particle phase was treated as a pseudo-fluid interacting
with the gas phase. The k–ε–kp two-phase turbulence model was used to model the gas
and particle turbulence, and the eddy dispersion model was adopted to quantify the
effect of turbulence on the combustion rates of volatiles, carbon monoxide and
hydrogen. The heterogeneous reactions considered in the model included char oxidation
with oxygen, carbon dioxide and moisture. The computational domain is shown in
Figure 2.44. Obviously, the raceway shape looks like a balloon.
Figure 2.45 shows the distributions of gas velocity vectors and gas temperature in the
computational domain. As indicated in Figure 2.45a, a high speed gas exiting the tuyere
was found to form a jet space inside the raceway, while a large scale recirculation zone
was generated between the jet and coke bed. Owing to well mix of oxygen (31.5% in
the hot blast gas) and fuel, the location of highest temperature (3115K) coincided with
the location of gas recirculation, as shown in Figure 2.45b. The effect of tuyere diameter
to the coal combustion was also examined. Figure 2.46 shows the coal burnout at the
exit of the computational domain. It can be seen that coal burnout was promoted from
60.6% to 85.3% when the tuyere diameter was enlarged from 0.156 m to 0.165m. It
could be explained by an increase of residence time of coal particles in the raceway
when the tuyere diameter was enlarged. It should be noted that the blast furnaces of
CSC have reduced their tuyere diameters for obtaining deeper penetration of hot blast
gas into the deadman, since it is thought of as an effective countermeasure to keep a
stable thermal level within the hearth.
69
Figure 2.45 Distributions of (a) gas velocity vectors and (b) gas temperature (K)
in the computational domain.
(a) (b)
Figure 2.44 Schematic of computational domain: (a) side view; (b) top view.
(a)
(b)
70
2.2.3.5 Model of Nogami et al.
Nogami et al. (2004) has reported a 3D transient analysis model for raceway and the
surrounding coke bed. In the model Finite Differencing Method (FDM) and Discrete
Element Method (DEM) were used for featuring the coke movement. As a result, the
trajectories of coke particles could be traced within and around the raceway.
Gasification rate of coke particles was calculated by shrinking core model, in which
chemical reaction on the surface of particle, mass transfer in boundary layer, and gas
diffusion inside particle were considered. The model was applied to the combustion
tests in an experimental coke bed as shown in Figure 2.47. The calculated raceway
shapes were validated by that obtained from the observations of the tests, as revealed in
Figure 2.48. It can be found that the raceway for PCI operation was slightly lager than
that for all coke operation. With the design of the combustion rig, the development of
raceway was constrained in both side walls of the rig, therefore the raceway could be
considered as 2-dimensional. Figure 2.49 shows (a) the calculated raceway shape, and
(b) the calculated gas velocity vectors in the lower zone of blast furnace. As indicated in
Figure 2.49b, the blast proceeded as a high speed jet along tuyere axis due to its inertia,
Figure 2.46 Coal burnout at the exit of the computational zone.
71
and there was no recirculation flow in the roof of the raceway. It is different from the
gas flow patterns in the raceway given by Shen et al. (2011) and Gu et al. (2008).
Figure 2.47 Schematic figure of hot model.
Figure 2.48 Comparison of calculated raceway shape with observation of
test. (a) All coke operation. (b) PCI operation.
72
2.2.3.6 Summary of 3-D model
As reported above, 3-D studies in PCI operation have been developed from blowpipe to
raceway, even to the lower zone of blast furnace. A summary of key 3-D models and the
sub-models applied can be found in Table 2.1. Discussion for the 3-D models is made
below:
(1) The segregation of coal particles can be found after exiting injection lance. Large
particles with larger momentum maintain their initial direction (the lance axial
direction), while fine particle disperse more widely.
(a)
(b)
Figure 2.49 Characteristics of raceway: (a) calculated raceway shape, and (b)
calculated gas velocity vectors.
73
(2) Owing to lack of O2 and CO2, the char burnout can not be promoted further when
the char particles enter the deadman. It should be noted that the char in the deadman
zone can be consumed by the direct reduction of FeO in slag.
(3) After leaving tuyere nose, blast proceeds as a high speed jet along tuyere axis due to
its inertia. In the region above the jet, the recirculation of gas may occur as a result
of low porosity around the raceway.
(4) The coal burnout in the raceway can be significantly promoted when the tuyere
diameter is enlarged. However, to keep a stable thermal level in the deadman, the
tuyere diameter has been reduced at CSC for obtaining deeper penetration of hot
blast gas into the deadman.
As revealed above, the 3-D models have been successfully developed for featuring the
coal combustion characteristics in the region of blowpipe-tuyere-raceway. The
parameters which are related to coal burnout are especially examined in the models. In
fact, higher pressure resistance (or poor permeability) has been encountered in the lower
zone of CSC’s blast furnaces when attempting to increase the PCI rates. For a stable
operation of blast furnace with high PCI rates, countermeasures for the reduction of
pressure loss caused by coal combustion in the raceway is needed at CSC.
2.3 Sub-models for integrated calculation
The combustion of pulverised coal in the region of blowpipe-tuyere-raceway includes
many physical and chemical processes as shown in Figure 2.50. After exiting the
injection lance tip, the coal particles are dispersed to the hot blast gas, while the coal
particles are heated by hot blast gas, and the moisture evaporates. Coal devolatilisation
then occurs after further heating. The released products, mainly hydrocarbon, are ignited,
which causes an increase in the temperatures. As a result, the devolatilisation of coal
can be enhanced. Finally, the residual char particles can be consumed by oxidants (O2,
CO2 and H2O). All of these processes take place sequentially with some overlap. Since
74
the residence time of coal particles in the combustion region is very short, generally less
than 20 ms, the characteristics of these physical and chemical processes are very
important for the effectiveness of a PCI system.
As mentioned above, the combustion of pulverised coal consists of several processes,
which can be described by some mathematical and empirical equations or simple model
component. These equations should be solved simultaneously to express entire coal
burning characteristics in the region of blowpipe-tuyere-raceway. In the CFD based
modelling for coal burning in the raceway, all the steps of geometry and grid
generations, boundary condition implementation and sub-model simulations are
integrated into a framework, as shown in Figure 2.51. Note that the use of appropriate
sub-models allows for a more accurate description of the coal burning in the combustion
region.
Figure 2.50 Illustration of combustion phenomena of pulverised coal (Ishii, 2000).
75
According to the experiments by Seeker et al. (1981), the combustion of a high volatile
coal in a high temperature environment (1750K) can be divided into three stages in
terms of residence time (Smoot and Smith, 1985): heat-up and devolatilisation (few ms
to 75 ms), followed by char oxidation, as shown in Figure 2.52. It implies that the coal
burnout in the raceway may be primarily contributed by the coal devolatilisation, rather
than by char oxidation and gasification. Therefore the sub-model review starts with the
coal devolatilisation, followed by other physical and chemical processes involved.
Solution
Gas phase
˙ Mass conservation
˙ Momentum conservation
˙ Turbulence model
˙ Energy conservation
˙ Turbulence combustion
model
Solid Phase
˙ Devolatilisation model
˙ Char combustion
˙ Reaction/ diffusion
controlling regime
˙ Momentum
conservation (particle
trajectory)
˙ Energy conservation
Initialisation
˙ Geometry (meshing)
˙ Boundary conditions
˙ Sub-models
˙ Energy conservation
˙ Turbulence combustion
model
Radiative Transport
Particle
properties
Radiative
energy
exchange
Figure 2.51 Framework of the CFD code and computational procedure of the
gas phase and solid (coal particle) phase (Du et al., 2007).
76
2.3.1 Devolatilisation of coal
Coal devolatilisation is a complex chemical and physical process. Various phenomena
are involved during coal devolatilisation, namely, heat transfer to and within the coal
particles, various bond-breaking and cross-linking reactions, and the transport of
volatile products. It should be noted that enhanced yields for rapid heating have been
correlated with “Q-factors”, which are the ratios of the weight loss after rapid heating to
the proximate volatile matter content of the coal, and values for various coals have been
reported (Anthony and Howard, 1976; Niksa and Lau, 1993; Yan et al., 2014).
As a matter of fact, a complete description of the chemical reactions that occur during
devolatilisation is not available. Practically, based on experimental results without
considering the chemical structure of coal and complicated physics process, simplified
models, such as the single overall reaction model (Badzioch and Hawksley, 1970) and
two-competing reaction model (Kobayashi et al., 1977), have been widely applied in the
40 μm
80 μm
Figure 2.52 Effects of coal size and residence time on physical and temperature
profile (Smoot and Smith, 1985)
77
CFD based simulations for pulverised coal combustion.
2.3.1.1 Single overall reaction model
The single overall reaction model proposed by Badzioch and Hawksley (1970) is the
simplest devolatilisation model, in which the kinetics of devolatilisation has been
simplified by assuming a first-order decomposition occurring uniformly throughout the
particle.
CharVMCoal k (2.1)
The coal conversion rate can be expressed as the lumped production rate of all volatile
species being proportional to the volatile matter yet to be released:
)( VVkdt
dV (2.2)
Where V∞ is the ultimate yield of volatiles at t = , i.e. the total volatile content of coal,
and k is the rate constant, usually expressed as an Arrhenius relationship.
pRTEAk /exp (2.3)
Where Tp is particle temperature, and A (pre-exponential factor) and E (activation
energy) are constants, determined experimentally for the coal. As coal devolatilisation is
not a single reaction but a wide range of overlapping decompositions, the use of a single
set of parameters to describe reactions occurring over a wide range of conditions may be
inadequate.
2.3.1.2 Two competing reaction model
Based on the assumption that the coal may decomposed via one of several possible
reaction paths depending upon the time-temperature history, two competing reaction
model has been developed by Kobayashi et al. (1977) to explain devolatilisation yields
78
as a function of time, temperature and heating rate. The two parallel and competing
reactions are given as follows:
1111 )1( 1 VYSYCoal k (low temperature) (2.4)
2222 )1( 2 VYSYCoal k (high temperature) (2.5)
where Y, S, and V denote stoichiometric coefficient, char, and volatile, respectively. The
relative importance of the two equations is mainly determined by temperature.
Specifically, when the temperature is low, the devolatilisation reaction is dominated by
route one (Equation 2.4). Alternatively, it is governed by route two (Equation 2.5) once
the temperature is relatively high. Accordingly, the devolatilisation reaction kinetics are
written as:
CYkYkdt
dV 2211 (2.6)
pRTEAk /exp 111 (2.7)
pRTEAk /exp 222 (2.8)
To get calculation results that can sufficiently reflect the pulverised coal burning
characteristics, in a research work at CSC (Du, 2001), the kinetic parameters for two
competing reaction model and those for single overall reaction model were validated
using the measured gas temperatures from an experimental combustion rig, as show in
Figure 2.53 (Burgess et al., 1983). The investigated operation conditions are
summarised in Table 2.2. Details of the kinetic parameters are given in Table 2.3. In the
calculation, the volatile combustion was dealt by the probability density function (PDF)
approach. Figure 2.54 shows the comparison of predicted temperature distributions with
the gas temperatures measured along the blowpipe. It depicts that the temperature was
spatially uniform when using the parameters of Kobayashi et al. (1977), indicating that
the rate constants gave unrealistic devolatilisation rates for blast furnace operations.
Upon inspection of the prediction given by the single reaction model with the
79
parameters of Takeda (1994), the gas temperature increased at about 0.25 m from the
lance tip, resulting from the release and ignition of volatile matters. As can be seen in
Figure 2.54, the calculated temperature kept rising with moderate rates in the following
process. Obviously, the rates in region of 0.4-0.5 m from the tip were behind the
measurements. Instead, when the parameters proposed by Ubhayakar et al. (1976) were
employed, the calculated temperature profile exhibited a good agreement with the
measurements. As revealed, the calculated temperature distribution could be partitioned
into two stages: it composed of rapid rise in the upstream region and progressive
increase in the downstream one. The behaviour in the first region arose from the PC
devolatilisation or pyrolysis reaction followed by the combustion of the emitted volatile
matters with oxygen. Obviously, in this stage, the devolatilisation reaction was
accelerated by the route two (Equation 2.5) when the coal particles were heated up by
volatile combustion. In regard to the second region, the slow increase in temperature
was attributed to the reaction between char and oxygen. In a word, after the PC is blown
into the blowpipe, the chemical reaction is primarily achieved by the gas-phase
combustion (i.e., homogeneous reaction) and then implemented by the solid-phase
oxidation (i.e., heterogeneous reaction).
Apart from the importance of kinetic parameters, the stoichiometric coefficients Y1 and
Y2 play essential roles for the determination of the total amount of volatile being
released during the combustion. Since the route one (Equation 2.4) represents
low-temperature devolatilisation, Y1 is set to the fraction of volatiles given by proximate
analysis. The high temperature yield Y2 (in Equation 2.5) is often estimated as being
related to Y1. A summary of the relation between Y2 and Y1 used in the PCI combustion
models (Suzuki et al, 1986; Guo et al., 2003; Aoki et al., 1993; Du and Chen, 2006;
Guo et al., 2005; Haywood et al., 1994) has been reported by Shen et al. (2008), as
shown in Figure 2.55. Clearly, the ratio of 1.5 (red line) employed in this research is
acceptable for the modelling in comparison with the experimental results (Ueno et al.,
1993; Niksa et al., 1984; Maloney and Jenkins, 1984). Table 2.4 shows the kinetic
80
parameters and the stoichiometric coefficients employed in the PCI combustion models.
Table 2.2 Operational conditions selected by Burgess et al. (1983).
Rig diameter, mm 50
Hot blast temperature, K 1243
Hot blast velocity, m/s 68
Coal particle diameter, μm 40
Volatile matter of PC (db), % 35.9
PC injection rate, kg/h 5.3
Table 2.3 Three sets of parameters used for predicting PC devolatilisation.
Kobayashi et al.
(1977)*
Ubhayakar et al.
(1976)*
Takeda (1994)**
A1, 1/s 200000 3.7×105 8.36×10
4
A2, 1/s 1.3×107 1.46×10
13 N/A
1E , kJ/mol 1.046×102 74 65
2E , kJ/mol 1.674×102 251 N/A
* Two competing reaction model
** Single overall reaction model
81
800
1000
1200
1400
1600
1800
2000
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Distance from lance tip, m
Ga
s te
mp
era
ture, K
Injection rate: 5.3 kg/h
Blast temperature: 1243K
O Burgess et al. (1983)
Ubhayakar et al. (1977)
Takeda (1994)
Kobayachi (1976)
Figure 2.53 Schematic of the Blowpipe/Tuyere (combustion test section)
assembly of the pilot scale raceway hot model.
Figure 2.54 Gas temperature distributions for pulverised coal burning in a reactor
from experimental measurement and numerical predictions using different
devolatilisation models (Du, 2001).
First stage Second stage
82
Y2/Y
1
Figure 2.55 Comparison of the relationships between Y1 and Y2 in the literature (Shen
et al., 2008).
83
Table 2.4 Kinetics of single overall reaction and two competing reaction models used in
the PCI calculation models.
Authors
Kinetic parameters Stoichiometric
coefficients
A (1/s) E (kJ/mole)
Y1
Y2 A1 A2 E1 E2
Takeda (1994) 8.36×104 65 1.5 VM
He et al. (1986) 5105.1 73.13 1.3VM
Sato et al. (1996)
-1.11×103Cb
2+1.96×
105Cb-8.5510
6
Cb: carbon in coal,
wt%
-2.26×102 Cb
2+
4.48×104Cb-
2.04×106
(0.96×10-3
Tb+4.6
×10-2
Cb-3.9)VM
Tb: B. T., oC
Ubhayakar et al. (1976) 5107.3 131046.1 74 251 VM 2Y1
Kobayashi et al. (1977) 5102 71037.1 104.6 167.4 0.3 1
Burgess et al. (1983) 5107.3 131046.1 74 251 VM 2 Y1
Jamaluddin et al.
(1986) 5107.3 131046.1 74 251 VM 2 Y1
Aoki et al. (1993) 5107.3 131046.1 149.6 251 VM 2 Y1
Picard (2001) 1.34×105 131046.1 74 251 VM 1.7Y1
Guo, et al. (2005) 5107.3 131046.1 150 251 VM 1.4+
1.6VM
Du and Chen (2006) 5107.3 131046.1 74 251 VM 1.5Y1
Gu et al. (2008) 5107.3 131046.1 74 251 VM 2 Y1
Shen et al. (2008) 5107.3 131046.1 150 251 VM
1.25Y12
+
0.92Y1
84
2.3.2 Char Oxidation
2.3.2.1 Field approach (Field, 1969)
Once the volatile component of coal is completely evolved, heterogeneous surface
reactions begin to consume the char fraction of the coal. The reactions occurring in the
raceway can be listed below:
C + 1/2 O2 CO (2.8)
C + CO2 2CO (2.9)
C + H2O CO + H2 (2.10)
The char oxidation (Equation 2.8) is exothermic, while char gasification with CO2 and
H2O are endothermic. The time required for consumption of char particles in a turbulent
environment can range from 30 ms to over hour. It should be noted that the char
reactions with CO2 and H2O are about 5-6 orders of magnitude slower than that with O2
(Smoot and Smith, 1985). Due to very short residence time for injected coal particles
within the raceway, the char burnout contributed by the gasification reactions may be
negligible.
The particle temperature affects the apparent reaction rate by shifting the reaction
controlling mechanisms. Figure 2.56 represents the effect of temperature on the reaction
rate of a char particle. In low temperature range (regime I), the chemical reaction may
determine the overall rate. The reaction rate in middle temperature range is controlled
by both chemical reaction and diffusion (regime II). In regime III (high temperature),
diffusion of reactants and products in the boundary layer limits the reaction rate
(diffusion control), while the temperature has little effect. In the actual PCI combustion
operation, reactions of char particles occur under various temperatures and gas
atmospheres. Therefore, the reaction and diffusion should be considered for the
estimation of the global reaction of char particles (Ishii, 2000; Picard, 2001).
85
The corresponding equations for global reaction of char particles proposed by Field
(1969) are given below:
RD
RD
kk
kk
oxp
ppA
dt
dm 2.11
where 2
pp πDA is the surface area of the coal, and oxp is the partial pressure of
oxidant species in the gas surrounding the combusting particle. kD and kR are the
diffusion rate and the kinetic rate, respectively, expressed by
0.75
p
1
2D
C
gp
D
TTk 2.12
pR
N
gR TETk R expAR 2.13
In this research, the mass diffusion–limited rate constant is 12
1 105C , the kinetics
limited rate pre–exponential factor (AR) and activation energy (ER) are 3050
kg/(m2-s-atm) and 179.4 kJ/mol (Smith, 1982), respectively. Besides, the temperature
exponent N is taken as zero (Smoot and Pratt, 1979). Table 2.5 summarises some
kinetics of char oxidation by PCI combustion models.
Figure 2.56 Rate-controlling regimes for char reactions (Smoot and Smith, 1985).
86
Table 2.5 Kinetics of char oxidation employed in the PCI combustion models.
Authors Chemical reaction rate, kR
Jamaluddin et al, 1987 )R
exp( R
p
RRT
EAk
AR=309 kg/m2-s-atm 1200<Tp<1800 K
ER=156.5 kJ/ mol
AR=3.15 kg/m2-s-atm 1800<Tp<2100 K
ER=87 kJ/ mol
Nogami, et al. (1992) )R
exp( R
p
RRT
EAk
AR=87100 kg/m2-s-atm Tp≦1273 K
ER=149.7 kJ/ mol
AR=-4.9+3.85×10-3
Tp Tp>1273 K
ER=0
Takeda and Lockwood
et al., 1997
)R
exp( R
p
RRT
EAk
AR=87140 kg/m2-s-atm
ER=102 kJ/ mol
Haywood et al. (1994)
Du and Chen (2006)
)R
exp( R
p
RRT
EAk
AR=3050 kg/m2-s-atm
ER=179.4 kJ/ mol
Picard (2001) )R
exp( R
p
RRT
EAk
AR=0.007 kg/m2-s-atm
ER=90 kJ/ mol
Gu et al. (2008) )R
exp( R
p
RRT
EAk
AR=1.813×103 m/s
ER=108.9 kJ/ mol
Aoki et al (1993)
He et al. (1986)
Sato et al., 1996
Nogami, et al. (2004)
g
p
R RTT
k )80001
exp(7260
87
2.3.2.2 Gibb Model (1985)
In Gibb approach (Gibb, 1985; Guo et al., 2005), the oxidation mechanism of carbon
can be characterised by the parameter Φ so that oxides are produced according to the
equation:
ΦC + O2 → 2(Φ-1) CO + (2-Φ) CO2 (2.14)
The value of Φ is assumed to depend on the particle temperature TP:
)exp(2
)1(2
p
SS
T
TA
(2.15)
where the constants are given by Gibb as AS=2500 and TS=6240 K (Gibb, 1985). An
analytical solution of the oxygen diffusion equation leads to the following equation for
the rate of change in char mass (mc):
11
32
1
1 ))((1
3
kkkM
M
edt
dm
COxy
CSC
(2.16)
The far field oxygen concentration ρ∞ is taken to be the time-averaged value obtained
from the gas phase calculation, and ρc is the density of the char. Physically, k1 is the rate
of external diffusion, k2 is the surface reaction rate, and k3 represents the rate of internal
diffusion and surface reaction. These are defined as follows:
21
PR
Dk (2.17)
P
R
R
kek )1(2 (2.18)
)exp(P
CpRR
T
TTAk (2.19)
aTkk PR
2
3 /)1coth( (2.18)
88
5.0)(eaD
kR
P
C (2.19)
The parameters include void fraction (e), volume/internal surface area ratio, a, and
particle radius (Rp). D is the external diffusion coefficient of oxygen in the surrounding
gas, and kc is the carbon oxidation rate, defined by the modified Arrhenius equation.
Besides, Gibb recommends a value for Dp an order of magnitude less than D. As
indicated in Table 2.1, Gibb approach has been adopted for char oxidation by Guo et al.
(2005) and Shen et al. (2008, 2009a, 2009b, 2011).
2.3.3 Turbulence model
In the region of blowpipe-tuyere-raceway, the fluid motion is practically fast. The
standard k-ε model has been used in the PCI combustion models (Aoki et al., 1993;
Haywood et al., 1994; Picard, 2001; Nogami et al., 2004; Shen et al., 2008). Some
modified k-ε models were also considered in the calculations (Takeda and Lockwood,
1997; Guo et al., 2005; Gu et al., 2008). For a better treatment in the mixing and
dispersion of coal particles in comparison with the conventional k-ε model, the RNG k-ε
model (Biswas and Eswaran, 2002) is thus applied in this research to simulate the
turbulent combustion (Du and Chen, 2006; Du et al., 2007; Du et al., 2015). The
complete formulation of the RNG k–ε turbulence model is given as follows:
ρεGkμρUk kefft (2.20)
ρεCGCk
εεμρUε k1εefft
*2 (2.21)
where k is the turbulence kinetic energy, ε is the kinetic energy dissipation rate, and Gk
is the generation of turbulence kinetic energy due to the mean velocity gradients and
expressed by:
89
UUkUUG t
T
tk
I
3
2)( (2.22)
UUkUUG tT
tk
I
3
2)( (2.23)
The coefficient *2εC is given by
3
3
22012.01
)38.4/1(
CCC*
ε (2.24)
where η=Sk/ε, and S is the modulus of the mean rate of strain. The coefficients Cμ, C1ε,
C2ε, and αt are empirical constants, and their values derived empirically are 0.0845,
1.42, 1.68, and 1.393, respectively.
2.3.4 Gas combustion in turbulent flow field
2.3.4.1 Probability density function (PDF) of turbulence chemistry
In this approach, individual species transport equations are not solved. Instead, the
solution of a single conserved scalar transport equation, the mixture fraction, is solved.
The individual component concentrations are derived from the predicted mixture
fraction distribution (Eghlimi and Sahajwalla, 1997; Zhou, 1993). The effect of
turbulence on the gas combustion is modelled using the partial equilibrium chemistry
model. An assumption is made that the reaction is mixing limited (i.e. fast chemistry)
and the diffusion coefficients of all the species are equal. (Note that this assumption is
reasonable since the temperature in the combustion zone of blast furnaces is extremely
high and the effects of turbulent convection dominate those of molecular diffusion.)
Given this simplifying assumption, the species equations can be reduced to a conserved
scalar quantity known as the mixture fraction f . The mixture fraction can be written in
terms of the atomic mass fraction as
Oi,Fi,
Oi,i
XXXX
f
, where iX is the
elemental mass fraction for element i . The subscript O and F denote the value at the
oxidizer stream inlet and the fuel stream inlet, respectively. It is essentially a numerical
90
construct used to describe the degree of scalar mixing between the fuel and the oxidant.
In accordance with the mixture fraction concept, the transport equations are given by:
f
σ
μfρ
t
g t,
ggU (2.25)
Zk
ερCfμCZ
σ
μZρ
g
g
g2Z
2
g t,1Z
t
g t,
gg
U (2.26)
The last two terms of Equation 2.26 represent, respectively, the production of
concentration fluctuation due to non–uniformity of mixture fraction and destruction rate
of the fluctuations. Note that in the present simulations, the values of tσ , 1ZC and
2ZC are specified as 0.85, 2.86 and 2.0, respectively. Importantly, the shape of the
assumed probability distribution for a variation of the mixture fraction, fp , is
described by the β–function form that more closely represents experimentally observed
features. The shape produced by this function is given by the following functions of
mean mixture fraction f and concentration fluctuation Z :
dff1f
f1ffp
1B1-A
1B1-A
(2.27)
1
Z
f1f fA (2.28)
f
A f1B (2.29)
Thus, given prediction of f and Z at each point in the flow field (Equations 2.25 and
2.26), the assumed PDF shape can be computed and used as the weighting function to
determine averaged values of variables.
2.3.4.2 Eddy break up and eddy dissipation models
Gaseous combustion in turbulent flow field can be modelled by eddy dissipation
approach (Magnussen and Hjertager, 1976), in which modelling concept assumes that
91
chemical reactions are instantaneous when molecular–level mixing of reactants occurs.
In such cases, the overall rate of reactions is essentially controlled by turbulent mixing
rather than chemical kinetics. Note that this assumption is reasonably applicable in the
combustion zone of blast furnaces where temperature is extremely high and reaction
rates are fast compared to reactant mixing rates. In turbulent flows, the mixing time is
dominated by the eddy properties, and therefore the reaction rate is proportional to the
rate of turbulent mixing (g
g
kε
), i.e.
S
m,mmin
ερC ox
fu
g
gA
g
fuk
R (2.31)
where Rfu is reaction rate of volatile, AC is an empirical constant, mox and mfu are the
time-average mass fraction of oxygen and volatile in the gas phase and S is the
stoichiometric oxygen required for volatile combustion. The sensitivity analysis of CA,
ranged from 4.0 to 0.5, to the coal burnout was carried out by Takeda (1994). It is found
that CA did not have a strong influence on a burnout profile. This parameter was
recommended by Takeda (1994) to be 0.8 because it gave a marginally better fit to the
measured data reported by Suzuki et al. (1984).
When eddy break up model (Spalding, 1971) is adopted, the reaction rate of volatile can
be written as:
1/2
prod
g
gA mε
ρCg
fuk
R (2.32)
where mprod is variance of mass fraction of gas product.
In practice, both models have been widely used in PCI combustion models as shown in
Table 2.1.
2.3.5 Lagrangian approach
For a stable operation of blast furnace with high coal injection rates, accurate
92
information related to the dispersion of coal particles is required, because it is critical
not only to control the devolatilisation and combustion of injected coal particles in the
raceway, but also to protect tuyeres from failure caused by coal abrasion. As indicated
in the 2-D and 3-D PCI combustion models summarised in Table 2.1, the Eulerian
model was used for describing the behaviours of gas phase, while Lagrangian approach
(Kuo, 1986) was used in tracking behaviours of individual coal particle. With the
approach, the coal particles are treated as discrete objects, and their motion is calculated
as the burning coal particles move through the combustion flow field (Toporov, 2014).
In the Lagrangian approach, the trajectory of the particle is calculated by integrating the
force on the particle which can be written as:
where the subscript “ p ” denotes the particulate phase, m is coal particle mass, gμ is
the molecular viscosity, pD is the particle diameter, DC is the drag coefficient given
by Morsi and Alexander (1972), and pRe is the relative Reynolds number, which is
defined as g
prel.gp μ
DρRe
U . The gas velocity fluctuation ( '
gu ) are randomly
sampled by assuming that they obey a Gaussian probability density function, so that
2
gg uζu , 2
gg vζv , 2
gg wζw (2.34)
where ζ is a normally distributed random number, and the remainder of the right–hand
side is the local RMS value of the velocity fluctuations. These values of the RMS
fluctuating components are obtained from solving the turbulence kinetic energy
equation and defined (assuming isotropy) as 3
2kwvu
g2
g
2
g
2
g .
Integration in time of Equation 2.33 yields the velocity of the coal particle at each point
along the trajectory via:
pggpDpgD
ppuReCπDμ
8
1
dt
mdUU
U f (2.33)
93
pi U
dt
dx (2.35)
Equations 2.33 and 2.35 are solved in each coordinates (xi) to calculate the trajectories.
2.3.6 Summary
The sub models used in the CFD based models are comprehensively reviewed in section
2.3. Due to very short residence time for the injected coal particles in the regions of
blowpipe, tuyere and raceway, the coal burnout in the combustion region is mainly
contributed by the volatile evolution, followed by the char combustion. Therefore, in
this research, the performance of single overall reaction and two competing reaction
model are compared, while the kinetic parameters, as well as the stoichiometric
coefficients, are validated by the experimental data reported by Burgess et al. (1985).
The calculated results show that the kinetic parameters and stoichiometric coefficients
used in this research can significantly reflect the coal combustion behaviours in the high
temperature environment.
2.4 Raceway shape
2.4.1 Observation of raceway
A cold model experiment was performed by Inatani et al. (1976) with the aid of a
high-speed camera to investigate the dynamic behaviour of coke in the raceway. As can
be seen in Figure 2.57, the raceway was divided into five regions, from A to E
exhibiting different features of coke movements. Each of tick interval made on the coke
flowing trajectories represents 0.001 second. In the region of the raceway immediately
downstream of the tuyere nose (labelled as section A in Figure 2.57), the resident coke
particles were accelerated by the blast supplied from the tuyere, while new particles
were introduced from the area of the raceway just above the tuyere nose (exit). In
section B, the hot blast generated a recirculation structure which caused the particles
falling in front of the tuyere nose to be caught by the gas stream and carried backward
to impact on the raceway wall (section C). In section D, the coke particles simply
94
circulated around point P. Finally, in section E, the coke particles delivered by the gas
flow accumulated with a high packing density and form an almost impermeable zone
referred to as the “bird’s nest”.
Using an endoscope enclosed by a cooling jacket, studies of the raceway phenomena by
photographic observations parallel and vertical to the axis of tuyere have been carried
out by Greuel et al. (1974) in an operating blast furnace (hot model). The observations
revealed that a hollow (jetting) space was formed in front of the tuyere. As shown in
Figure 2.58, coke particles fell in the flow of the hot blast gas from the tuyere exit, and
were accelerated towards the centre of the furnace. Contrary to the observations in a
cold model reported by Inatani et al. (1976), the coke particles did not circulate in the
raceway. Notably, no coke recirculation within the raceway was also observed in the hot
model experiment of Nogami et al. (2004), which will be discussed in more detail latter.
Owing to short residence time of coke in the raceway, a complete combustion of coke in
the raceway could not be achieved. Instead, it was presumed that main reaction between
the blast oxygen and the coke carbon took place deeper in the furnace. Table 2.6 shows
the comparison of raceway observed by the coal model and the hot models.
Figure 2.57 Schematic illustration of raceway structure.
95
Table 2.6 Comparison of raceway observations in cold and hot models.
To analyse the coal combustion behaviours in the lower zone of blast furnace, a proper
raceway shape should be taken in the calculation model. The raceway shapes used in the
calculation models were determined by means of:
(1) observation (Kuwabara et al., 1981; Takeda and Lockwood, 1997);
(2) tuyere probing (Gu et al., 2008);
(3) DEM modelling (Nogami et al., 2004);
Characteristics
Cold model
by Inatani et al. (1976)
Hot models
by Greuel et al. (1974),
Nogami et al. (2004)
Shape Balloon-like with a jetting space Hollow space with the shape
of curved hose
Entrance of coke Above the tuyere exit Above the tuyere exit
Coke recirculating Yes No
Combustion space
Figure 2.58 Representation of the movement of coke through the raceway.
96
(4) Eulerian approach (Aoki et al., 1993); and
(5) unpublished research (Jamaluddin et al., 1986; Shen et al., 2011).
In early phase of model development in this study (Du et al., 2004), the raceway was
assumed to be a cylindrical jetting space according to the observation by an endoscope
(Greuel et al., 1974). To achieve a more accurate calculation, a mathematical model
based on Eulerian approach coupled with CFD has been developed to predict the
raceway configuration of blast furnace at CSC (Du, 2011).
2.4.2 CSC’s raceway shape prediction model
The raceway prediction model was developed using the Eulerian–Eulerian multi–fluid
model implemented in a commercial CFD code (Du, 2011). It is noted that this model is
based on the fundamental concept of interpenetrating continua for multiphase mixtures
(Gidaspow, 1994). In performing the simulations, the conservation equations (i.e. mass,
momentum, energy and species) are derived by averaging the local instantaneous
balances, and are solved for each individual phase. Moreover, the different phases
present within the same control volume at the same time are characterized by means of
phase volume fractions. In the model, the energy transfer between solid (coke particles)
and gas phases and the heterogeneous reactions within the furnace are taken into
consideration. Calculation approaches for gas and solid phases can be described as
follows:
Gas phase
i. The gas mixture is an incompressible ideal fluid.
ii. The gaseous turbulent combustion rate can be described by the eddy
dissipation model (Magnussen and Hjertager, 1976).
Solid phase (coke particles)
97
i. The coke has the form of granular particles with a smooth and spherical
morphology.
ii. The reaction rate of the coke combustion and gasification processes can be
calculated using the kinetic / diffusion–limited rate model.
iii. The diameter of respective coke particles is kept constant.
iv. The coke degradation mechanisms are ignored.
Furthermore, the considered chemical reactions are shown in Table 2.7. Heterogeneous
reactions of carbonaceous fuels are assumed to be of first–order irreversible with respect
to the oxygen and carbon dioxide. In the model, the conservation equations of mass,
momentum, energy and species are constructed for each phase. The equations for gas
and coke are described in Eulerian coordinate system. Moreover, the different phases
present within the same control volume at the same time are characterised by means of
phase volume fractions. The conservations equations for gas and solid phases can be
written as
he i,ggggg mραρα
tU
(2.36)
he i,sssss mραρα
tU
(2.37)
where the subscript g and s denotes the gas and solid phase; α , ρ , and U are the
volume fraction, density, and mean velocity, respectively. The terms on the right–hand
side of Equations 2.36 and 2.37 express the mass transfer between phases due to
heterogeneous reactions.
Momentum equation:
he i,ssggggg
ggggggg
mβραpα
ραραt
UUUgτ
UUU (2.38)
98
he i,ssgsssss
sssssss
mβραPpα
ραραt
UUUgτ
UUU
(2.39)
where p is the gas pressure, τ is the stress–strain tensor, g is the gravitational
acceleration, β is the interphase momentum transfer coefficient ,and sP is the solid
pressure, which arises due to the both of translation and collision for particles in the
solid phase. The last two terms on the right–hand side of Equations 2.38 and 2.39 are
the source terms due to interaction between phases.
The energy equations for both phases are:
o
ihe i,
o
iho i,gsgg
t
g t,
g p,g
ggggggg
ΔhmΔhωαQTPr
μCλ
hραhραt
U
(2.40)
gssssssssss QTλhραhραt
U
(2.41)
where h is the sensible enthalpy, λ is the thermal conductivity, T is the mean
temperature, pC is the specific heat, sggs QQ is the intensity of heat exchange
between the gas and solid phases, o
ih is the formation enthalpy per unit mass of
species i , and g t,μ is the gas phase turbulence viscosity and is described later. In the
simulations, the turbulent Prandtl number, tPr , is set to be 0.85. For Equation 2.40, the
last two terms on the right–hand side are the source terms due to chemical reactions.
The rate of energy transfer between phases sgQ is modelled according to the correlation
of Gunn (1978).
The species equation can be described below:
he i,ho i,gi g,
t
g t,
i g,ggi g,gggi g,gg mωαYSc
μDραYραYρα
t
U
(2.42)
99
where i g,Y and i g,D are the mass fraction and molecular diffusivity of species i in
the gas phase, respectively. tSc is the turbulent Schmidt number and is set to be 0.7.
The last two terms on the right–hand side of Equation 2.42 are net rates of production of
species i form homogeneous and heterogeneous reactions respectively. Since coke is the
only species of solid phase, the species equation of solid phase is reduced to equation
2.37.
Table 2.7 Chemical reactions in coke–packed furnace model.
Homogeneous reactions (gas–gas) Heterogeneous reactions (gas–solid)
CO + 1/2 O2 → CO2 C + 1/2 O2 → CO
C + CO2 → 2 CO
2.4.2.1 Validation of raceway shape prediction model
To evaluate the performance of the model, the combustion experiment using the coke
bed rig developed by Nogami et al. (2004), as shown in Figure 2.49, is modelled. Table
2.8 indicates the hot blast gas conditions and properties of coke–packed bed in the
calculation. Figure 2.59 compares the numerical and measurement results given by
Nogami et al. (2004). As shown in Figure 2.59a, the predicted raceway shape, which is
featured by the voidage contour of 0.4 (Mondal et al., 2005), for all coke operation is
only slightly smaller than the observation, showing the model is performed acceptable
accuracy in the prediction of raceway shape. Figure 2.59b shows the variation of the gas
compositions along the central axis of the tuyere (i.e. A→B in Figure 2.59a). Note that
for each gas, the symbols and lines denote the experimental observations and the
numerical predictions, respectively. It is evident that the numerical results predicted by
the present model are concurred with the experimental results of Nogami et al. (2004).
100
Table 2.8 Operating conditions and properties of coke–packed bed.
Hot blast Coke
Temperature, K 1100 Diameter, mm 25-35
Flow rate, Nm3/h 710 Density, kg/m
3 1081
Oxygen content, % 22 Initial volume fraction, - 0.55-0.65
Shape factor, - 0.7
2.4.2.2 Prediction of raceway shape in an operating blast furnace
The model was applied to predict the raceway shape of CSC’s No3 BF, in which 32
tuyeres with inner diameter of 140 mm are introduced into the blast furnace at an
inclination angle of 5 degrees in the downward direction. Figure 2.60a shows the
geometry and dimensions of the furnace. In the calculation, only 1/32 sector of blast
Figure 2.59 Comparison of numerical and experimental results (Nogami et al., 2004)
of: (a) raceway shape; and (b) gas composition distribution along central axial of
tuyere (Du, 2011).
(a) (b)
101
furnace corresponding to a single tuyere of the furnace was considered (Figure 2.60b).
The typical operating parameters of the furnace for all coke operation were used as the
inlet boundary conditions (Table 2.9).
Table 2.9 Typical operating parameters of No3 BF for all coke operation.
Blast pressure, atm 4.5
Blast flow mass flow rate, kg/s 3.9
Blast temperature, K 1423
Blast oxygen content, % 21
Coke size, mm 30
Initial voidage 0.4
Figure 2.60 Profile of CSC’s No3 blast furnace: (a) main dimensions (unit: m); (b)
calculation domain of a single tuyere.
(a) (b)
102
Figure 2.61 shows the calculated voidage distribution under all coke operation
conditions of the furnace. In this research, the raceway shape is featured by the voidage
contour of 0.4. It is seen that the depth of the raceway is around 0.95 m. In addition, it is
observed that the raceway is composed of an inlet region with a relatively high voidage
and a transition zone characterized by a reduction in voidage toward the raceway
boundary. The movement of coke is much slower than that of coal particles and hot
blast gas within the raceway, therefore the cave could be thought of as a porous media.
As indicated in Figure 2.62, a simplified computational domain is proposed for the 3D
coal combustion in this work. Obviously the predicted raceway shape is similar with
that observed by Greuel et al. (1974).
a
b
Figure 2.61 Void fraction contours in combustion zone of 3D coke packed
furnace model: (a) top view; (b) side view.
103
2.4.3 Summary of raceway shape prediction
In the cold raceway experiment model of Inatani et al. (1976), the raceway was featured
as a balloon, and it was a hollow space with a shape of cured hose in the hot model
(Greuel et al., 1974; Nogami et al., 2004). Notably, recirculting coke particles are found
in the raceway of cold model. On the other hand, the coal particles in the raceway of the
hot model are blown towards the centre of the furnace.
In this research, the raceway shape and size is determined by Eulerian-Eulerian
multi–fluid model implemented in a commercial CFD code. The measured 2-D raceway
shape reported by Nogami et al. (2004) was applied for the validation of calculated
raceway shape. Based on the operation conditions of CSC’s No3 blast furnace, the
voidage distribution in the raceway are determined. The voidage contour of 0.4 is used
as the boundary of the raceway in this research.
2.5 Injection of biofuel into blast furnace
The utilisation of alternatives to coal in PCI can abate the consumption of fossil fuels
and, in the case of biomass, mitigate CO2 emissions. If biomass can be used in blast
Raceway depth
950 mm
Blowpipe--Tuyere
405 mm 795 mm
Figure 2.62 The simplified calculation domain. Note that αg is the volume fraction
of gas inside the raceway.
104
furnaces as an alternative fuel to coal, it is anticipated that CO2 emissions from the steel
industry and ironmaking can be lessened to a certain extent. However, application of
biomass as an industrial fuel is limited due to its high moisture, low bulk and energy
densities as well as hard grindability. The upgrade of biomass can be fulfilled via
torrefaction and carbonization or pyrolysis where biomass is thermally degraded in an
inert or oxygen-free environment. The torrefaction temperature is in the range of
200-300o C (Peng et al., 2013; Lu et al., 2012; Sabil et al., 2014), and carbonisation is
operated at temperatures of 300–500oC (Abdullah and Wu, 2009). As reported by
Babich et al. (2010), the injection of charcoal fines has been successfully practiced in
some small charcoal blast furnaces in Brazil with injection rates of 100 to 150 kg/tHM.
2.5.1 Combustion experiments and modelling
Combustion experiments of four charcoals and a high volatile coal using the
combustion rig shown in Figure 2.1 (Mathieson et al., 2005) have been carried out by
BSL (Mathieson et al, 2012). Table 2.10 and 2.11 show the key properties of the
samples and operating parameters of the tests respectively. Figure 2.63 provides a
summary of combustion performance of the coal (C-HVM) and the three hardwood
charcoals as a function of the VM of the samples with an air cooled coaxial lance and
interpolated to an O/C of 2.0. Also included are results for previously studied PCI coals
at the same conditions, and the trend of increasing burnout with VM that was
established. It is observed from Figure 2.63 that the burnout of the hardwood charcoal
samples was (a) a function of VM, (b) of similar slope to that for the PCI coals, and (c)
the trend line for the charcoals was located approximately 40% above the trend line for
the PCI coals. The higher performance in burnout given by the charcoals might be
resulted from higher specific areas by heat pretreatment and fragmentation in
combustion. This implies that superior combustion performance and therefore higher
injection rates can be expected for similar hardwood charcoals at relatively low VM
contents. As shown in Figure 2.64, the pressure drop measured across to the tuyere was
105
proportional to the VM of the injectant. This is expected to be related to the amount of
additional gas from the combustion of the volatile matter within the tuyere.
Table 2.10 Key properties of the bulk coal and charcoal samples.
Table 2.11 Key properties of the bulk coal and charcoal samples.
106
Torrefaction and burning characteristics of bamboo, oil palm, rice husk, bagasse, and
Madagascar almond were studied and compared with a high-volatile bituminous coal
using the drop tube furnace developed by Du et al. (2010) at CSC to evaluate the
Figure 2.63 Burnouts as a function of volatile matter of the injectants with an
air cooled lance and O/C = 2.0. Comparison is made with previous results for
PCI coals.
Figure 2.64 Differential pressure across the tuyere as a function of the volatile
matter of the injectants.
107
potential of biomass consumed in blast furnaces (Chen et al., 2012). The schematic of
the reaction system for testing combustibility of fuel is demonstrated in Figure 2.65.
The particle size of solid fuels was in the range of 74 to 149 μm, and the reaction
temperature was 1000oC. The higher heating value (HHV) ratios (enhancement factors)
for raw and torrefied biomasses and coal are listed in Table 2.12. From a calorific point
of view, Madagascar almond is the most sensitive biomass to torrefaction. Specifically,
with torrefaction temperatures of 250 and 300oC, its HHV was amplified by factors of
1.36 and 1.54, respectively. For the torrefaction temperature of 300oC, the calorific
values of the biomasses were close to that of coal, except for rice husk, which had
enhancement factors of around unity. This reveals that 300oC is a feasible operating
condition to transform the biomasses into solid fuels resembling high-volatile
bituminous coal. The burnout versus fuel ratio of the samples is demonstrated in Figure
2.66. The fuel ratio is defined as the content ratio of FC to VM. The fuel ratio was in the
range of 0.13–1.4 in the experiments. Once the biomasses underwent torrefaction, the
burnout tends to decay, whereas the fuel ratio shifted rightwards in the diagram. In
contrast to the experimental results (Figure 6.21) reported by Mathieson et al. (2012),
the burnout of the biofuels did not their exhibit superior performance on burnout as their
fuel ratios were higher than that of HV PCI coal. This might result from the differences
in gas flow patterns in both combustion systems.
108
Table 2.12 Enhancement factor of higher heating value.
Figure 2.65 Schematic of the reaction system (1) cylinder; (2) carrier gas; (3)
secondary gas; (4) rotameter; (5) hopper; (6) preheater; (7) lance; (8) DTF; (9)
thermocouple; (10) ceramic tube; (11) heater; (12) sampling probe; (13) cooling water;
(14)cyclone; (15) residual solid particles; (16) induced suction fan; (17) exhausted gas.
109
The combustion of Taiheyo coal (VM: 44.6%) and Oak char (VM: 27.1%) in a 2-D
furnace (Figure 2.67) was simulated and compared by Wijayanta et al. (2014). Table
2.13 shows the computational conditions. Figure 2.68 are the temperature distributions
for both fuels at 23 wt.% O2 (Figure 2.68a) and at 27 wt.% O2 (Figure 2.68b). The
calculated results showed that Taiheiyo coal achieved a higher temperature distribution
in the furnace than that achieved by Oak char. Obviously, the higher temperature caused
more volatile being released from Taiheiyo coal. In other words, the burnout of the fuels
is strongly related to the volatile contents. Besides, an increase in flame temperatures
for both fuels could be achieved when the oxygen level is increased from 23% to 27%.
Figure 2.66 Distributions of burnout versus fuel ratio of raw and torrefied
biomasses as well as a HV coal.
110
Table 2.13 Computational conditions for biofuel injection.
Figure 2.68 Temperature profiles at an injection rate of 36 (kg solid fuel) / (1000
Nm3 feed gas).
Figure 2.67 Geometry and computational domain used in numerical simulation.
111
2.5.2 Summary of the biofuel injection
(1) The burnout results provided by the turbulent combustion rig indicate the trend line
for the charcoals is located approximately 40% above the trend line for the PCI
coals. The superior performance in combustion may be resulted from higher specific
area of the biofuel by heat pre-treatment and fragmentation in combustion
(Mathieson et al., 2012). Upon inspection of the experimental results by the drop
tube furnace, this advantage is not reproduced by the biofuels tested in the drop tube
furnace (Chen et al., 2012). This may result from the differences in gas flow
patterns in both combustion systems.
(2) The pressure drop measured across to the tuyere is proportional to the VM of the
injectant. This is expected to be related to the amount of additional gas from the
combustion of the volatile matter within the tuyere.
(3) The promotion of torrefaction temperature results in an increase in the calorific
value of biofuel, whereas the burnout of biofuel is decreased.
2.6 Summary
The experimental and numerical studies on pulverised coal combustion in PCI operation
have been reviewed in this chapter. Under conditions simulating blast furnace
environments, the experiments were carried out using the combustion rigs with/without
coke bed to investigate the influence of operation factors, including coal properties, hot
blast gas conditions and injecting facilities, on the coal combustion behaviours.
Alternatively, the drop tube furnace provides a better view at the combustion properties
of coal when one attempts to evaluate the combustion efficiency of coal.
In the 1980s, 1-D mathematical models were established to study the coal combustion
behaviours in raceway. However, the most important phenomena, the dispersion of the
coal particles to hot blast gas were not efficiently simulated. By the 1990s, some studies
have been carried out to develop the 2-D coal combustion models, in which the coal
particle trajectory was determined by Lagrangian approach. In the models, the
112
calculation domain and lance arrangement were assumed to be axi-symmetric. Recently
3-D CFD based models have been intensively developed to deal with the complicated
configurations of lance arrangement and raceway shape. Through parametric studies,
many countermeasures for promoting the coal burnout and the performance of blast
furnace have been proposed.
Practically, improvement of permeability of the raceway is one of key factors to achieve
high productivity with high PCI rates. Notably, none of the previous models consider
the relation between the pressure loss (permeability resistance) and the coal combustion
within the raceway.
2.7 Methodology
To improve the coal combustion efficiency and stability of raceway operation, the
combustion characteristics in the regions of blowpipe, tuyere and raceway were
numerically and experimentally studied in this research. The development of the
calculation model begun with the model validation. Secondly, only blowpipe and tuyere
were considered as the calculation domain. In the third phase, a cylindrical space with
the same diameter of tuyere exit was used as the raceway in the calculstion. In the final
phase, the raceway shape was determined by the Eulerian-Eulerian multi-fluid model.
The factors which influence the pressure loss and coal burnout were especially studied.
In the coal combustion experiment, a drop tube furnace was established to evaluate the
combustion performance of PCI coal, coal blend and biofuel in an environment with
high heating rates (>104 K/s). The volatile release and particle formation characteristics
in the blast furnace were also studied. The experiment results of the drop tube furnace
have been applied by CSC’s blast furnace operators as guidance for the selection PCI
coals and operation conditions.
113
CHAPTER 3
NUMERICAL PREDICTION AND PRACTICAL IMPROVEMENT
OF PULVERIZED COAL COMBUSTION IN BLAST FUTRNACE
A CFD base coal combustion model in the regions of blowpipe and tuyere is developed
and validated in this chapter. The performance of single lance and double lance
injection are compared. Earlier ignition is found with the operation of double lance in
comparison with the single one. Accordingly, the double lance injection became the
standard operation of CSC’s blast furnace in 2002.
Du, S. W. and Chen, W. H. (2006), Numerical prediction and practical improvement of
pulverized coal combustion in blast furnace, International Communications in Heat and
Mass Transfer, vol. 33, p. 327-334.
114
ABSTRACT
The burning characteristics of pulverized coal in blowpipe and tuyere at two different
injection patterns are simulated numerically, to aid improving the practical performance
of blast furnace. With the condition of the same fuel and oxidant mass flow rates, the
predictions indicate that the combustion efficiency of pulverized coal using
double-lance can be substantially enhanced compared with that using single lance.
Accordingly, the pulverized coal injection in a practical blast furnace was modified
from single lance to double-lance. As a result, the practical injection rate of the
pulverized coal in the blast furnace was increased from 110 kg/tHM to 153 kg/tHM,
revealing that a profound decrease in operating cost of the blast furnace has been
implemented.
Keywords: Blast furnace; Pulverized coal; Blowpipe and tuyere; Combustion; Injection.
115
Nomenclature
pA Coal particle surface
pC Specific heat of coal particle
E Activation energy
h Convective heat transfer coefficient
k Reaction rate constant or turbulent kinetic energy
pm Coal particle weight
t Time
pT Temperature of coal particle
T Gas temperature
xi Spatially coordinate
21 , YY Mass fractions of emitted volatile at low and high temperatures
Greek Symbols
ε Dissipation of turbulent kinetic energy
p Emissivity of coal particle
μ Viscosity
σ Stefan-Boltzmann constant (=5.67×10-8
W/m2.K
4)
Subscripts
f Fuel
o Oxidizer
p Coal particle
∞ Gas phase
116
3.1 Introduction
It is known that coal plays an important role in energy developments and industrial
applications. For example, since the industrial revolution occurred in eighteenth century
(Boubel, et al., 1994), a considerable amount of coal has been utilized for power
generation in steam engines. In the last decade of 19th
century, pulverized coal (PC)
began to be used in cement industry for heating drying kilns (Singer, 1984). Nowadays,
pulverized coal is widely applied in coal-fired power plants for producing electricity
(Sami et al., 2001; Hinrichs and Kleinbach, 2002); it is also extensively used in
metallurgical industry for refining metals. As far as the process of ironmaking is
concerned, conventionally, coke, the product of high-temperature pyrolysis of coal,
serves as an important reactant in reducing iron ores into steels in blast furnaces
(Vamvuka et al., 1996). However, because of higher price of coke compared with
pulverized coal, the technique of pulverized coal injection (PCI) from tuyere has been
developed for several years to partially replace the consumption of coke. In other words,
the operating cost of the blast furnace can be substantially reduced if the injection rate
of PC is promoted significantly.
The pulverized coal can be used as auxiliary fuel in a blast furnace and possesses the
merit of reducing operating cost. Nevertheless, it should be addressed that, if coal
particles in combustion zones undergo incomplete combustion, the unburned or residual
char will accumulate in the blast furnace in which the char is depleted by means of
reaction with slag and carbon dioxide (Takahashi et al., 2002; Iwanaga, 1991). If the
accumulation rate of the char in the furnace is larger than the depletion rate, the
movement of hot blast will be retarded. This results in a pressure fluctuation which
further suppresses the operation of the blast furnace. In consequence, enhancing the
burning rate of PC and reducing the accumulation ratio of unburned char is one of
available methods to stabilize the performance of the blast furnace.
Because the operation of PCI is highly relevant to the performance of the blast furnace,
the purpose of the present study is to predict the combustion characteristics of
117
pulverized coal in a blast furnace through numerical simulation. By varying the
injection pattern of PC, its impact on the burning behaviors of the PC in the blast
furnace will be evaluated. Furthermore, based on the obtained results, a practical
strategy in improving PC combustion will be adopted.
3.2 Mathematical Formulation
3.2.1 Burning process of pulverized coal
A schematic diagram of the internal structure of a blast furnace is demonstrated in
Figure 3.1. Attention of the present study is focused on the pulverized coal combustion
in the regions of blowpipe and tuyere in the blast furnace. As shown in the Figure, when
coal particles are injected into the blowpipe, they will immediately immerse in a
high-temperature environment filled with hot blast and thereby experience rapid
heating, devolatilization reaction of the coal, oxidization of the volatile matters with hot
blast, combustion of residual unburned char, and gasification of the char. Recognizing
the above characteristics, it is known that the devolatilization reaction initiates coal
combustion, implying that the selection of parameters to model the devolatilization
reaction is of the utmost importance in predicting the PC combustion. Therefore, in the
current study, the initial chemical reaction of coal particles will be tested and verified to
ensure the validation of the numerical method.
3.2.2 Momentum and energy balance of a coal particle
Considering a moving coal particle, when it is assumed to be spherical and the
Lagrangian framework is used, the trajectory of the particle can be obtained by solving
a single particle momentum equation. That is, the rate of change of momentum is equal
to external forces on the particle. On account of very small coal particles investigated, it
is proper to neglect body force and only drag force is considered during computation.
Consequently, the equation of motion of the particle is expressed as:
118
p
p
p Fdt
vdm (3.1)
In regard to the energy balance, with conceiving the coal particle as a lump system,
the heating of the particle is carried out by convection and radiation; thus the
temperature of the particle can be described by the energy equation as the
following:
)()(44
ppppp
p
pp TTATThAdt
dTCm (3.2)
3.2.3 Model of devolatilization of coal particle
When one is concerned with the devolatilization process of coal particles, it depends
strongly on the heating rate, reaction time, and coal grade, and so forth. In fact, as the
heating process is fast, the volatile matters emitted from the coal is larger than the
analyzed result of ASTM, rendering that Q factor is larger than one. To describe the
devolatilization process more realistically, two-competing devolatilization model
Figure 3.1 A schematic diagram of internal structure in a blast furnace.
119
(Smoot and Smith, 1985) is employed. The two parallel and competing reactions are
given as follows:
) ( )1( 1111 etemperaturlowVolatileYCharYcoal
k (3.3)
) ( )1( 2222 etemperaturhigh VolatileYCharYcoal
k (3.4)
Furthermore, the reaction kinetics is written by:
CoalYkYkdt
dV )( 2211 ; )/exp( 111 pRTEAk and )/exp( 222 pRTEAk (3.5)
where V and R are mass fraction of volatile matter and universal gas constant,
respectively. In examining the preceding model, it is apparent that the parameters Y1, k1,
Y2, k2, E1, and E2 have a vital influence in predicting the devolatilization process. The
appropriate values will be suggested later.
3.2.4 Turbulent combustion model
In the gas phase the fluid motion is fast, the k model is thus applied to simulate the
turbulent combustion. In the operation of PCI, following the release of volatile matters
from coal particles, oxygen will encompass the volatile, yielding the diffusion flame
combustion. In such a situation, mass fraction probability density function (PDF) model
(Kobayashi et al., 1977) is an appropriate method to approach the reaction phenomena.
The model is established based on the concepts of mixture fraction, mix-is-burnt
(Eghlimi and Sahajwalla, 1997; Zhou, 1993), and probability density function. For a
system just having two reactants, consisting of fuel and oxidant, the PC combustion can
be approximated by a single-step reaction as:
ProductkgiOxidantkgiFuelkg )1( 1 (3.6)
The coefficient i represents the stoichiometric balance between the fuel and oxidant.
When the turbulent transport coefficients of reactant and oxidant in the flow field are
summed to be equivalent, employing the Zeldovich transformation the combined mass
fraction X can be obtained as the following:
iMMX of / (3.7)
120
and the mixture fraction f is defined by:
0
0
XX
XXf
f
(3.8)
where Ofof XXMM and , , , stand for mass fractions of fuel and oxidant as well as
combined mass fractions on the fuel and oxidant sides, respectively. The mixture
fraction f is a conservative scalar, and its value at a control volume can be calculated via
the solution of its instantaneous conservation equation for f (time-averaged):
m
it
t
i
i
i
Sx
f
xfu
xf
t
)()()(
(3.9)
In the above equation, mit Sx and , , , designate density (g/cm3), dynamics viscosity
(N.s/m2), spatial coordinate, and source term stemming from the reaction of coal into
the gas phase, respectively. Meanwhile, t is a computational parameter whose value
is given by 0.9 (Zhou, 1993). In the framework of PDF, mean square value of
concentration fluctuation g can be calculated through the following equation
gk
Cx
gC
x
g
xgu
xg
tdtg
it
t
i
i
i
2
)()()( (3.10)
where gC and dC are the computational parameters and they are given by 2.8 and
2.0, respectively. According to mixture fraction f, molar fraction of each gas species,
density, and temperature in control volume can be calculated.
3.3 Results and discussion
3.3.1 Numerical validation and parameter selection
Previous to simulating the physical phenomena, accurate selection of the parameters in
the devolatilization model has to be carried out. To achieve this goal, the presently
predicted results are compared with the experimental data of Burgess et al. (1983) to
confirm the validation of the simulation. The investigated conditions are summarized in
Table 3.1. In the meantime, two sets of parameters, reported by Kobayashi et al. (1977)
and Ubhayakar et al. (1976), are tested for comparison each other. Details of the
121
parameters are given in Table 3.2 and the predicted temperature distributions in the
blowpipe and tuyere are displayed in Figure 3.2. It depicts that the temperature is
spatially uniform when using the parameters of Kobayashi et al. (1977), implying that
the devolatilization reaction in the reactor is not exhibited. Clearly, the preceding result
is inconsistent with the experimental measurement. Regarding the parameters of
Ubhayakar et al. (1976), as shown in Fugure 3.2, the predicted temperature distribution
is close to the experimental data. It follows that the proposed parameters of Ubhayakar
et al. (1976) is capable of providing a more realistic prediction. Because of this, their
parameters are employed in the current study.
In examining the calculated temperature distribution, it is noteworthy that the curve is
characterized by monotonic increase with increasing distance away from the lance exit.
The profile can be partitioned into two stages; it composes of rapid rise in the upstream
region and progressive increase in the downstream one. The behavior in the first region
arises from the PC devolatilization or pyrolysis reaction followed by the combustion of
the emitted volatile matters with oxygen. In regard to the second region, the slow
increase in temperature is attributed to the reaction between char and oxygen. In a word,
after the PC is blown into the blowpipe, the chemical reaction is primarily achieved by
the gas-phase combustion (i.e., homogeneous reaction) and then implemented by the
solid-phase oxidation (i.e., heterogeneous reaction).
Table 3.1 Operational conditions selected by Burgess et al. (1983)
Reactor diameter, mm 50
Hot blast temperature, K 1243
Hot blast velocity, m/s 68
Coal particle diameter, μm 40
Volatile matter of PC (db), % 35.9
PC injection rate, kg/h 5.3
122
Table 3.2 Two sets of parameters used for predicting PC devolatilization.
Kobayashi et al. (1977) Ubhayakar et al. (1976)
1Y 0.3 VM
2Y 1 1.5×Y1
A1, 1/s 200000 3.7×105
A2, 1/s 1.3×107 1.46×10
13
1E , kJ/mol 1.046×102 74
2E , kJ/mol 1.674×102 251
Distance (m)
Ga
ste
mp
era
ture
(K)
0 0.2 0.4 0.6 0.8 1800
1000
1200
1400
1600
1800
2000
Kobayashi et al.
Ubhayakar et al.
Burgess et al.
2nd stage1st stage
Figure 3.2 A comparison of gas temperature distribution among experimental
measurement and two devolatilization models.
123
3.3.2 Impact of injection pattern
As mentioned previously, the emphasis of the present work is upon the burning
characteristics of pulverized coal in blowpipe and tuyere. To evaluate the behaviors of
PC combustion with different injection patterns, a typical running condition of the blast
furnace at China Steel Corporation (CSC), as shown in Table 3.3, is simulated.
Meanwhile, the physical geometries of the blowpipe and tuyere are illustrated in
Figure3.3. When attention is placed on the influence of injection pattern upon the PC
combustion, two different cases, consisting of single lance and double-lance, are
calculated where the mass flow rates of PC and carrier gas are fixed. Accordingly, the
diameters of the lances in the cases of single lance and double-lance are 20 and 14 mm,
respectively. To provide a reference for indicating combustion efficiency, the burning
ratio of PC is defined as:
%100(%) i
e
M
Mratioburning (3.11)
where Me and Mi are the PC weight-loss at the exit of tuyere and the original PC weight
at the entrance of blowpipe, respectively. The calculations suggest that, once the single
lance is modified to the double-lance, the burning ratio is substantially promoted from
4.9% to 12.2%. To proceed farther into the recognition of the burning mechanisms,
Figure 3.4 displays the isothermal contours in the blowpipe and tuyere in accordance
with the performances of the single lance and the double-lance. In the both cases,
because the PC and carrier gas are at room temperature prior to entering the blowpipe,
the temperatures in the vicinity of the entrance are relatively lower, as observed. When
comparing the isothermal contours in the downstream, it can be found that the ignition
of the latter case occurs earlier than that of the former. This obviously reflects that the
operation of the double-lance can facilitate the mixing between the PC and hot blast,
whereby the production rate of unburned char adjacent to the exit of the tuyere is
reduced.
124
Table 3.3 Operating conditions of PCI at CSC.
Hot blast conditions
Temperature: 1423 K; Pressure: 4.5 atm;
Mass-flow-rate: 3.9 kg/s; Oxygen content: 21 %.
Properties of PC
FC: 55.09%; VM: 35.13%; Ash: 6.23%; Moisture:
3.55%.
Particle distribution of
PC
90μm: 5%; 63μm: 25%; 45μm: 55%; 20μm: 15%.
Others Lance angle: 15o; Lance internal diameter: 20mm;
Carrier gas mass-flow-rate: 0.026 kg/s; PC injection
rate: 0.4 kg/s; Heat loss of tuyere: 900,000 W/m2.
180 mm
450 mm
140 mm
100 mm
Lance exit Tuyere Blowpipe
Figure 3.3 A schematic diagram of blowpipe and tuyere as well as their sizes.
125
3.3.3 Practical improvement of blast furnace
The aforementioned results have provided a practical insight into the performance of the
blast furnace. Based on the simulations, the injection pattern in one of the blast furnaces
in CSC has been the redesigned through changing the single lance to the double-lance.
After that, the injection rate of PC has been promoted to a great extent, from 110
kg/tHM (ton of hot metal) to 153 kg/tHM. For this reason, the goal of reducing the
operating cost of the blast furnace has been accomplished sufficiently.
PC
Blowpipe Tuyere
(a)
(b)
Figure 3.4 Isothermal contours in blowpipe and tuyere under the operations of (a)
single lance and (b) double-lance injections.
126
3.4 Conclusions
By utilizing two different injection patterns of pulverized coal, the burning
characteristics of the pulverized coal in the blast furnace have been examined. The
numerical simulations elucidate that the performance of PCI by means of double-lance
is capable of providing a superior burning, in contrast to the original single lance
design. This is attributed to the fact that the double-lance injection is conducive to
mixing between pulverized coal and hot blast, resulting in earlier ignition of the fuel.
The practical injection pattern of the PC in the blast furnace was modified from the
double-lance to the single lance, in accordance with the foregoing numerical
predictions. As a result, the injection rate of the PC has been amplified by a factor of
40%, from 110 kg/tHM to 153 kg/tHM. In summary, the numerical study has provided a
useful insight into the practical improvement of the blast furnace performance.
127
CHAPTER 4
PERFORMANCES OF PULVERIZED COAL INJECTION IN
BLOWPIPE AND TUYERE AT VARIOUS OPERATIONAL
CONDITIONS
The factors affecting the coal combustion are numerically studied in this chapter. The
calculated results provide useful insights for the assessment of blast, tuyere and PCI
operation conditions assisting improvement of coal burnout.
Du, S. W., Chen, W. H.and Lucas, A. J. (2007), Performances of pulverized coal
injection in blowpipe and tuyere at various operational conditions, Energy Conversion
and Management, vol. 48, p. 2969-78.
128
ABSTRACT
Combustion efficiencies of pulverized coal in blowpipes and tuyeres under various
operational conditions are numerically predicted to recognize the performance of
pulverized coal injection in a blast furnace. A variety of parameters including injection
pattern of pulverized coal, oxygen content in hot blast, inlet temperature of the hot blast,
and mass flow rate of coal carrier gas are taken into consideration. The effect of
installing a ceramic sleeve around the tuyere on the pulverized coal combustion is also
evaluated. The predictions indicate that pulverized coal combustion is highly related to
the injection pattern, hot blast temperature, mass flow rate of the carrier gas, and
installation of ceramic sleeve, whereas it is insensitive to the oxygen concentration. The
present study is carried out based on the practical operational conditions of the blast
furnace at the China Steel Corporation. Consequently, the obtained results have
provided a useful insight into the operation of pulverized coal injection for improving
the blast furnace performance in the future.
Keywords: Blowpipe and tuyere; Blast furnace; Pulverized coal; Injection; Combustion
129
Nomenclature
pA Coal particle surface
C Coal
pC Specific heat of coal particle
E Activation energy
f Mixture fraction
F Fuel
h Convective heat transfer coefficient
k Reaction rate constant or turbulent kinetic energy
pm Coal particle weight
M Mass fraction
O Oxidant
R Universal gas constant
S Source term
t Time
pT Temperature of coal particle
T Gas temperature
v Mass fraction of solid lost as volatiles
V Volatile
X Combined mass fraction
xi Spatially coordinate
21 , YY Stoichiometric coefficients of emitted volatiles at low and high
temperatures
Greek Symbols
130
ε Dissipation of turbulent kinetic energy
p Emissivity of coal particle
μ Viscosity
σ Stefan-Boltzmann constant (=5.67×10-8
W/m2.K
4)
ρ Density
Subscripts
f Fuel
o Oxidizer
p Coal particle
∞ Gas phase
131
4.1 Introduction
Pulverized coal (PC) has become an important auxiliary fuel in the iron and steel
industry since the technique of pulverized coal injection (PCI) was developed for
ironmaking (Babich et al., 1996). When pulverized coal is injected into blast furnaces
through blowpipes and tuyeres, because of the reactions of devolatilization, gasification,
and combustion as well as the formation of unburned char, the coal becomes sources of
heat and reductant in raceways (Ohno et al., 1994). For this reason, PC is extensively
employed in blast furnaces to partially replace the metallurgical coke at the present time
(Babich et al., 1996; Ohno et al., 1994; Chung and Hur, 1997). In fact, utilizing PCI
also possesses the advantages of reducing operation costs of blast furnaces and
decreasing emissions of carbon dioxide. This arises from the fact that the price of coal is
relatively lower than that of coke and, from the viewpoint of energy conversion and
management, the PCI is more efficient than metallurgical coke. Accordingly, under the
situation of stabilizing blast furnaces, how to extend the PC injection rate for increasing
yield in the ironmaking process has become a prime concern for blast furnace engineers
(Babich et al., 1996; Ariyama et al., 1994).
When one attempts to recognize the combustion situations of PC in a blast furnace in
order to evaluate the performances of the furnace at various operating conditions, it is
practically difficult by virtue of intrinsic high-temperature environment and close
system of the furnace. Also the practical analysis will expose one to a high risk
environment. Over the past several decades, on account of rapid progress in computer
simulating capability as well as advanced development in numerical algorithm,
computational fluid dynamics (CFD) has become a powerful tool to aid understanding a
variety of scientific and industrial phenomena. For instance, Chen et al. (2000a; 2000b)
developed a numerical code with multi solid progress variables to simulate coal
gasification in an air blown entrained flow gasifier. Their simulation illustrated that
carbon conversion was independent of devolatilization rate, but sensitive to the
chemical kinetics of heterogeneous reactions on char surfaces, and less sensitive to a
132
change in coal particle size. On the other hand, Choi et al. (2001) numerically predicted
coal gasification in an entrained flow gasifier with slurry feed. By varying O2/coal ratio,
their developed models in predicting carbon conversions and syngas concentrations
agreed with the measured results. Aside from the difficulty of practical analysis and the
safety issue mentioned above, other obvious merits of CFD include that the numerical
predictions can be achieved in a short time and their analyses are much more economic
than the practical measurements.
In order to increase the performance of blast furnaces, according to the realistic
operation of a blast furnace at China Steel Corporation (CSC) the preset study is
intended to investigate PC combustion in blowpipes and tuyeres by means of a
numerical simulation. A variety of operating parameters, composed of injection pattern
of PC, oxygen content and inlet temperature of hot blast, as well as mass flow rate of
coal carrier gas, will be taken into consideration to account for their impacts on the
burning behaviors of PC. The PC combustion in the blast furnace and the
presence/absence of installing ceramic sleeve is also evaluated. From these predictions,
the obtained results will be adopted as potential countermeasures to enhance the
productivity of the blast furnace in the future.
4.2 Methodology
Schematic diagram of the internal structure of the investigated blast furnace is
demonstrated in Figure 4.1. Attention of the present study will be focused on the
pulverized coal combustion in the regions of blowpipes and tuyeres. Physically, the
phenomena involve fluid dynamics and reactions of the gas phase and solid particles.
The relevant governing equations are stated below.
133
4.2.1 Gas-phase continuity and momentum equations
In the gas phase, with the assumptions of steady-state flow and Newtonian fluid, the
continuity and momentum equations are:
m
pi
i
Sux
(4.1)
and
u
p
j
i
ji
ji
i
Sx
u
xx
puu
x
(4.2)
where the subscript p shows sources to the corresponding property due to the presence
of particulate phase. For the accurate and efficient predictions of the turbulent mixing
and dispersion of injected pulverized coal into the hot blast (>150 m/s), the RNG
Figure 4.1 A schematic diagram of internal structure in blast furnace.
134
(Re-Normalization Group) k model is adopted to predict turbulent combustion
(Biswas and Eswaran, 2002). This is because that the RNG k model can provide a
better treatment in the mixing and dispersion of coal particles in comparison with the
conventional k model.
Considering the operation of PCI, when coal particles are heated, volatile matters will
be released to react with oxygen, resulting in diffusion flame combustion (Khalil, 1982).
To approach the gaseous combustion, it is proper to assume that the chemical reactions
are fast compared to fluid mixing rate (mixed-is-burned); hence the mixture fraction
probability density function (PDF) model (Sivathanu and Faeth, 1990) is employed. In
the mixture fracture PDF frame, individual species transport equations are not
considered. Instead, the mixture fraction transport equation is solved. The mixture
fraction, f, can be written in terms of the mass fraction as
iOiF
iOi
XX
XXf
(4.3)
where Xi represents the mass fraction for some element i and the subscripts F and O
respectively stand for the values at the fuel and oxidant sides. The mixture fraction f is a
conserved scalar and its value in a control volume can be calculated from the solution of
its time-averaged ( f ) instantaneous conservation equation:
p
it
t
i
i
i
Sx
f
xfu
xf
t
(4.4)
In the above equation, pt S and designate dynamics viscosity and source term,
stemming from the reactions of coal in the gas phase, respectively. Meanwhile, t is a
computational parameter whose value is given by 0.9 (Jones and Whitelaw, 1982). In
the framework of PDF, mean square value of concentration fluctuation g can be
calculated through the following equation:
gk
Cx
gC
x
g
xgu
xg
tdtg
it
t
i
i
i
2
(4.5)
where gC and dC are the computational parameters and they are given by 2.8 and
2.0, respectively (Jones and Whitelaw, 1982). According to the mixture fraction f, molar
135
fraction of each gas species, density, and temperature in every control volume can be
obtained.
4.2.2 Coal particle momentum and energy equations
When coal particles are injected into a blowpipe via injection lance, the pathway of
individual moving particle can be tracked by solving a single particle momentum
equation. It is known that a particle’s motion is subjected to relative velocity between
the gas phase and solid phase. If the coal particle is assumed to be spherical and its body
force is neglected, resulting from investigating very small particle, the equation of
motion of the particle can be expressed by means of Lagrangian framework as the
following:
ppDp
p
p uuuuCddt
dum
8
1 2 (4.6)
In regard to the energy balance, if we postulate that the coal particle is a lump system
(Myers, 1971), the heat conducted into the whole particle is equivalent to the convective
and radiative heat transfer onto the particle surface; thus the temperature of the particle
can be described by:
44
ppppp
p
pp TTATThAdt
dTCm (4.7)
In general, the residence time of coal particles in combustion zone is about 20 ms
(Steiler, et al., 1996) where the environmental temperature is very high. The result is
that the particles experience rapid heating. Specifically, the heating rate of coal particles
is commonly in the order of 105 K/s. Following the rapid heating, hot blast surrounding
the particles will trigger a sequence of physical and chemical reactions. The reactions
include devolatilization of the coal, combustion of volatiles and unburned char, as well
as gasification of the char (Steiler et al., 1996). It has been illustrated (Smoot and Smith,
1985) that significant devolatilization will be started when temperature is as high as
about 650K. Since the characteristic times of coal devolatilization and volatiles
combustion are much shorter than that of unburned char combustion and gasification
136
(Smoot and Smith, 1985), the devolatilization process becomes dominant mechanism in
the initial period of injection. To describe the devolatilization process, the following
parallel, first order, irreversible reactions (Kobayashi et al., 1977) are employed:
) ( )1( 11111 etemperaturlowVYSYC
k (4.8)
) ( )1( 22222 etemperaturhigh VYSYC
k (4.9)
where C, Y, S, and V denotes coal, stoichiometric coefficient, char, and volatile,
respectively. The relative importance of the two equations is mainly determined by
temperature. In other words, as long as the temperature is low, the reaction is dominated
by Equation 4.8. Alternatively, it is governed by Equation 4.9 once the temperature is
relatively high. Accordingly, the reaction kinetics is written to:
CYkYkdt
dv 2211 (4.10)
pRTEAk /exp 111 (4.11)
pRTEAk /exp 222 (4.12)
In the aforementioned equations, v is the mass fraction of volatiles, and the reactions are
characterized by 21 EE . The framework of the CFD code and overall computational
procedures of the gas phase and the solid phase are illustrated in Figure 4.2. In
examining the preceding model, it is apparent that the parameters Y1, k1, Y2, k2 , E1, and
E2 have a vital influence in predicting the devolatilization process (Du and Chen, 2006).
In the present study, two sets of parameters respectively recommended by Kobayashi et
al. (1977) (Model 1) and Ubhayakar et al. (1977) (Model 2) are applied to predict PC
combustion and they are compared with the experimental data of Matheson, et al.
(2005). The predicted results under the situations of low and high PC injection rates are
plotted in Figures 4.3a and 4.3b, respectively. The operational conditions and the
properties of the coal are also included in Figure 4.3a. Obviously, the model 2 is
capable of providing accurate predictions in the current phenomena when compared to
137
the model 1. Consequently, the model 2 is employed in the study. The profiles of the
model 2 shown in Figure 4.3 also suggest that the PC combustion can be simulated well
in the developed numerical method.
Gas phase
˙ Mass conservation
˙ Momentum conservation
˙ Turbulence model
˙ Energy conservation
˙ Turbulence combustion
model
Solid Phase
˙ Devolatilization model
˙ Char combustion
˙ Reaction-rate controlling
regime
˙ Momentum conservation
(particle trajectory)
˙ Energy conservation
Initialization
˙ Geometry (meshing)
˙ Boundary conditions
˙ Sub-models
˙ Energy conservation
˙ Turbulence combustion
model
Solution
Radiative Transport
Particle
properties
Radiative
energy
exchange
Figure 4.2 Framework of the CFD code and computational procedure of the gas
phase and solid (coal particle) phase.
138
4.3 Results and discussion
In general, pulverized coal combustion in blowpipes and tuyeres is likely to be
influenced when certain important operating parameters are altered. The present work is
aiming to provide a reference for improving the performance of PCI. To recognize the
burning behaviors of coal particles under various operational parameters, a baseline
case, in accordance with the typical running conditions of a practical blast furnace in
Figure 4.3 Gas temperature distributions for pulverize coal burning in a reactor
from experimental measurement and numerical predictions.
139
CSC is chosen. The operation conditions (Table 4.1) of the case are used as the
boundary conditions in the baseline calculation. Detailed physical geometries of the
investigated blowpipe and tuyere are sketched in Figure 4.1 as well. In the following
discussion, moving dynamics for coal particles in blowpipes and tuyeres as well as the
influences of some possibly important operating parameters on the particles burning
will be examined.
Table 4.1 Operating conditions (base case) of PCI at CSC.
Hot blast conditions Temperature: 1423 K;
Pressure: 4.5 atm;
Mass flow rate: 3.9 kg/s;
Oxygen content: 21 %.
Proximate analysis of
pulverized coal
FC: 55.09%;
VM: 35.13%;
Ash: 6.23%;
Moisture: 3.55%.
Particle distribution of
pulverized coal
90μm: 5%;
63μm: 25%;
45μm: 55%;
20μm: 15%.
Other conditions Lance angle: 15o;
Lance internal diameter: 20mm;
Carrier gas mass flow rate: 0.026 kg/s;
PC injection rate: 0.4 kg/s;
Heat loss of tuyere: 900,000 W/m2.
140
Figure 4.4 Trajectories and residence times of coal particles under the operation of
the base case.
4.3.1 Trajectories and residence times of coal particles
Figure 4.4 demonstrates the trajectories and residence times of coal particles under the
operation of the base case. Upon inspection of the trajectories, it is evident that the
mixing between coal particles and hot blast is insufficient in the blowpipe and the
upstream of the tuyere. This will delay the ignition of the fuel and result in poorer
reactions. Figure 4.4 also reveals that the residence times of the particles in the
blowpipe and tuyere range from about 4 to 7 ms.
4.3.2 Injection pattern
The preceding observation suggests that if the mixing between the solid phase and the
gas phase can be intensified to a certain extent, it will be conducive to heating,
141
devolatilization, and combustion of the PC. At present, three different methods,
including injection through single lance (base case), single lance with larger diameter,
and double-lance are calculated and compared with each other. They are denoted by
cases 1, 2, and 3, sequentially. Since the mass flow rates of the PC and carrier gas are
controlled to be the same in all the cases, the diameters of the lances corresponding to
the three cases are 20, 25.4, and 14 mm, respectively. On the other hand, to figure out
the performance of the PCI, a parameter of burning ratio (BR) is evaluated which is
defined as the ratio between the PC weight-loss at the exit of tuyere to the original PC
weight at the entrance of blowpipe. As seen in Figure 4.5, once the diameter of the
single lance is enlarged from 20 mm (case 1) to 25.4 mm (case 2), the BR will be
increased from 4.9% to 8.4%. It is inferred that the enhancement of the PC
combustion is due to the decrease in the velocities of the particles, thereby elongating
their residence times. In addition to increasing residence time, physically, if the
originally single lance is divided into double-lance, it will facilitate particles dispersion
and thereby enlarge the mixing between fuel and oxidant. For this reason, as shown in
Figure 4.5, when the single lance injection is modified into the double-lance injection
(case 3), the BR is further promoted to 12.2%.This argument can be verified by
examining the practical combustion situations of PC in cases 1 and 3 which are
displayed in Figure 4.6. From the Figures, it is obvious that coal particles extending
outward away from the centerline of the tuyere in case 3 is much more pronounced than
that in case 1. As a whole, the relative values of the BR in the three cares are 1:1.7:2.5,
revealing that an appropriate design in lance arrangement is capable of sufficiently
increasing the performance of PCI.
142
Figure 4.5 Burning ratios of PC at various injection patterns.
(a) (b)
Figure 4.6 Combustion situations of pulverized coal in (a) case 1 and (b) case 3.
143
hot blast temperature (K)
bu
rnin
gra
tio
(%)
1350 1400 1450 1500 1550 16000
2
4
6
8
10
12
4.9%(base case)
6.2%
9.3%
mass flow rate of carrier gas (kg/s)
bu
rnin
gra
tio
(%)
0 0.01 0.02 0.03 0.04 0.05 0.060
2
4
6
8
10
7.2%
2.5%
4.9%(base case)
oxygen content (%)
bu
rnin
gra
tio
(%)
21 23 25
4.6
4.8
5
5.2
5.4
4.9%(base case)
5.1% 5.1%
4.3.3 Oxygen concentration and hot blast temperature
Intuitively, pulverized coal combustion in blowpipes and tuyeres will be intensified if
oxygen concentration in hot blast is increased. To understand the role played by oxygen
concentration upon the PC combustion, the burning ratios of PC at three different
oxygen concentrations, consisting of 21% (base case), 23%, and 25%, are simulated and
demonstrated in Figure 4.7. The Figure depicts that the BR is increased a bit as the O2
concentration is enriched from 21% to 23%. Specifically, only 0.2% of increment in the
BR is developed. Once the O2 concentration is further enlarged to 25%, it is noteworthy
that the BR ceases increasing. The reason causing this feature is that the characteristic
times for particles traveling in the tuyere are very short, as illustrated in Figure 4.4,
whereas time required for unburned char to react with oxygen is much longer.
Consequently, coal devolatilization becomes the dominant reaction in the tuyere and the
reaction is insensitive to the oxygen concentration. It follows that the enriched oxygen
almost plays no part in the enhancement of PC combustion.
Figure 4.7 Burning ratios of pulverized coal at various oxygen concentrations.
144
4.3.4 Hot blast temperature
Considering the effect of hot blast temperature, unlike the oxygen concentration, Figure
4.8 reveals that increasing temperature has a remarkable effect on the BR. In other
words, corresponding to the hot blast temperatures of 1423K (base case), 1473K, and
1523K, the values of BR are 4.9%, 6.2%, and 9.3%, respectively. This arises from the
fact that the utilized devolatilization model is principally determined by particle
temperature, which is closely related to the hot blast temperature. Therefore, the higher
the hot blast temperature, the more effective the coal reactions, as predicted.
4.3.5 Mass flow rate of carrier gas
For the mass flow rate of carrier gas, Figure 4.9 shows the burning ratios of the PC
at the mass flow rates of 0.015 kg/s, 0.026 kg/s (base case), and 0.05 kg/s,
respectively. In contrast to the base case, when the mass flow rate is decreased to
0.015 kg/s, the BR grows greatly, from 4.9% to 7.2%. Alternatively, when the
mass flow rate is increased to 0.05 kg/s, the BR declines to 2.5%. This elucidates,
in short, that the BR rises markedly as long as the mass flow rate of the carrier gas
is decreased. This is the result that the inlet carrier gas is in the state of room
mass flow rate of carrier gas (kg/s)
bu
rnin
gra
tio
(%)
0 0.01 0.02 0.03 0.04 0.05 0.060
2
4
6
8
10
7.2%
2.5%
4.9%(base case)
oxygen content (%)
bu
rnin
gra
tio
(%)
21 23 25
4.6
4.8
5
5.2
5.4
4.9%(base case)
5.1% 5.1%
hot blast temperature (K)
bu
rnin
gra
tio
(%)
1350 1400 1450 1500 1550 16000
2
4
6
8
10
12
4.9%(base case)
6.2%
9.3%
Figure 4.8 Burning ratios of pulverized coal at various hot blast temperatures.
145
temperature. As soon as the gas is blown into the blowpipe, contrary to the
function of the hot blast, the cooling effect stemming from the carrier gas will be
exhibited, rendering the coal reactions being suppressed. For PC in CSC, its
transport pertains to dilute phase transportation (DPT). In order to avoid PC choke
in pipes, the blown amount of coal particles is generally controlled below 18 kg per
kg of carrier gas. Therefore, despite the advantage of decreasing the mass flow rate
of the carrier gas observed above, it should be emphasized that we are still unable
to noticeably reduce the mass flow rate. Furthermore, according to the present
simulation and under the limitation mentioned above, how to promote the
efficiency of the pulverized coal injection via decreasing the mass flow rate of the
carrier gas will become an important and practical issue in the near future.
4.3.6 Installation of ceramic sleeve
Apart from the operations, redesigning equipment is another achievable method to
improve the performance of PCI. In the practical operation of the blast furnace, cooling
water is commonly used in the tuyere to lessen the damage caused by high-speed PC
erosion. Recently, a new method of installing ceramic sleeve around tuyere was
Figure 4.9 Burning ratios of pulverized coal at various mass flow rates of carrier gas.
146
developed in CSC to diminish the damage. Because the ceramic sleeve possesses an
adiabatic feature, one is able to assume that no heat is transferred through the wall of the
sleeve and this assumption is adopted in the simulation. As a result, the BR is promoted
to 7.7% when compared to the base case of 4.9% which is not adiabatic situation. In
consequence, the installation of the ceramic sleeve can simultaneously protect the tuyere
and promote PC burning while isolating the energy.
4.4 Conclusions
Proper operations of PCI are highly related to fuel consumption and energy
management in blast furnaces. By selecting various operating conditions in a blast
furnace, consisting of injection pattern, oxygen concentration and inlet temperature of
hot blast, mass flow rate of carrier gas, and construction of ceramic sleeve, the burning
characteristics of the pulverized coal in blowpipe and tuyere in the blast furnace has
been extensively examined. When tracking the trajectories of coal particles, it is
observed that the mixing of the coal particles and hot blast plays a crucial role in
heating, devolatilization, and combustion of the PC. Considering the injection form, the
performance of PCI by means of single lance with larger diameter or double-lance is
conducive to providing a superior burning, in contrast to the original single lance
design. This results from coal particles characterized by longer residence times and
better mixing between the fuel and oxidant. The improvement in PC burning is
especially significant for the double-lance injection. Besides, the simulations also
indicate that either increasing hot blast temperature or decreasing the mass flow rate of
carrier gas enable us to promote the burning ratio of the PC in the tuyere. However, in
view of the dominant reaction of the devolatilization in the tuyere, the combustion
efficiency of PC is hardly affected at all when the concentration of oxygen is enriched.
With regard to the design of device, the numerical prediction reveals that the installation
of adiabatic ceramic sleeve can reduce the heat loss of hot blast effectively, whereby the
147
PC burning ratio is enhanced greatly. The present study has provided a number of
practical insights into the improvement of blast furnace performance.
148
CHAPTER 5
PRACTICE OF HIGH PRODUCTIVITY AT NO3 BLAST FURNACE
OF CHINA STEEL CORPORATION
The coal blend (mixture of low and high volatile coals) combustion within a simplified
raceway is analysed through CFD in this chapter. The pressure loss due to the coal
combustion in the raceway can be abated when the coal blend is practiced. It was
confirmed by the plant trials at CSC’s No3 blast furnace. Consequently, the coal blend
injection has become standard practice at CSC’s blast furnaces since 2003.
Du, S. W., Yeh, C. M., Yang, M. K. and Ho, C. K. (2004), Practice of high productivity
at No.3 blast furnace of China Steel Corporation, Proceedings of AISTech Conference,
Tennessee, USA, p. 195-204.
149
ABSTRACT
To meet ongoing rise in the market demand for steel products, China Steel has carried
out series of studies for promoting its productivity since 2000. No3 blast furnace was
exemplified in this paper. Some countermeasures to improve the stability of blast
furnace operation have been practiced as follows: (1) decreasing the slag volume for the
reduction of energy consumption, (2) taking proper charging patterns based on the
burden trajectory measurement to establish terrace in the vicinity of wall, and (3)
adopting one-bit drilling to stabilize the tapping process. As a result, the hot metal
production was gradually increased from 7469 (2000) to 8167 t/d (2002). To promote
the performance of the blast furnace further, a calculation model has been developed to
investigate the coal combustion behaviours in the raceway. The calculation results
indicated that the pressure resistance (pressure loss) in the raceway could be reduced
with the decrease of the volatile content of the PCI coal. With this advantage, the flow
rate of hot blast air may be increased for the increase of the hot metal production. This
was successfully confirmed by the plant trials of coal blend (adding low volatile coal
into the high one) injection in the furnace in 2002. As a result, the PCI operation at CSC
was shifted from high volatile coal injection to coal blend injection in 2003. The coal
blend injection was of key importance in upraising the hot metal production of the
furnace from 8167 t/d in 2002 to 8322 t/d in 2003, even reaching its record high 8536
t/d in September 2003.
Key words: Blast furnace; High productivity; Combustion model; Coal blend injection
150
5.1 Introduction
China Steel Corporation (CSC) is the only integrated steel producer in Taiwan. It
operates four blast furnaces, No1 to 4, having inner volumes of 2434, 2850, 3400 and
3400 m3 respectively. N
o3 blast furnace completed its first campaign on 18
th October
1999 (from November 1987) achieving an average productivity of 1.98 t/d-m3 (inner
volume base), and was again blow-in on 15th
January 2000. Table 5.1 shows the main
features of the furnace. The recovery of steel market has continued worldwide,
especially in Asia, since 2002. Therefore, to promote the productivity has become a
major challenge for No 3 blast furnace in its second campaign.
This paper describes the research works and countermeasures taken for raising the
productivity of No3 blast furnace.
Table 5.1 Main features of CSC’s No3 blast furnace.
5.2 Development of low flux sinter
The energy required for blast furnace process is mainly supplied from sensible heat of
the hot blast air and the combustion heat produced by reactions between the oxygen and
Inner volume, m3
3400
Working volume, m3 2850
Hearth diameter, m 12.5
Number of tapholes 4
Number of tuyeres 32
Top pressure (kPa) 2.3
Blast temperature, ℃ 1150-1180
Charging equipment Bell-less
Cooling system Stave cooler
151
fuels within the raceway. Therefore, the production rate of blast furnace is mainly
determined by blast volume supplied, and can be expressed as the quotient below:
Production Rate (t/h) = (Hot blast supplied through tuyere, m3/h)/(Specific hot blast
required for producing one tonne of hot metal, m3/tHM) (5.1)
Equation 5.1 suggests the production rate can be increased with the decrease of the
energy consumption for generating hot metal. Table 5.2 shows the typical energy
consumption of No3 blast furnace in 1999 (late period of the first campaign). It was
found the energy consumed by slag was significant in the process. Therefore, the
reduction of the slag volume might be one of effective countermeasures for energy
saving and for raising the productivity of the blast furnace. In CSC, the blast furnace
slag is mostly from sinter which accounts for more than 70% of the charged ferrous
burden and is fluxed by adding serpentine, silicon sand and limestone in the sinter plant.
Theoretically, the slag volume can be lowered as the fluxes addition is decreased.
Concerning negative effects on the properties of sinter due to decreasing the fluxes, an
investigation into low fluxes sintering has been carried out (Hsieh et al., 2002). The
experimental results given by sinter pot tests showed that under the conditions of
reducing SiO2 content (from 5.6% to 4.5%), increasing basicity (CaO/SiO2) slightly and
reducing MgO content slightly in sintering, the negative effect was limited and
negligible. According to the results, SiO2 content of sinter was reduced gradually, and
the slag volume was consequently lowered as shown in Table 5.3.
152
Table 5.2 Typical energy consumption in the late period of the first campaign.
Table 5.3 Reduction of SiO2 in sinter and slag volume.
1999 2000 2001 2002 2003
SiO2 content in sinter,% 5.60 5.04 4.84 4.74 4.72
Slag volume,kg/tHM 295 275 265 264 265
5.3 Establishment of burden terrace
The blast furnace has been operated with V shape burden profile since its first
campaign. Owing to unstable movement of the burden resulting in its collapse,
especially in the high productivity operation, the blast furnace has experienced sharply
increase in wall heat loss. To obtain stable movement of the descending burden, a new
charging pattern is required to establish a terrace of the burden profile in the vicinity of
the wall.
×103kcal/tHM %
Heat of Solution Loss 320.5 27.0
Reduction Heat by H2 12.0 1.0
Reduction Heat of Non-Ferrous Metal 12.5 1.1
Decomposition Heat of Moisture 106.8 9.0
Sensible Heat of Top Gas 73.5 6.3
Sensible Heat of Hot Metal 318.5 26.9
Sensible Heat of Slag 156.2 13.2
Heat Loss to Cooling Water 49.3 4.2
Heat Loss 134.3 11.3
Total 1186.6 100
153
Thus an acoustic sensor system (Figure 5.1) was developed to measure the falling points
of the discharged burdens from the rotating chute (Ho, 2000), and the trajectories of the
burdens were also modelled. From the measured falling points and the calculated
trajectories of the burdens, it was found the charged burdens hit the wall heavily and
rolled down towards the centre of the furnace; consequently the V shape burden profile
was formed. Obviously, this was caused by the charging pattern employed in the first
campaign, in which the rotating chute was located at an angle of 49o to the centre of the
furnace in the beginning of charging. A new charging pattern was developed and tested.
The main change of the new pattern was to reduce the angle from 49o to 46
o. As shown
in Figure 5.2, after taking that pattern into practice, the burden profile was shifted from
V shape to M shape, and a long terrace (1.2-1.5m) in the vicinity of wall was
successfully formed. Consequently, the wall heat loss was decreased (Table 5.4) due to
stable peripheral gas flow.
Above Burden Probe
Bar
Chute
AEo
Oscillator
PC
OreCoke
AEi
Pipe
N2
Li Lo
Above Burden Probe
Bar
Chute
AEo
Oscillator
PC
OreCoke
AEi
Pipe
N2
Li Lo
Figure 5.1 AE sensor system for measuring burden falling point.
154
Table 5.4 Variation of wall heat loss.
Wall heat loss, M cal/h
2000 8418
2001 8313
2002 7539
2003 7708
5.4 Development of one bit drilling method
At CSC, soaking bar tapping has been employed since 1984. In the soaking bar
operation, the percussion bar soaked was pulled out by the drilling machine when the
taphole was plugged by mud. However, it was found the oxygen lancing process was
always needed in the final stage of opening. Consequently, the soaking bar tapping
failed to match the gradually increased productivity of the furnace, and fluctuation of
blast pressure occurred due to high liquid level in the hearth. To solve these problems, a
(a 修改前)
前
焦炭層
爐壁
爐中心區域
燒結礦層
Coke TerraceCoke Terrace
(b after)
Coke Layer
Wall
Ferrous Layer
(a before)
(a) (b)
Figure 5.2 Burden profile before (a)/ after (b) changing charging pattern.
155
simplified tapping method called as one bit drilling was developed at CSC for replacing
the soaking bar tapping (Wu and Hsieh, 2003).
For drilling through the taphole in one bit, the diameter of the percussion bar was
increased from 40-45 mm to 42-48 mm, and the pressure of cooling nitrogen was
increased from 6 to 14 atm for better cooling. From the operation results as shown in
Table 5.5, it was found the new drilling practice has become more efficient resulting in
less consumption of the oxygen lance and percussion bar. Additionally, the tapping
frequency has been reduced due to longer life of the taphole.
Table 5.5 Comparison between soaking bar tapping and one bit drilling.
5. 5 Coal blend injection
5.5.1 Analysis of permeability of the furnace
To stably increase the productivity, it is essential to maintain a good permeability of the
furnace. At CSC, the permeability resistance is defined as:
1000B
ΔPk
v
t (5.2)
where
tΔP is the pressure drop between the tuyere exit and the top of the furnace, atm
Bv is the blast volume, Nm3/min
Soaking bar tapping One-bit tapping
Oxygen lance consumed, piece/month 800 207
Percussion bar consumed, piece/ month 498 197
Average number of tapping, -/day 9.3 8.8
156
Equation 5.2 indicates that the permeability can be improved by decreasing the pressure
drop across the furnace. Figure 5.3 shows the pressure distribution of the furnace across
the furnace. It is clear from the Figure, the maximum pressure gradient happens in the
lower zone of the blast furnace, especially in the area of the raceway, where pulverised
coal is injected into and combusted with oxygen. In other words, the pressure drop in
the lower zone of the furnace is influenced by the pulverised coal injection (PCI)
operation, and the pressure gradient in the lower zone might be abated by changing PCI
operation parameters.
At CSC, it was presumed the permeability of the furnace might be affected if the
accumulation rate of unburnt char exceeded its consumption rate in the furnace.
Therefore, high volatile coals have been injected solely into the furnace since it
commenced PCI operation. On the other hand, the advantages of injecting low volatile
coal has been reported below (Willmers, 1989):
(1) Coke replacement ratio increases due to higher calorific value of the low volatile
coal;
(2) The blast momentum decreases with decreasing volatile content of coal injected.
2.2
2.4
2.6
2.8
3
3.2
3.4
3.6
3.8
4
0 5 10 15 20 25 30 35 40 45
Vertical Distance from Tuyere Exit, m
Press
ure, a
tm
Raceway Area
Figure 5.3 Typical pressure distribution of No 3 blast furnace.
157
It suggested the performance of the furnace may be improved by blending low volatile
coal into high volatile coal. Therefore, a further study was needed to compare the
performance of high volatile coal injecting and blend coal injecting.
5.5.2 Coal combustion model within tuyere-raceway area
A calculation model based on a CFD code has been developed to analyse the coal
combustion behaviour within a tuyere-raceway area of blast furnace. The raceway was
assumed to comprise a cylindrical jetting space which had the same diameter (14cm) as
the tuyere exit. Figure 5.4 shows the physical geometry of the tuyere-raceway area. The
burning history of pulverised coal injected can be expressed as the following process
(Takeda, 1994):
(1) rapid heating;
(2) devolatilisation;
(3) oxidization of volatile matters released with hot blast;
(4) combustion of residual unburnt char; and
(5) char gasification.
Figure 5.4 Physical geometry of combustion region.
158
Having examined the aforementioned processes, it was illustrated that the residence
time of coal particle is about 20 ms (Steiler et al., 1996), while the second and third
reactions are implemented within 100 ms. As regards the char combustion, its
characteristic time is in the order of one second. The time required for completing the
last reaction (i.e., the char gasification) is even longer (Smoot and Smith, 1985).
Recognizing these characteristic times, it is known that the devolatilisation reaction
initiates the coal combustion, implying that the selection of parameters in modelling the
devolatilisation reaction is of the utmost importance. To describe the coal
devolatilisation process more realistically, two-competing devolatilisation model(8)
was
employed. The two parallel and competing reactions are given as follows:
Coal 1k
(1-Y1) Char1 + Y1Volatile (high temperature) (5.3)
Coal 2k
(1-Y2) Char2 + Y2Volatile (low temperature) (5.4)
Furthermore, the reaction kinetics can be written as:
Coal )YkY(kdt
dV2211 (5.5)
)/RTEexp( Ak p111 (5.6)
)/RTEexp( Ak p222 (5.7)
In examining the preceding model, it is apparent that the parameters of A1, A2, Y1, Y2,
E1, and E2 have a vital influence in predicting the devolatilisation process. According to
the validation of the model, the parameters suggested by Ubhayakar et al. (1976) were
adopted in the calculation (Table 5.6). In the gas phase, because the average blast
velocity faster than 160 m/s, k-εmodel was thus applied to simulate the turbulent
combustion. In the operation of PCI, following the release of volatile matters from the
coal particles, the oxygen will encompass the volatile, yielding the diffusion flame
combustion. In such a situation, mass fraction probability density function (PDF) model
(Zhou, 1993) is an appropriate method to approach the reaction phenomena. For a
159
system just with two reactants, consisting of fuel and oxidant, the combustion can be
approximated by a single-step reaction as:
Product kg i)(1Oxidant kg i Fuel kg 1 (5.8)
where the coefficient i represents the stoichiometric balance between the fuel and
oxidant. When the turbulent transport coefficients of reactant and oxidant in the flow
field are assumed to be equivalent, by employing the Zeldovich transformation, the
combined mass fraction can be obtained as the following:
iMMX of / (5.9)
where Mf and Mo respond to the mass fractions of fuel and oxidant, respectively.
The mixture fraction f is further defined by
f≡( X-Xo )/( Xf-Xo ) (5.10)
where Xo and Xf express values of X in the fuel side and oxygen side respectively.
Because f is a conservative scalar, its time-average, f , instantaneous conservation
equation in a control volume can be written as:
Sx
f
xfu
xf
t it
t
i
i
i
)()()(
(5.11)
where σt is a computational parameter and it is given as 0.9. On the other hand, in the
framework of PDF, the mean square value of concentration fluctuation, g, can be
calculated through the following equation
160
gk
Cx
gC
x
g
xgu
xg
td
itg
it
t
ii
i
2)()()()( (5.12)
where Cg and Cd are the computational parameters and they are given as 2.8 and 2.0,
respectively. The operation parameters used in the base case calculation is listed in
Table 5.7.
Table 5.6 Parameters of devolatilisation kinetics.
Table 5.7 PCI Operation condition used in the calculation.
Y1, - VM analyzed value
Y2,- 1.5×Y1
A1, 1/s 3.7×105
A2, 1/s 1.46×1015
E1, kJ/mol 74
E2, kJ/mol 251
Hot blast conditions Properties of PC Other operating conditions
Temperature: 1423K
Pressure: 4.5atm
Mass flow rate: 3.9kg/s
O2 content: 21%
FC:55.09%
VM:35.13%
Ash:6.23%
Moisture:3.55%
Particle size
distribution:
90μm:5%
63μm:25%
45μm:55%
20μm:15%
Lance angle: 15o
Lance inner diameter: 20mm
Carrier gas flow rate:0.026 kg/s
Injection rate of PC: 0.4kg/s
Heat loss of tuyere: 900000W/m2
161
5.5.3 Calculation results and discussion
In the base case calculation, the combustion efficiency of the high volatile coal reached
81.5%, and it was 63% when the volatile content of the coal was reduced to 25% (coal
blend case). Figure 5.5 shows the trajectories and residence times of coal particles under
the operation of the base case within the reaction area. It is evident that the mixing of
the coal particles and hot blast is poor, resulting in insufficient oxygen within the coal
plume as shown in Figure 5.6, where the released volatile may turn into soot, which is
not favourable in the blast furnace ironmaking process.
As can be seen in Figure 5.7, the smoother pressure distribution and less pressure loss
along the combustion region area is found with the coal blend operation in comparison
with the base case. This indicates that a decrease in volatile content of coal injected can
effectively abate the pressure loss due to less volatile released to the gas and moderate
gas volume expansion in the combustion area. Although the coal blend injection
generate more unburnt char than injecting high volatile coal, the performance of the
furnace can be improved as long as the consumption rate of the unburnt char exceeds its
accumulation in the furnace. The calculation results encouraged the use of coal blend
injection to improve the performance of the furnaces of CSC.
PC
C
鼓風嘴
Jet Zone
Coal Plum
Jet Zone
Tuyere
Residence Time, ms
Figure 5.5 Trajectories and residence time of coal particles in the combustion
region.
Coal plume
Jet zone
Tuyere
162
5.5.4 Plant trial of coal blend injection
A series of coal blend trials were carried out in the end of 2002, with the low volatile
coal in coal blend increasing gradually from 30% to 50%. Table 5.8 compares the
changes of pressure drop (resistance) in the lower zone of the furnace and permeability
before and after injecting coal blend. The operation results indicate the pressure loss in
the lower zone of the furnace was abated, and permeability of the furnace was improved
Oxygen Concentration, -
Low Oxygen Zone
Oxygen Concentration, -
Oxygen Concentration, -
Low Oxygen Zone
Figure 5.6 Oxygen concentration contour at cross section along combustion
region.
3.4
3.5
3.6
3.7
3.8
3.9
4
4.1
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Horizotal Distance from Tuyere Exit, m
Pres
sure
, a
tm
VM=35%
VM=25%
Raceway
Tuyere Exit
Figure 5.7 Pressure distribution along combustion region from lance exit.
Distance from lance exit, m
163
when the low volatile coal was added. Having verified of the improvement of the
operation, the coal blend injection now becomes standard operation of CSC’s blast
furnaces. With the advantage in lower pressure resistance, the increase of PCI rate or
hot blast flow rate for higher productivity can be achieved.
Table 5.8 Changes of pressure drop and permeability for coal blend injection.
100% HV coal 30% C coal 40% C coal 30% F coal
Pressure drop between
blast and P1, atm 0.68 0.46 0.53 0.58
Permeability, - 0.286 0.267 0.259 0.257
* P1 is located 2.08 meters above tuyere level.
** C and F are low volatile coals with volatile content of 12 and 12.5% respectively.
5.6 Increase of hot metal production in No3 blast furnace
Table 5.9 shows the performance of the furnace on the hot metal production. Generally,
the increase of hot metal production from 2000 to 2002 was mainly contributed by the
countermeasures mentioned in sections of 5.2 to 5.4. On the other hand, the coal blend
injection was of key importance in upraising the productivity from 8167 t/d (2.4 t/m3-d)
in 2002 to 8322 t/d (2.45 t/m3-d) in 2003, even reaching its record high 8536 t/d (2.51
t/m3-d) in September 2003.
Table 5.9 Hot metal Production in CSC’s No 3 blast furnace.
2000 2001 2002 2003
Average blast air flow rate, Nm3/min 5335 5440 5493 5579
Average daily production rate, t/d 7469 7912 8167 8330
Productivity,t/d-m3
2.20 2.33 2.40 2.45
164
5.7 Conclusions
In the period of 2000 to 2003, the operation of CSC’s No3 blast furnace was improved
by means of (1) decreasing the slag volume for the reduction of energy consumption of
the furnace, (2) taking proper charging patterns based on the burden trajectory
measurement to establish terrace in the vicinity of wall, and (3) adopting one-bit drilling
to stabilize the tapping process. As a result, the hot metal production was gradually
increased from 7469 (2000) to 8167 t/d (2002). As revealed in the coal combustion
model in the raceway and plant trials, the pressure resistance in the lower zone of blast
furnace can be abated when low volatile coal is added to the high one in the PCI
operation. With this advantage, the flow rate of hot blast air can be increased for
upraising the hot metal production. In 2003, the PCI operation at CSC’s blast furnaces
was shifted from high volatile injection to coal blend injection. With the advantage of
the coal blend injection, the hot metal productivity was promoted from 8167 (2002) to
8330 t/d (2003), even reaching its record high 8536 t/d (2.51 t/m3-d) in September 2003.
165
CHAPTER 6
BURNING CHARACTERISTICS OF PULVERIZED COAL WITHIN
BLAST FURNACE RACEWAY AT VARIOUS INJECTION
OPERATIONS AND WAYS OF OXYGEN ENRICHMENT
This chapter evaluates the performance of injection lances associated with different
ways of oxygen enrichment in terms of pressure loss and coal burnout within the
regions of blow pipe, tuyere and raceway, which is featured by a voidage contour of 0.4.
Du, S. W., Yeh, C. P., Chen, W. H., Tsai, C. H. and Lucas, J. A. (2015), Burning
characteristics of pulverized coal within blast furnace raceway at various injection
operations and ways of oxygen enrichment, Fuel, vol. 143, p. 98-106
166
ABSTRACT
In this research, coal combustion behavior across the regions of blowpipe, tuyere, and
raceway of blast furnace are numerically examined. Three different lance
configurations, including a single lance, a double air-cooled coaxial lance, and an
oxy-coal lance with different oxygen enrichment patterns, are taken into consideration.
The coal combustion efficiency by the double lance injection is 5.1% higher than that
by single lance injection. From the calculated temperature by the oxy-coal lance, coal
ignition is retarded due to the cooling effect of enriched oxygen flowing through the
lance annulus, resulting in the moderation of pressure loss within the raceway. Most
importantly, the enriched oxygen becomes the combustion enhancer in the downstream
of coal plume after ignition is triggered. Consequently, the coal burnout under the
oxy-coal lance injection is comparative to that under the double air-cooled lance
injection. The performance of blast furnace may be improved with the advantages
provided by the oxy-coal lance injection. Compared with the single lance injection, coal
trajectories under the oxy-coal lance injection are closer to the tuyere exit due to the
higher inertia force of coal particles against hot blast. This should be taken into account
for the designs of the oxy-coal lance.
Keywords: Blast furnace; Pulverized coal injection; Oxy-coal lance injection;
Combustion efficiency; Pressure loss; Numerical simulation.
167
Nomenclature
pA Coal particle surface (m2)
C Coal
C0 Inertial loss coefficient (m-1
)
pC Specific heat of coal particle (J kg-1
K-1
)
Cμ, C1ε, C2ε Empirical constants for turbulence model
Dc Coke diameter in the coke bed (m)
Dp Coal diameter (m)
E Activation energy (kJ mol-1
)
f Mixture fraction
fD Drag force from a particle (N)
F Fuel
Gk Generation of turbulence kinetic energy (kg m-1
s-3
)
H Total enthalpy (J kg-1
)
h Convective heat transfer coefficient (J kg-1
)
k Turbulent kinetic energy (m2 s
-2)
k1, k2 Devolatilization rate constant (s-1
)
M Mass fraction
pm Coal particle weight (kg)
O Oxidant
Qreac Reaction heat (J kg-1
)
q Heat transfer from a particle (W)
Prt Turbulent Prandtl number
R Universal gas constant (kJ mole-1
K-1
)
S Source term (kg m-2
s-2
)
S1, S2 Char
t Time (s-1
)
pT Temperature of coal particle (K)
T Gas temperature (K)
U Mean velocity (m s-1
)
v Mass fraction of solid lost as volatiles
V1, V2 Volatile released at low and high temperatures
168
X Combined mass fraction
xi Spatially coordinate
21 , YY Mass fractions of emitted volatile at low and high temperatures
Greek Symbols
α Porosity
αt Empirical constants for turbulence model
ε Dissipation of turbulent kinetic energy (m2 s
-2)
p Emissivity of coal particle
λ Thermal conductivity (W m-1
K-1
)
μ Viscosity (kg m-1
s-1
)
μeff Effect viscosity of gas (kg m-1
s-1
)
μt Turbulent viscosity (kg m-1
s-1
)
σ Stefan-Boltzmann constant (=5.67×10-8
W m-2
K-4
)
σt Prandtl number of turbulence kinetic energy
ρ Density (kg m-3
)
Subscripts
f Fuel
o Oxidizer
p Coal particle
g Gas phase
169
6.1 Introduction
A number of new ironmaking processes have been developed over the last several
decades; however, blast furnaces are still the most important and commonly employed
facilities for hot metal production due to their superiority in productivity and heat
utilization (Geerds et al., 2011; Chen, et al., 2012; Suopajärvi et al., 2014). In order to
reduce iron ore into iron, metallurgical coke is fed from the top of the blast furnace.
Meanwhile, pulverized coal is injected and burned at the bottom of the furnace to
provide heat for the reduction reactions (Du, et al., 2010). On account of mass
consumption of coal for hot metal production, ironmaking is an energy-intensive
industry and a large amount of CO2 is emitted into the atmosphere (Chen et al., 2011;
Porzio et al., 2013; Hammond and Norman, 2014).
During the operation of blast furnace, blast air heated to temperatures of 1100–125oC is
blown into the furnace through tuyeres, and reacts with coke in raceways to generate
heat and reduction gases for iron ore reduction. To diminish the consumption of
expensive coke, some cheaper auxiliary fuels, such as oil, natural gas, and pulverized
coal, have been used as the substitutes of coke and injected through lances into
raceways. Due to the relatively low price and abundant reserve of coal in comparison
with other fossil fuels, nearly half of blast furnaces in the world (47.7%) use pulverized
coal injection (PCI), while only 4.1% use oil, 11.9% use gas, and 0.2% use plastic
injection (Schott, 2012). For a stable PCI operation, high coal burnout along with a low
pressure loss (high permeability) is always desirable in the regions of blowpipe, tuyere,
and raceway. However, it is difficult to simultaneously implement the two situations
because the enhancement of coal combustion intensity may raise the pressure loss of
blast flow within the raceway.
China Steel Corporation (CSC) is the only integrated steel producer in Taiwan, and has
four blast furnaces located in Kaohsiung and two in Taichung. For the cost reduction of
fuel and its stable supply, the auxiliary fuel injected at CSC has been changed from oil
to pulverized coal since 1988 (Du et al., 2001). In an attempt to increase PCI rate and
170
stabilize blast furnace operation, simulation models by computational fluid dynamics
(CFD) code have been developed at CSC to investigate the flow patterns of injected
coal and gas temperature distributions in combustion zone. Upon inspection of the
predicted trajectories of coal particles in the regions of blowpipe and tuyere (Du et al.,
2007), it was found that the dispersion of injected pulverized coal into hot blast was
poor when single lance injection was operated. This resulted in relatively low
combustion efficiency of coal and soot formation (Chen et al., 2007; Chen et al., 2008).
Besides, from the calculated temperature contours within the regions of blowpipe and
tuyere (Du and Chen, 2006), pulverized coal ignition could be triggered earlier under
the operation of double lance injection when compared to the single one. This
intensified the combustion efficiency of pulverized coal. For this reason, the pulverized
coal injection system in the blast furnaces of CSC was modified from the conventional
single lance to a double air-cooled coaxial lance system in 2001 (Yeh et al., 2002).
Instead of cooling air, the coal combustion temperature could be promoted with oxygen
flowing through the annulus of coaxial lance (Gudenau et al., 1994). This oxy-coal
injection technology has been adopted in many blast furnaces (Chung and Hur, 1997;
Peters et al., 2009; Austin et al., 2011; Hartig et al., 2011). The PCI operation of CSC’s
blast furnaces might be improved if the cooling air for the coaxial lance is replaced by
the enriched oxygen. However, the information regarding the application of oxygen
enrichment in pulverized coal injection remains insufficient. To recognize the
influences of lance configuration and oxygen enrichment pattern on the performance of
PCI operation, a CFD based simulation model has been established in this study. Three
different lance configurations, consisting of a single lance, a CSC’s double lance, and
an oxy-coal lance, are taken into account. The numerical predictions are able to provide
a useful insight into the coal combustion behavior in the regions of blowpipe, tuyere,
and raceway of a blast furnace. In addition, detailed discussion is made to reveal the
impact of enriched oxygen on coal ignition and pressure loss across the combustion
zone.
171
6.2 Methodology
The gas-particle flow and coal combustion in tuyere and raceway were calculated using
FLUENT V12 code. The flow field and temperature distribution were described using
3-D, steady-state Reynolds-averaged Navier–Stokes equations in association with the
RNG (Re-Normalization Group) k–e turbulence model. The pulverized coal particles
were treated as a dispersed phase using a Lagrangian method subject to the assumption
that each particle followed a discrete trajectory without interactions with any of the
other particles. The mathematical formulation is described below.
6.2.1. Gas-particle flow
6.2.1.1. Gas phase
Continuity:
mρU (6.1)
where ρ is the density, U is the mean velocity, and m is the mass transfer rate from
particulate to gas phase.
Momentum:
Deff
T
eff fρUUCUkpUUμρUU2
1
3
20 (6.2)
where μeff, p, and fD are the effective viscosity, pressure, and the drag forces of a
particle, respectively. Based on the Ergun equation (Ergun, 1952), the inertial loss
coefficient is expressed as 3
cαDα13.5C 0 where Dc represents the particle
diameter in coke bed and α is the voidage, which is defined as the total volume of the
voids divided by the total volume of the coke region.
Energy:
qHμ
C
λρUH
t
t
p Pr (63)
where H is the total enthalpy, λ is the thermal conductivity, Cp is the specific heat, μt is
the turbulent viscosity which is modeled by RNG k–ε turbulence model, and q is the
172
heat source from the particle phase. The turbulent Prandtl number, Prt, set in this study
is 0.85.
6.2.1.2 Single particle in dispersed phase
Mass:
mdt
dmp (6.4)
where the subscript p denotes the particulate phase, and m is coal particle mass.
Momentum:
ppDpD
ppUUUUρCπDf
dt
Umd 2
8
1 (6.5)
where Dp is the particle diameter, and CD is the drag coefficient given by Morsi and
Alexander (1972).
Energy:
reac
p
pgpppgp
pp
p,s Qdt
dmTTσεπD TTλNuπDq
dt
TmdC 442
(6)
where Cp,s , εp , σ , and Qreac. stand for the heat capacity of the particle, particle
emissivity, Stefan–Boltzmann constant, and heat released by the surface reaction,
respectively. The Nusselt number, Nu, is computed using the correlation of Ranz and
Marshall (1952a; 1952b).
6.2.1.3 Turbulence model
For the accurate and efficient predictions of the turbulent mixing and dispersion of
injected pulverised coal into hot blast (>160 m s-1
), the RNG k-ε model was adopted to
predict turbulent combustion (Biswas and Eswaran, 2002). This is because that the RNG
k-ε model can provide a better treatment in the mixing and dispersion of coal particles in
comparison with the conventional k-ε model. The complete formulation of the RNG k–ε
turbulence model is given as follows:
ρεGkμρUk kefft (6.7)
ρεCGCk
εεμρUε k1εefft
*2 (6.8)
173
where k is the turbulence kinetic energy, ε is the kinetic energy dissipation rate, and Gk
is the generation of turbulence kinetic energy due to the mean velocity gradients and
expressed by
UUkUUG tT
tk
I
3
2)( (6.9)
2kCteff (6.10)
The coefficient *2εC is given by
3
3
22012.01
)38.4/1(
CCC*
ε (6.11)
where η=Sk/ε, and S is the modulus of the mean rate of strain. The coefficients Cμ, C1ε,
C2ε, and αt are empirical constants, and their values derived empirically are 0.0845,
1.42, 1.68, and 1.393, respectively.
6.2.2 Turbulent combustion
When coal particles are heated, volatile matters will be released to react with oxygen,
resulting in diffusion flame combustion (Khalil, 1982). Owing to high blast temperature
(>1100 °C), it is reasonable to assume that the combustion reactions are fast compared
to fluid mixing rate. Therefore, the mixture fraction probability density function (PDF)
model (Sivathanu and Faeth , 1990) was employed. In the mixture fraction PDF frame,
individual species transport equations were not considered. Instead, the mixture fraction
transport equation was solved. The mixture fraction f in terms of the mass fraction is
written as
iOiF
iOi
XX
XXf
(6.12)
where Xi represents the mass fraction for some element i and the subscripts F and O
stand for the values at the fuel and oxidant sides, respectively. The mixture fraction f is a
conserved scalar and its value in a control volume can be calculated from the solution of
its time-averaged ( f ) instantaneous conservation equation:
p
t
t Sfσ
μfρU
(6.13)
174
In the above equation, t and pS designate dynamics viscosity and source term,
respectively, stemming from coal reactions in the gas phase, respectively. Meanwhile,
t is a computational parameter whose value is given by 0.85 (Jones and Whitelaw,
1982). In the PDF framework, the mean square value of concentration fluctuation g was
calculated through the following equation:
gk
ερCfμCg
σ
μρUg d
2
tg
t
t
(6.14)
where gC and dC are the computational parameters and they are 2.86 and 2.0,
respectively (Jones and Whitelaw, 1982). According to the mixture fraction f, molar
fraction of each gas species, density, and temperature in every control volume were
obtained.
6.2.3 Devolatilization of coal
The burning history of injected pulverized coal is featured by the following sequences:
(1) rapid heating; (2) devolatilization; (3) volatile oxidization; (4) residual char
combustion; and (5) char gasification. In examining the characteristic times of the
aforementioned reactions, the residence time of coal particle is around 20 ms (Steiler et
al., 1996), while the devolatilization reaction is implemented within 10-200 ms (Smoot
and Smith, 1985). With regard to char combustion, its characteristic time is in the order
of 1 s. The time required for completing char gasification is even longer (Smoot and
Smith, 1985). It is known that the devolatilization reaction initiates the coal combustion,
implying that the selection of parameters in modelling the devolatilization reaction is of
the utmost importance. To describe the coal devolatilization process more realistically, a
two-competing devolatilization model (Kobayashi, et al., 1977) was employed. The two
parallel and competing reactions are given as follows:
1111 )1( 1 VYSYC k (low temperature) (6.15)
2222 )1( 2 VYSYC k (high temperature) (6.16)
where C, Y, S, and V denote coal, stoichiometric coefficient, char, and volatile,
respectively. The relative importance of the two equations is mainly determined by
temperature. Specifically, when the temperature is low, the devolatilization reaction is
dominated by Equation 15. Alternatively, it is governed by Equation 16 once the
175
temperature is relatively high. Accordingly, the devolatilization reaction kinetics is
written as:
CYkYkdt
dv 2211 (6.17)
pRTEAk /exp 111 (6.18)
pRTEAk /exp 222 (6.19)
In a previous study (Du and Chen, 2006), it has been found that the kinetic parameters
suggested by Ubhayakar et al. (1977) could accurately predict the devolatilization
processes of coal particles and were thus adopted in this work. The Arrhenius rate
constant and activation energy in Equation 18 are set as 3.7 × 105 s
-1 and 74 kJ mol
-1,
respectively, while they were 1.46 × 1013
s-1
and 251 kJ mol-1
in Equation 19. The
stoichiometric parameter Y1 is taken as the volatile matter measured in proximate
analysis of coal, and Y2 is equal to 1.5Y1 (Du and Chen, 2006). The relationship of Y2
and Y1 fits the experimental data well with coals containing 10 to 40% volatile matter
(Shen et al., 2008). The developed CFD model has been validated by means of the
comparison of gas temperature between simulations and experimental measurements
(Du and Chen, 2006; Du et al., 2007).
6.2.4 Physical geometry and operating conditions
The operating conditions of CSC’ blast furnaces are given in Table 6.1 and used as the
boundary conditions in the numerical predictions. Detailed physical geometries and
sizes of the blowpipe, tuyere, and raceway are sketched in Figure 6.1. The coaxial lance
for double and oxy-coal system used a coal conveying pipe of half 1/2 inch stainless
steel tube (schedule 80), which were contained in a 1 inch tube (schedule 40).
Alternatively, a 3/4 inch straight pipe (schedule 80) was adopted as the single lance. The
placements for the all lance configurations studied in this work were the same at 9o with
respect to the centreline of the blow pipe. The tips of oxy-coal and single lance were
located at the centerline of the tuyere. The size and shape of the raceway were
determined by the porosity contour of 0.4, as shown in Figure 6.1a, and the raceway
was thought of as a porous medium (Du, 2011; Yeh et al., 2012). Because the specific
surface of injected pulverized coal was much higher than that of coke, only coal burning
176
was considered in the simulation. In the oxy-coal lance injection, 40–100% of enriched
oxygen was designed to go through the annulus of the coaxial lance, and the rest,
namely, from 60% to 0%, was added into the blast.
Table 6.1 A list of fuel properties and operating conditions.
Properties of pulverized coal
Proximate analysis (wt%, dry basis) Size distribution (wt%)
FC: 70.66%
VM: 20.02%
Ash: 8.32%
90μm: 5%
63μm: 25%
45μm: 55%
20μm: 15%
Hot blast conditions Other operating conditions
Temperature: 1423 K
Pressure: 4.5 atm
Flow rate of cold blast air: 3.753 kg s-1
Flow rate of enriched oxygen: 0.174 kg s-1
Single and double lance: 100% enriched
oxygen (0.174 kg s-1
) being added to
blast (O2 content: 25%)
Oxy-coal lance: 40-100% enriched oxygen
flowing through the lance, and the rest
proportion (60-0%) being added to
blast
Pulverized coal transportation
(1) Injection rate of pulverized coal:
0.51kg s-1
(2) Carrier gas (air) flow rate: 0.017 kg s-1
(3) Carrier gas and PC temperature: 298 K
Gas for the annulus of coaxial lance
(1) Cooling air flow rate for double lance:
0.0017 kg s-1
(2) Oxygen flow rate for oxy-coal lance:
40 to 100% of enriched oxygen (0.174
kg s-1
)
(3) Gas temperature: 298 K
177
6.3 Results and discussion
6.3.1 Trajectories of coal particles
Figure 6.2 shows the trajectories of coal particles at the three different injection patterns
under the operating conditions given in Table 6.1. Meanwhile, the residence times of
coal particles are presented by different colors. As a whole, the residence time of coal
particles within the combustion region is less than 25 ms, regardless of which injection
pattern is operated. The dispersion of coal particles at the exits of the lances under the
CSC’s double lance injection (Figure 6.2b) is superior to those with single lance
injections (Figure 6.2a and c). Figure 6.2c displays the dispersion characteristics of
larger (e.g., 90 μlm) and smaller (e.g., 20 μm) particles under the operation of oxy-coal
lance injection to reveal the segregation of coal particles in terms of particle sizes. In the
region of lance exit, the larger particles (90 μm) tend to travel along the lance direction,
(a)
(b) (c)
Figure 6.1 Schematics of (a) physical sizes of computational domain as well as the
arrangements of (b) CSC’s double air-cooled lance and (c) single and oxy-coal lance
(α is the porosity within the raceway).
178
resulting from their higher inertia force against the hot blast. Under the influence of the
turbulent fluctuation of the blast, the smaller particles (20 μm) detach themselves from
the centerline of the lance quickly. This suggests the dispersion of injected coal particles
may be improved when the particle size is reduced. In examining the particle
trajectories under the operations of the single lance injection (Figure 6.2a) and oxy-coal
lance injection (Figure 6.2c), it is noted that the coal plume under the oxy-coal lance
injection is closer to the tuyere inner wall. Obviously, the coal particles with the
enriched oxygen flowing through the annulus of the oxy-coal lance could travel longer
along the direction of the injection lance.
(a) single lance
90 µm
20µm
(c) oxy-coal lance
(b) CSC’s double air-cooled lance
Residence time, s
Figure 6.2 Distributions of coal particle trajectory and residence time under (a)
single lance, (b) double air-cooled lance, and (c) oxy-coal lance injections.
179
6.3.2 Oxygen consumption within the combustion region
In a previous study (Du and Chen, 2006), it was found the coal burning was not
sensitive to the oxygen concentration in the blast within the region of blowpipe and
tuyere, in which the dominant mechanism of reactions was coal devolatilization rather
than char oxidation. When the raceway was taken into account, notable improvement in
coal combustion could be accomplished by oxygen enrichment (Shen et al., 2009b). It
follows that the operation of oxygen enrichment may play an important role in
intensifying coal combustion in the raceway. With conventional oxygen enrichment
(stove oxygen enrichment) in the single and the double lance injections, as can be seen
in Figure 6.3a and b, the oxygen in the coal plume is rapidly consumed by the volatiles
liberated from coal particles. Figure 6.3a and b depict that a certain amount of hydrogen
is remained within the coal plume. In other words, a volatile rich region is exhibited
within the coal plume due to insufficient oxygen. The unburnt volatiles in the coal
plume may undergo secondary pyrolysis reactions and be converted into tiny aerosols,
composed of soot particles and tar droplets (Chen et al., 2007; Chen et al., 2008). The
formation of solid carbon aerosols will raise energy loss, as a consequence of
incomplete combustion of fuel. Moreover, these tiny aerosols particles will cause
problems of their collection in the flue gas cleaning (Steeghs, 1992). When the oxy-coal
lance injection is operated, coal particles are surrounded by the enriched oxygen in the
upstream of the coal plume, as shown in Figure 6.3c. A comparison to Figure 6.4a and b,
the volatile rich region with lower levels is pushed downstream (Figure 6.4c). This
reflects that the oxy-coal injection consumes more oxygen for the volatile combustion
in the downstream after the chemical reactions are triggered. It is also implies that the
generation of tiny aerosols can be efficiently abated.
180
Mole fraction of oxygen, -
(a) Single lance
(b) CSC’s double air-cooled lance
(c) Oxy-coal lance
Figure 6.3 Distributions of oxygen mole fraction under (a) single lance, (b) double
air-cooled lance, and (c) oxy-coal lance injections.
181
6.3.3 Ignition and temperature distribution
Figure 6.5 shows the isothermal contours in the regions of blowpipe, tuyere, and
raceway under the three lance injection patterns. Fuel ignition under the single lance
injection occurs in the vicinity of the tuyere exit (Figure 6.5a). With the operation of the
double lance injection, fuel ignition is triggered earlier (Figure 6.5b), stemming from
the better dispersion of coal particles and the increased contact surface between the
particles and hot blast. For the two injection patterns, low temperature zones within the
diffusion flames are observed, and they are consistent with the volatile rich zones within
the coal plumes (Figure 4a and b). The low temperature zones are attributed to the
endothermic reaction of char gasification (C + CO2 = 2CO). The developed diffusion
flame, which partition oxygen and volatiles into two different zones, is a result of the
Mole fraction of hydrogen, -
(a) Single lance
(b) CSC’s double air-cooled lance
(c) Oxy-coal lance
Figure 6.4 Distributions of hydrogen mole fraction under (a) single lance, (b)
double air-cooled lance, and (c) oxy-coal lance injections.
182
reactions between oxygen and volatiles. The flame configuration for the oxy-coal
injection is fairly different from those found in the single and double lance injections.
Figure 6.5c depicts that fuel ignition and diffusion flame obviously shift toward the
downstream of the coal plume in the raceway. This is attributed to the cooling effect of
the enriched oxygen to the combustion. Moreover, the low temperature zone in the
flame under the oxy-coal injection is not as significant as those under the single and the
double air-cooled lance injections where diffusion flames accompanied by
low-temperature cores are exhibited. To farther into the recognition of the burning
characteristics of the oxy-coal lance injection, the gaseous temperature profiles along
the centerline of the tuyere at different proportions of enriched oxygen are plotted in
Figure 6.6. The profile under the single lance injection is also provided for comparison
because the lance is arranged at the same position as the oxy-coal lance. Under the
single lance injection, the rapid increase in gas temperature is located in the vicinity of
tuyere exit where the peak temperature is 2635 K. The temperature sharply drops to
1952 K, as a consequence of insufficient oxygen within the diffusion flame. The peak
temperatures for 40% (60% is added to the blast), 60%, 80%, and 100% of enriched
oxygen flowing through the annulus of coaxial lance are 2785 K, 2781 K, 2804 K, and
2785 K respectively. As expected, the peak temperatures for the oxy-coal injection are
located behind that for the single lance injection due to the cooling effect of the
enriched oxygen. Moreover, for a higher proportion of enriched oxygen, the peak
temperature is located at the deeper location of the raceway. The higher the proportion
of enriched oxygen injected through the annulus of oxy-coal lance, the lower the
magnitude of the gas temperature drop after reaching the peak level. This indicates that
coal combustion can be intensified at a higher oxygen level. Despite the late ignition
resulting from the cooling effect in the beginning of injection, the enriched oxygen
becomes a combustion enhancer to facilitate coal burning in the downstream of the coal
plume as long as the reactions are triggered.
183
6.3.4. Combustion efficiency of coal particles
Temperature, K
(a) Single lance
(b) CSC’s double air-cooled lance
(c) Oxy-coal lance
Figure 6.5 Distributions of isothermal contours under (a) single lance, (b) double
air-cooled lance, and (c) oxy-coal lance injections.
Distance (m)
Ga
ste
mp
era
ture
(K)
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2300
600
900
1200
1500
1800
2100
2400
2700
3000
Lance exit Tyuere exit
Single lance
40%60%80%100%
Figure 6.6 Distributions of gas temperature along the centreline of tuyere under
single lance injection and oxy-coal lance injections at different proportions of
enriched oxygen.
184
6.3.4 Combustion efficiency of coal particles
Figure 6.7 presents the profile of coal combustion efficiency (burnout) under the
oxy-coal lance injection at different proportions (40–100%) of enriched oxygen. The
combustion efficiencies under the operations of the single lance and the double lance
injections with 25% oxygen in the blast are also given in the figure for comparison. The
coal combustion efficiency is defined as the reduction percentage of combustible
portion in coal (Ishii, 2000). It is apparent that the combustion efficiency is improved
when the injection is changed from the conventional single lance injection to the double
or oxy-coal lance injection. Specifically, when the single lance injection is replaced by
the double lance injection, the combustion efficiency rises from 51.3% to 56.4%,
accounting for 5.1% of improvement. This can be explained by the better dispersion
(Figures 6.2 and 6.5) and earlier ignition of coal particles. The combustion efficiency of
the injected coal particles with oxy-coal lance injection goes up with increasing the
proportion of enriched oxygen, and it is in the range of 55.1–56.5%.; it is even higher
than that of the double lance injection when 100% proportion of enriched oxygen is
injected through the annulus.
185
Proportion of enriched oxygen through oxy-coal lance (%)
Co
alco
mb
ustio
ne
ffic
ien
cy
(%)
40 50 60 70 80 90 10051
51.5
52
52.5
53
53.5
54
54.5
55
55.5
56
56.5
57
Single lance
CSC's double air-cooled lance
Oxy-coal lance
6.3.5 Pressure loss
Practically, the pressure loss is an index of permeability resistance in the blast furnace.
To evaluate the performance of the lances in the regions of blowpipe, tuyere, and
raceway, it is essential to analyze the pressure loss due to the generation of gas and heat
from coal combustion. Figure 6.8 shows the profiles of pressure loss across the regions
of blowpipe, tuyere, and raceway at the three injection patterns. In comparison with the
single and double lance injections, relatively low resistance in the regions of blowpipe,
tuyere, and raceway is exhibited when the oxy-coal lance injection is operated. This is
attributed to the occurrence of ignition away from the tuyere exit (Figure 6.5c),
rendering that the effect of gas expansion is not as significant as the other two injections.
Obviously, the oxy-coal lance injection enables to fulfill two contradictory conditions at
the same time in the regions of blowpipe, tuyere and raceway: (1) to retard the coal
Figure 6.7 A comparison of coal combustion efficiency among single lance
injection, double air-cooled lance injection, and oxy-coal lance injections with
different proportions of enriched oxygen.
186
combustion for moderating the pressure loss in the upstream of coal plume; and (2) to
enhance coal combustion and reduce unburnt char generation in the downstream of coal
plume. Taking these advantages from the oxy-coal lance injection, blast furnaces can be
operated with more blast for higher productivity, or with higher PCI rate for lower fuel
cost, thereby achieving the goal of hot metal production with energy saving.
6.4 Conclusions
A three-dimensional CFD model for simulating coal combustion within the regions of
blowpipe, tuyere, and raceway under the operating conditions of CSC’s blast furnace
has been presented in this research. The performances of three different injection
patterns, in terms of coal burnout and pressure loss caused by the combustion, have
Proportion of enriched oxygen through oxy-coal lance (%)
Pre
ssu
relo
ss
(atm
)
40 50 60 70 80 90 1000.25
0.27
0.29
0.31
0.33
0.35
0.37
0.39
0.41
0.43
0.45
Single lance
CSC's double air-cooled lance
Oxy-coal lance
Figure 6.8 A comparison of pressure loss among single lance injection, double
air-cooled lance injection, and oxy-coal lance injections with different
proportions of enriched oxygen.
187
been examined as well. The predicted trajectories of coal particles show the poor
dispersion of injected coal particles. The dispersion can be improved when the coal size
is reduced. Owing to more contact surface between the injected coal particles and the
hot blast under the double air-cooled lance injection, early ignition is triggered; but this
causes higher pressure loss. With oxy-coal lance injection, the coal plume is surrounded
by enriched oxygen. Fuel ignition is delayed, as a consequence of the cooling effect of
the enriched oxygen. However, coal combustion is intensified at the downstream of the
coal plume, resulting from more oxygen consumed for the reaction. This can abate the
generation of tiny aerosols composed of soot particles and tar droplets. Compared to the
single and double air-cooled lance injections, less pressure loss across the combustion
region is exhibited from the oxy-coal lance injection in that the ignition position is away
from the tuyere exit. Increasing the proportion of enriched oxygen under the oxy-coal
lance injection facilitates the coal burnout, and the coal combustion efficiency is slightly
higher than that of the double lance injection when all enriched oxygen is used in the
lance. By virtue of higher combustion efficiency and lower pressure loss under the
oxy-coal lance injection, the performance of blast furnace may be improved. Compared
with the single lance injection, the coal particle trajectories under the oxy-coal lance
injection are closer to the tuyere exit due to the higher inertia force of the particles
against the hot blast. This should be taken into account for the designs of oxy-coal lance
injection. This model has provided useful information for understanding coal
combustion characteristics in the regions of blowpipe, tuyere, and raceway, and this is
conducive to the operation of pulverized coal injection in blast furnaces with
energy-saving.
188
CHAPTER 7
VOLATILE RELEASE AND PARTICLE FORMATION
CHARACTERISTICS OF INJECTED PULVERIZED COAL IN
BLAST FURNACES
This chapter experimentally investigates the volatile release and the generation of char
particle and tiny aerosols in the region of coal plume, where the oxygen is insufficient.
The experiment results provide directions for improving PCI operation.
Chen, W. H., Du, S. W. and Yang, T. H. (2007), Volatile release and particle formation
characteristics of injected pulverised coal in blast furnace, Energy Conversion and
Management, vol. 48, p. 2025-33
189
ABSTRACT
Volatiles release and particle formation for two kinds of pulverized coals (a
high-volatile bituminous coal and a low-volatile bituminous coal) in a drop tube furnace
are investigated to account for the reactions of pulverized coal injection in blast
furnaces. Two different sizes of feed particles are considered; one is 100-200 mesh and
the other 200-325 mesh. By evaluating the R-factor it is found the devolatilization
extent of larger feed particles is relatively poor. However, the swelling behavior of
individual or two agglomerated particles is pronounced which is conducive to the
gasification of chars in blast furnaces. In contrast, for the smaller feed particles,
volatiles liberated from the coal particles can be improved in a significant way as a
result of the amplified R-factor. This enhancement can facilitate the performance of
gas-phase combustion. Nevertheless, the residual char particles are characterized by
agglomeration, implying that the reaction time of the char particles will be elongated,
thereby increasing the possibility of furnace instability. Double-peak distributions in
char particle size are observed in some cases. This is possibly resulting from the
interaction of plastic state and blowing effect at particle surface. Considering the
generation of tiny aerosols composed of soot particles and tar droplets, the results
indicate that their production is highly sensitive to the volatile matter and elemental
oxygen contained in coal. Comparing the reactivity of the soot to that of the unburned
char, the former is always lower than the latter. Consequently, the lower the soot
formation, the better the blast furnace stability.
Keywords: Blast furnace; Pulverized coal; Drop tube furnace; Particle size; Soot.
190
7.1 Introduction
In pulverized coal injection (PCI) operation, the cheaper pulverized coal is injected with
hot air (1100-1200℃) into blast furnaces (BFs) as a substitute for coke. The economic
benefits of the PCI include a reduction in the cost of hot metal, resulting primarily from
decreased coke consumption and an increase in hot metal production. In addition, the
coal is consumed directly without going through the cokmaking plant, the PCI is also
thought to be environmentally friendly because it helps to reduce CO2 emissions (Du et
al., 2001). When coal particles are injected and proceeding from the blowpipe, tuyere,
and then to the raceway, as shown in Figure 7.1, they will experience rapid heating (the
heating rate ranges from 104 to 10
5 K/s), devolatilization, gas-phase combustion, and
char combustion and gasification (Ishii, 2000; Smoot and Smith, 1985). These processes
obviously include homogeneous and heterogeneous reactions. Regarding the particles
dynamics of the pulverized coal, while the coal particles are liberating volatiles they
may undergo swelling (Yu et al., 2003), fragmentation (Hurt and Davis, 1999), and
agglomeration (Shampine, et al., 1995), and eventually evolve into unburned char
particles, which can be consumed by CO2 in the furnace afterwards. Because the typical
residence time for coal particles reaching bird’s nest is around 20 ms (Steiler, et al.,
1996), the reactions between the gas phase and the solid phase are mainly governed by
devolatilization and pyrolysis followed by gas-phase combustion. In contrast, unburned
char combustion and gasification due to the interaction among carbon, oxygen, and
carbon dioxide are relatively unimportant in that the characteristic times of the
heterogeneous combustion and gasification are by far longer than the other reactions
mentioned above (Smoot and Smith, 1985).
In the past, in order to increase pulverized coal injection rate (PCR), a variety of
theoretical and experimental methods have been carried out. For example, Ohno et al.
(1994) investigated PC combustion in a raceway cavity by deriving theoretical
formulas; they also experimentally developed a coke-packed furnace to evaluate the
effect of mixing between the PC and oxygen on the combustion rate. Babich et al.
191
(1996) analyzed the effect of coal grinding on PC burning. Though an increase in coal
grinding level can raise coal combustion intensity, it also increases energy-waste of coal
grind. Hence an optimum coal grinding for PCI into BFs was suggested. The
configuration of injection lance was also studied to achieve higher combustion
efficiency. Chung and Hur (1997) experimentally conducted a coaxial lance with
enriched oxygen going through the annulus to improve coal combustion. They found
that the foregoing design is capable of increasing raceway depth and reducing bird’s
nest thickness in consequence of the enhancement of coal reactions. Ariyama et al.
(1994) experimentally studied pulverized coal combustion in tuyere zone by means of
single-lance and double-lance injections. It was illustrated that the performance of coal
combustion through the double lance injection is better than that by the other one. This
is a result of superior particle dispersion. Similarly, recent numerical predictions of Du
and Chen (2006) suggested that the double lance injection in a blast furnace can
facilitate the ignition of coal particles compared to that of single lance injection. They
also reported that the PCR in a practical BF has been promoted since the double-lance
injection was carried out, revealing that the double-lance injection does possess the
merit of increasing the practice of the PCI.
To pursue cost reduction, high PCR is always the operation target of a BF; however,
once the injection rate is promoted to a certain extent, it will seriously affect the furnace
stability. According to the PC trajectories calculated by Du et al. (2004), the dispersion
of coal plum to hot blast in the raceway is poor, resulting in low oxygen concentration
within coal plum. High PCR practice will not only produce more unburned char, but
also trigger secondary reactions and thereby generate a bit of soot (Fletcher et al.,
1997). The operation might go worse when the accumulation rate of the unburned char
and soot is faster than their consumption, mainly by CO2 inside of the BF. This results
in destroying the permeability of the furnace, excessive coke erosion, as well as
undesirable temperature distribution and cohesive zone shape (Kalkreuth, et al., 2005).
Accordingly, it is recognized that the formation of the unburned char and soot within
192
the coal plume and their reactivity in the BF have a vital effect on the stability of the
furnace operation. For these reasons, the main interests of the present study are on the
characteristics of volatile release from PC as well as related features of unburned char
and soot such as char particle size distribution, surface and internal structures of char,
and reactivity of char and soot. To provide a useful insight into the practice of PCI in
BFs, a drop tube furnace (DTF) will be developed to simulate coal particle reactions in
coal plum region under high-temperature and inert environments. The reactivity of the
unburned char and soot will also be evaluated through thermogravimetric analysis
(TGA). The obtained results enable us to provide a reference for choosing coals in the
performance of BFs.
7.2 Experiment
Two different coals, denoted by coal F and coal L, respectively, serve as the basis of the
present study. These coals are used in the form of PCI for the purpose of getting hot
Bird’s nest
Raceway
Injection
lance
Blowpipe
Tuyere
Coke
Pulverized coal Char particles
Figure 7.1 Schematic diagram of pulverized coal injection and internal structure of
blast furnace around raceway.
193
metal in BFs and their basic properties such as proximate analysis, elemental analysis,
and heating values are listed in Table 7.1. As seen in the table, the volatile matters
(VMs) of the coal F and coal L are 14.67% and 37.26%, respectively. According to the
coal classification of the ASTM, it is known that the former pertains to low-volatile
bituminous coal whereas the latter is classified into high-volatile bituminous coal. When
investigating the characteristics of coal reaction, conventionally, a variety of devices
such as thermogravimetry (TG) (Alonso et al., 2001), heated wire grid (HWG)
(Carpenter and Skorupska, 1995), and drop tube furnace (DTF) (Card and Jones, 1995;
Lee et al., 1996) can be employed. Corresponding to the TG, HWG, and DTF the
heating rates are approximately in the orders of 0.1-1, 103, and 10
4-10
5 K/sec (Anthony,
et al, 1976), respectively. As mentioned in the introduction, in general, the heating rate
for coal particles in raceway is around 104-10
5 K/sec. Consequently, it is proper to use
DTF rather than TG or HWG to simulate coal reactions in the raceways of BFs.
Because of this, a DTF and a standard testing procedure are conducted in the present
work. Alternatively, when the reactivity of char and soot is evaluated, the characteristic
time of heterogeneous reaction between these particles and carbon dioxide in BF is in
the order of second, thermogravimetric analysis is thus performed.
Regarding the developed DTF, as a whole, the entire system can be partitioned into
feeding, reactor, collection, and control subsystems. The feeding unit is constructed by a
310 stainless steel tube which is enveloped by a water jacket and cooled by water to
reduce the possible damage caused by the high temperature furnace. Pulverized coal is
stored in a hopper and fed into the furnace by means of a screw feeder. A hitter with
periodically striking the feed tube is mounted beside the tube to avoid bridging
phenomenon in the tube when charging coal particles. With regard to the reactor, its
diameter and length are 70 mm and 1000 mm, respectively, and the material of the
reactor is composed of crystal Al2O3 which can sustain up to 1600℃. The traveling time
of coal particles within the reactor is around 0.5 second. After the pulverized coal passes
through the reactor, the formed unburned char particles are directly collected from the
194
bottom of the reactor due to inertial impact. The residual particles will travel through a
cyclone in which the medium-size particles are caught by means of centrifugal force.
Thereafter, the tiny aerosols such as soot particles and tar droplets are captured on a
filter. For the heating element, the furnace is heated by a SiC rod which can promote the
reaction temperature up to 1650℃ and a R-type thermocouple is installed in the reactor
to provide a reference temperature of the controller. The reaction temperature is
controlled by a PID (proportional band integral derivative) temperature controller and a
SCR (silicon controlled rectifier) power controller.
As far as the experimental procedure is concerned, when PC is sent into the DTF, pure
nitrogen with volume flow rate of 1 L/min is blown as well to aid carrying the fuel into
the furnace. In addition, preheated gas (N2) with the temperature of 350℃ and volume
flow rate of 1 L/min is also sent into the furnace. The pulverized coal and the preheated
gas are in a co-current pattern. Two different particle size ranges, consisting of 100-200
mesh and 200-325 mesh, are considered to account for coal particles reactions in the
DTF. Moreover, three different reaction temperatures composed of 1000, 1200, and
1400℃ are included in that these temperatures cover the reaction environments of the
raceway in BFs. To ensure the measuring quality, previous to experiments air with fixed
flow rate is blown into the reaction system and then measured at the system exit. This
guarantees that no gas leakage occurs in the system. Following the collection of the
unburned char, particle size distributions are analyzed by laser particle size analyzer
(Coulter LS100) in which the particle sizes ranging from 0.4 to 1000 μm can be
measured. With regard to the reactivity of char particles and soots, they are analyzed in
a TG with the heating rate of 8℃/min. Before that, the soots are washed by acetone and
then dried to separate tar. Carbon dioxide with the volumetric flow rate of 2 L/min is
sent into the TG because the gasification of char and soot in BFs is simulated.
195
Table 7.1 Proximate and ultimate analyses of the investigated coals
7.3 Results and discussion
7.3.1 Devolatilization extent
Volatile release (or devolatilization) extent from PC in a short time can be evaluated by
examining the value of R-factor (Gibbins et al., 1993). The parameter R-factor stands
for volatile release ratio under rapid heating condition to standardized volatile matter
(VM) test from the ASTM. Hence, the higher the R-factor, the better the devolatilization
extent for coal particles travelling in a reaction zone. That is to say, the coal featured
with high R-factor will facilitate volatile liberation from coal particles, thereby
intensifying gas-phase combustion. The R-factor, which is established based on the ash
tracer technique, is defined by
coal) daf (% coal ofcontent VM prox.
VR
where V, which essentially represents volatile yield, is given as
100chardry %char ofcontent ash
coaldry % coal ofcontent ash 1
V
Coal Coal F Coal L
Classification (ASTM) Low-volatile
bituminous
High-volatile
bituminous
Proximate Analysis
(wt %)
Volatile matter 14.67 37.26
Moisture 1.10 1.74
Fixed carbon 76.35 52.51
Ash 7.89 8.50
Ultimate analysis
(wt %, dry-ash-free)
C 91.46 83.19
H 4.15 5.60
N 1.78 1.75
S 0.57 0.53
O (diff.) 2.04 8.82
Heating value (kcal/kg) 7410 7060
196
The R-factor values of the two coals under the three reaction temperatures are displayed
in Figure 7.2. As a whole, the R-factor is obviously affected by both the reaction
temperature and feed particle size. For the case of larger feed particles (i.e., 100-200
mesh), as shown in Figure 7.2a, it is difficult to completely liberate volatiles from the
two coals in consequence of R-factor being smaller than unity. The value is larger than
one only when the reaction temperature is as high as 1400℃. Once the feed particle size
is reduced to 200 to 325 mesh, Figure 7.2b depicts that the R-factor is enlarged greatly,
with the exception of the coal L at 1000oC. It follows that the devolatilization process
can be improved in a significant way if the feed particle size is decreased to a certain
extent. The improvement in the devolatilization extent is particularly remarkable for the
low-volatile bituminous coal (coal F). Accordingly, it should be illustrated that, despite
higher VM contained in the coal L, its R-factor is still lower than that of the coal F. This
implies that the devolatilization extent is not determined by the VM in nature.
7.3.2 Particle formation of the low-volatile bituminous coal
Subsequently, particle size distributions of the coal F before and after reactions are
demonstrated in Figure 7.3. For the coal F with larger feed particles, once the coal
Temperature (0C)
R-f
acto
r
900 1000 1100 1200 1300 1400 15000
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Coal F
Coal L
(a) Mesh 100-200
Temperature (0C)
R-f
acto
r
900 1000 1100 1200 1300 1400 15000
0.25
0.5
0.75
1
1.25
1.5
1.75
2
(b) Mesh 200-325
Figure 7.2 Profiles of R-factor of two different coals at various reaction
temperatures.
197
Particle diameter (m)
Vo
lum
e(%
)
10-1
100
101
102
1030
1
2
3
4
5
6
7
8
9
10
11
12
Raw coal
1000oC
1200oC
1400oC
(a)
Partialfragmentation
Swelling
Particle diameter (m)
Vo
lum
e(%
)
10-1
100
101
102
1030
1
2
3
4
5
6
7
8
9
10
11
12
Raw coal
1000oC
1200oC
1400oC
(b)
Swelling
Agglomeration
experiences reactions, it is evident that the size distributions of unburned char are not as
sharp as that of the feed coal no matter what the reaction temperature is. The curves also
depict that part of the char particles become smaller whereas some particles tend to
extend. It is thus recognized that the coal particles are simultaneously characterized by
fragmentation and swelling while they are evolving into unburned char. When the peaks
of the curves are examined in detail, prior to reaction the maximum value of the curve
occurs at d=130.7μm (the solid line shown in Figure 7.4), where d designates particle
diameter. In addition, corresponding to the three curves of the reaction temperatures
1000, 1200, and 1400oC, Figure 7.4 depicts that the maximum values or peaks take
place at d=130.7, 117.4, and 105.5μm, respectively. The maximum value decreases with
increasing reaction temperature, therefore it is inferred that the peaks are dominated by
little debris separated from their parent particles, as a consequence of thermal impact
upon the particle surface stemming from rapid heating in the DTF. To provide a more
clear observation on the unburned char formation, the scanning electron microscope
(SEM) images of the char particles are demonstrated in Figure 7.5. It is apparent that, in
the cases of larger feed particles (Figures 7.5a-5c), most char particles are characterized
by individual particles and accompanied by a few debris. The figures also indicate that
increasing reaction temperature makes the char particles rounder.
Figure 7.3 Particle size distributions of coal F before and after experiencing reactions
with (a) larger feed particles and (b) smaller feed particles.
198
Temperature (oC)
Pe
ak
loca
tio
n(
m)
800 1000 1200 1400 160050
100
150
200
250
300
Larger feed particlesSmaller feed particles
Second peak
First peakPeak location of feed coal
Temperature (oC)
Pe
ak
loca
tio
n(
m)
800 1000 1200 1400 160050
100
150
200
250
300
Larger feed particlesSmaller feed particles
Second peak
First peak
Peak location of feed coal
Figure 7.4 Peak locations of particle size distributions for coal F before and after
experiencing reactions.
Figure 7.5 SEM images of unburned chars of coal F at larger feed particles (a-c) and
smaller feed particles (d-f).
(a) 1000℃ (b) 1200℃ (c) 1400℃
(d) 1000℃ (e) 1200℃ (f) 1400℃
199
When the coal F with smaller feed particles is considered, unlike the preceding
behaviors, the three curves shown in Figure 7.3b shift rightward to a great extent; the
case of 1400oC even exhibits a double-peak distribution. In examining the maximum
values of the curves, it occurs at d=76.46 μm for the feed coal (the dashed line shown in
Figure 4) whereas they develop at 85.11 and 94.74 μm for the unburned chars under the
situations of 1000 and 1200oC, respectively. Physically, on account of smaller feed
particles, heat transferred from the gas phase to the particle phase is more effective,
rendering that more volatiles are released, as elucidated in Figure 2. This results in the
characteristic of particle fragmentation being intensified. At the same time, the plastic
state on the particles surfaces and the collisions among the particles make them
agglomerate each other. As a result, as presented in Figures 5d-5f, the generated char
particles are featured by agglomeration instead of the individual particles (Figures
5a-5c). In regard to the case of 1400oC, the first and the second peaks appear at 76.46
and 248.6μm, respectively. It is known that the Stefan flow (Chen and Jiang, 2000)
stemming from particle surface will be elicited as long as volatiles are liberated. If the
blowing effect of the Stefan flow is high to a certain extent, the tendency of particles
gather will be suppressed somewhere. This might be the mechanism which causes the
phenomenon of the double-peak distribution. In a word, there are two major forces at
play while coal particles are evolving into chars. The first one is the plastic state which
will trigger agglomeration; the second one is the blowing effect which will prohibit char
particles to combine together.
7.3.3 Particle formation of the high-volatile bituminous coal
When attention is focused on the high-volatile bituminous coal (viz., the coal L) with
larger feed particles, a comparison with Figure 7.3a reveals that the curves shown in
Figure 7.6a move rightward to a certain extent after undergoing reactions. This is
because the particles swelling and agglomeration are more significant, resulting from
200
higher VM owned by the coal L. The peak of the feed coal is located at d=105.7 μm (the
solid line shown in Figure 7.7). Once the coal experiences devolatilization,
corresponding to the reaction temperatures of 1000, 1200, and 1400oC the peaks move
to 200.6, 180.3, and 161.9 μm, respectively. In light of these char particles being larger
than the feed particles approximately by a factor of 2, it is realized that the chars
formations near the peaks are no longer dominated by individual particles but mainly
characterized by two particles agglomeration. The aforementioned mechanism can be
verified by observing the SEM images of the char particles (Figures 7.8a to 8c) where
the majority of char particles exhibit low level accumulation.
Particle diameter (m)
Vo
lum
e(%
)
10-1
100
101
102
1030
1
2
3
4
5
6
7
8
9
10
11
12
Raw coal
1000oC
1200oC
1400oC
(a)
Swelling
Agglomeration
Particle diameter (m)
Vo
lum
e(%
)
10-1
100
101
102
1030
1
2
3
4
5
6
7
8
9
10
11
12
Raw coal
1000oC
1200oC
1400oC
(b)
Agglomeration
Figure 7.6 Particle size distributions of coal L before and after reactions with (a)
larger feed particles and (b) smaller feed particles.
201
Temperature (oC)
Pe
ak
loca
tion
(m
)
800 1000 1200 1400 160050
100
150
200
250
300
Larger feed coalSmaller feed coal
Second peak
First peak
Peak location of feed coal
Temperature (oC)
Pe
ak
loca
tion
(m
)
800 1000 1200 1400 160050
100
150
200
250
300
Larger feed coalSmaller feed coal
Second peak
First peakPeak location of feed coal
Figure 7.7 Peak locations of particle size distributions for coal L before and after
experiencing reactions.
Figure 7.8 SEM images of unburned chars of coal L at larger feed particles (a-c) and
smaller feed particles (d-f).
202
When the situations of the smaller feed particles are taken into account, the curves
shown in Figure 7.6b indicate that the double-peak distributions develop in the cases of
1000 and 1200oC. From the profiles shown in Figure 7.7, the values of the second peak
are larger than that of the first peak approximately by a factor of four. It is inferred that
the second peaks yielded are mainly contributed by over two char particles together.
The SEM images shown in Figures 8d and 8e clearly suggest the situations of particles
accumulation. With regard to the result of 1400oC, both the Stefan flow and particles
fragmentation are further enhanced. In such a situation, the DTF tends to behave as a
mixed flow reactor and the char agglomeration is in a more uniform environment. As a
result, the characteristic of the double-peak distribution disappears and the observation
of particles accumulation is provided in Figure 7.8f.
To proceed farther into the recognition of particles formation under the condition of
larger feed particles, the SEM cross-sectional images of the feed coal F and coal L as
well as their char particles with the reaction temperatures of 1000 and 1400oC are
demonstrated in Figure 7.9 Basically, the internal structures of the feed particles of the
coal F and coal L are compact (Figures 7.9a and 9b). Following the reaction with the
temperature of 1000oC, the interior of the char of the coal F becomes porous
accompanied by some larger holes (Figure 7.9c). In view of the higher VM of the coal
L, volatiles emission from the coal is more pronounced so that the holes inside the char
particles are larger (Figure 7.9d). With promoting the reaction temperature to 1400℃,
the char interiors of both the coal F and coal L are further emptied (Figures 7.9e and 9f).
It is not surprising because more volatiles are transported into the gas phase compared
to the results of 1000oC. In the study of Wall et al. (2002), char internal structures were
categorized into three groups in accordance with geometric parameters and porosity.
Upon inspection of the SEM cross-sectional images shown in Figure 7.9, the three
groups are also obtained. Specifically, Figure 7.9c pertains to group III where the char
porosity is lower and the wall is very thick. Alternatively, the internal structures with
203
group I and group II can be observed in some char particles shown in Figures 9d to 9f
since the porosity is higher and the average wall thickness is thinner.
(a) Feed coal F (b) Feed coal L
(c) Char particles from coal F at 1000oC (d) Char particles from coal L at 1000
oC
(e) Char particles from coal F at 1400 oC (f) Char particles from coal L at 1400
oC
Group III
Group II
Group I
Group I
Group II
Group II
Figure 7.9 SEM images of feed coal and unburned char particles shown in
cross-section.
204
7.3.4 Aerosol formation and reactivity
It is known that soot and tar particles will be produced due to volatiles secondary
reactions if the fuel undergoes incomplete combustion. By virtue of higher crystal
structure of carbon in soot which causes lower reactivity, the accumulated soot in blast
furnaces will reduce gas permeability. The weight percentages of soot and tar from the
two coals reactions at the three reaction temperatures are plotted in Figure 7.10 where
the feed particles of 100-200 mesh are tested. For the coal F (or the low-volatile
bituminous coal), Figure 7.10a suggests that the generation of soot and tar increases
with increasing reaction temperature where the weight percentage ranges from 0.13 to
0.25. This is, of course, attributed to more volatiles release which results in more soot
and tar production. Alternatively, for the high-volatile bituminous coal, namely, the coal
L, one can find that the weight percentage is in the range from 1.8 and 2.2. These values
are by far higher than the coal F suggesting that soot and tar formation is highly
sensitive to the VM content. Contrary to the behavior shown in Figure 7.10a, it is of
interest that the weight percentage from the coal L decreases with temperature. It is
inferred that the preceding feature is a consequence of high elemental oxygen contained
in the coal, as shown in Table 7.1. At higher temperature, the reaction between volatiles
and oxygen is more complete. Hence, increasing temperature reducing the formation of
the aerosol particles, as observed.
7.3.5 Reactivity of char and soot
Under the condition of 1400℃ and 100-200 mesh feed particles, the thermogravimetric
analyses of the unburned char and soot from the coal F and coal L reacting with carbon
dioxide are shown in Figure 7.11. It is observed that the decaying trends of both the
unburned char and soot from the coal F are slower than that from the coal L. This
elucidates that the reactivity of the char and soot from the low-volatile bituminous coal
is worse than that from the high-volatile bituminous coal. Furthermore, from the
viewpoint of furnace stability, the performance instability of furnaces may be excited if
205
Temperature (0C)
We
igh
tp
erc
en
tag
e(%
)
We
igh
tp
erc
en
tag
e(%
)900 1000 1100 1200 1300 1400 15000
0.1
0.2
0.3
0.4
0.5
0.6
1.5
1.6
1.7
1.8
1.9
2
2.1
2.2
2.3
2.4
2.5
Coal F
Coal L
the low-volatile bituminous coal is used. In addition, for either the coal F or coal L, the
reactivity of soot is slower than the unburned char. It follows that the lower the soot
production ratio, the higher the gas permeability in blast furnaces. In light of the
obtained results, the reaction physics of the two coals are summarized in Table 7.2.
Temperature (0C)
We
igh
tp
erc
en
tag
e(%
)
600 700 800 900 1000 11000
20
40
60
80
100
Coal F unburned charCoal L unburned charCoal F sootCoal L soot
Figure 7.11 Thermogravimetric analyses of the produced unburned chars and soots
at 1400oC.
Figure 7.10 Profits of soot and tar formations with respect to reaction
temperature.
206
T
ab
le 7
.2
Sum
mary
of
reacti
on p
hysi
cs
of
the t
wo c
oals
207
7.4 Conclusions
Experimental studies on the characteristics of volatiles release, particles formation, and
reactivity of unburned char and soot of two different coals have been achieved in the
present study. The purpose was to recognize the fundamental physics of pulverized coal
reactions in blast furnaces. An examination of R-factor reveals that either decrease feed
coal particle size or increase reaction temperature can efficiently enhance volatile
liberation. As a whole, the value of the R-factor of the larger feed particles is relatively
small; however, the swelling behavior of the char particles is conducive to the
subsequent char combustion and gasification reactions in blast furnaces because of the
surface enlargement of individual or two agglomerated particles. On the contrary, for
the finer feed particles the devolatilization extent is more complete. This can facilitate
the gas-phase combustion. However, the char formation is dominated by over two
particles agglomeration which will disadvantage the furnace stability. The double-peak
distributions in particle size are exhibited in some cases possibly due to the interaction
between softening stage and blowing effect of the Stefan flow. Furthermore, in view of
volatiles release from particles interiors, three different internal structures in the char
particles are observed and they depend strongly on the VM and reaction temperature.
Considering the generation of tiny aerosols which consist of soot particles and tar
droplets, it is mainly determined by the content of VM and elemental oxygen. Finally,
from the examinations of the unburned char and soot in TG, we find that the reactivity
of the unburned char from the low-volatile bituminous coal is lower than that from the
high-volatile bituminous coal. Moreover, the reactivity of the soot is lower than the
char. The obtained results have provided a very useful insight into the choice of injected
pulverized coal from the viewpoints of gas-phase combustion as well as char and soot
gasification in blast furnaces.
208
CHAPTER 8
PULVERIZED COAL BURNOUT IN BLAST FURNACE
SIMULATED BY A DROP TUBE FURNACE
This chapter describes the method developed at China Steel Corporation to evaluate the
combustion efficiency of PCI coals using a drop tube furnace. The method is currently
employed at CSC for the selection of PCI coals.
Du, S. W., Chen, W. H. and Lucas, J. A. (2010), Pulverised coal burnout in blast
furnace simulated by a drop tube furnace, Energy, vol. 35, p. 576-581.
209
ABSTRACT
Reactions of pulverized coal injection (PCI) in a blast furnace were simulated using a
drop tube furnace (DTF) to investigate the burnout behavior of a number of coals and
coal blends. For the coals with the fuel ratio ranging from 1.36 to 6.22, the experimental
results indicated that the burnout increased with decreasing the fuel ratio, except for
certain coals departing from the general trend. One of the coals with the fuel ratio of
6.22 has shown its merit in combustion, implying that the blending ratio of the coal in
PCI operation can be raised for a higher coke replacement ratio. The experiments also
suggested that increasing blast temperature was an efficient countermeasure for
promoting the combustibility of the injected coals. Higher fuel burnout could be
achieved when the particle size of coal was reduced from 60-100 to 100-200 mesh.
However, once the size of the tested coals was in the range of 200 and 325 mesh, the
burnout could not be improved further, resulting from the agglomeration of fine
particles. Considering coal blend reactions, the blending ratio of coals in PCI may be
adjusted by the individual coal burnout rather than by the fuel ratio.
Keywords: Combustion; Burnout; Pulverized coal injection (PCI); Drop tube
furnace (DTF); Blast furnace; Blend; fuel ratio.
210
8.1 Introduction
Blast furnaces are a crucial and the most commonly employed facility in ironmaking
processes. The interior of a blast furnace is filled with alternating layers of coke and ore
burden charged from the top of the furnace to initiate the production of hot metal (Ishii,
2000). The fed coke descents to the hearth of the furnace and constructs a porous coke
bed named deadman. The formed hot metal and slag inside the blast furnace can
penetrate through the deadman and flow downward to the bottom of the blast furnace.
Meanwhile, blast air heated to the temperature of 1100-1250°C is blown into the furnace
through the tuyeres. As a result, a cavity called raceway is formed around the exit of a
tuyere. In the raceway, the heated air takes part in the combustion of deadman coke to
generate heat and reducing gases. The heat and reducing gases are used for the
production of hot metal. To reduce the consumption of coke, some cheaper auxiliary
fuels, such as oil, natural gas, pulverized coal (Perlov, 1987) and even plastic wastes
(Ziebik and Stanek, 2001), have been used as the substitute of coke and injected through
the injection lance into the raceway. It should be emphasized that the technique of
auxiliary fuel injection can not only promote the productivity of a blast furnace but also
increase the flexibility of practical operation. For example, the oxygen concentration in
hot blast can be enriched (Jianwei, et al., 2003), thereby intensifying the combustion of
the auxiliary fuel. Since the second energy crisis occurred in 1979, the technique of
pulverized coal injection (PCI) has become a vital method in ironmaking process due to
the relatively lower price and abundant reserve of coal compared to other fossil fuels
(Toyoda, 1983).
The operation of PCI was introduced into the blast furnaces of China Steel Corporation
(CSC) in 1987 (Lai et al., 1994). Since that, a high PCI rate is always a desirable
operation target of CSC to reduce the cost of fuel. To make the injected coals ignited
earlier in the tuyere for higher combustibility, the coals with low fuel ratio, which
generally had high ignitibility (Kurose, et al., 2004), were solely injected into the blast
furnaces since CSC commenced the PCI operation. The fuel ratio is defined as the
211
weight ratio of fixed carbon to volatile matter in a raw coal (Kurose, et al, 2004).
Practically, it is known that the operational cost for producing hot metal can be reduced
if the consumed fuel rate (coke rate plus PCI rate) based on per metric ton of hot metal
is decreased. To approach this goal, some coals with high fuel ratio, which have higher
calorific values, have been blended into low ones at CSC since 2002 (Du et al., 2004).
On the other hand, the past research has suggested that soot formation from coal
reaction with low fuel ratio will exceed that with high fuel ratio, and the reactivity of
soot is lower than that of unburned char (Smoot and Pratt, 1979; Chen et al, 2008).
Consequently, another advantage for coals with high fuel ratio injected into the blast
furnaces is that the formation of soot can be abated (Chen et al., 2007). For these
reasons, increasing the blending ratio of a coal with high fuel ratio in PCI coals has
become an important target in the operation of CSC’s blast furnaces. When coal
particles are injected into the blowpipe-tuyere-raceway area, by virtue of very short
residence time (Steiler et al., 1996) and poor dispersion of the particles into the blast
(Du and Chen, 2006; Du, et al., 2007), the generation of unburned char from the
operation of high PCI rate is unavoidable. This is especially significant when a coal
with high fuel ratio is injected. Although it was reported that the unburned char trapped
in the furnace can be consumed by CO2 and H2O and by the direct reduction of FeO
(Iwanaga, 1991), the permeability of gas and liquid in the furnace may be adversely
affected if the accumulation rate of unburned char exceeds its consumption rate.
Therefore, if one intends to achieve a smooth operation of blast furnace, it is essential to
evaluate the combustion efficiency of PCI coals, especially for the coals with high fuel
ratio.
To figure out the behavior of coal reaction in blast furnaces, many experiments of coal
combustion have been carried out using combustion rigs to simulate the situations
around blowpipes. The combustion rig was attached to an empty combustion chamber
(Mathieson, et al., 2005; Ueno et al., 1993) or to a coke bed in some cases (Yamagata,
et al., 1992; Ariyama et al., 1994). However, in a CSC’s internal research for the
212
combustion experiments using a combustion rig connected with an empty combustion
chamber (Du et al., 2001), many operation difficulties have been encountered; they
included pulverized coal preparation, char sampling, temperature measurement and
keeping a constant coal injection rate. As a matter of fact, it has also been found that the
experiments were a time, manpower and cost consuming process. Over the years a
number of studies concerning coal combustion using drop tube furnaces have been
reported. Despite the inherent difference between the realistic combustion environment
of the raceway and the drop tube furnace (DTF), the DTF is still an effective device
when one attempts to evaluate the combustion performance of PCI coals in an
environment with high heating rate (>104
K/s). In the present study, a DTF combustion
system has been developed and the effects of some parameters, composed of coal type
(or fuel ratio), reaction temperature, particle size and blending ratio of binary coal, on
the combustion efficiency have been taken into consideration. The obtained results are
able to provide useful information to the operation of PCI in blast furnaces.
8.2 Experiments
8.2.1 Reaction system
The experiments were performed by dropping pulverized coal particles into a drop tube
furnace reaction system, as shown in Figure 8.1. The DTF system consisted of a feeding
subsystem, a reactor and a particle collection subsystem. The feeding subsystem was
composed of a hopper, a screw feeder, a lance and an electric hammer. The hammer was
arranged to periodically knock the lance during the experiments to aid in dispersing coal
particles into the carrier gas. The reactor was heated by a SiC heating element
controlled by a PID (proportional band integral derivative) temperature controller and a
SCR (silicon controlled rectifier) power controller. An R-type thermocouple was
mounted in the reactor to detect the temperature in the reaction zone and the detected
temperature was used as the reference of the SCR. Two ends of the reactor were sealed
by high temperature o-rings. The char particles were collected by a cyclone through a
213
sampling probe. The sampling probe was enveloped by a water jacket and the water
jacket was filled with cooling water so as to protect the probe. A suction motor was
connected to the exit of the cyclone to keep the coal particles moving straightforward
into the probe. The position of the probe was adjustable for controlling different
residence times, and a cooling nitrogen stream was arranged at the entrance of the probe
to prevent further reaction of the collected char particles when they entered the probe.
1. Carrier gas 2. Secondary gas 3. Rotameter
4. Hopper 5. Preheater 6. Thermocouple
7. DTF 8. Heater 9. Sampling probe
10. Cooling water 11. Container 12. Cyclone
13. Unburned char 14. Pump 15. Exhausted gas
14
15
1
3
9
11
5
8
6 2
7
13
10
4
12
Figure 8.1 Schematic of the reaction system.
1. Carrier gas 2. Secondary gas 3. Rotameter
4. Hopper 5. Preheater 6. Thermocouple
7. DTF 8. Heater 9. Sampling probe
10. Cooling water 11. Container 12. Cyclone
13. Unburned char 14. Pump 15. Exhausted gas
14
15
1
3
9
11
5
8
6 2
7
13
10
4
12
Figure 8.1 Schematic of the reaction system.
214
8.2.2 Experimental procedure and conditions
The tested coals were dried in a nitrogen-purged oven at 100-105°C for 24 hours and
then pulverized and sieved to different size fractions for experiments. During the
experiments, the tested coal particles stored in the hopper were fed by a screw feeder.
The coal particles and carrier gas (2L min-1
) were introduced from the top of the reactor
and they travelled along the centerline of a ceramic tube with the internal diameter of
70mm. To lessen the significant agglomeration of the coal particles, the feeding rate of
the coal particles was controlled to be as lower as 5 g h-1
. Additional secondary gas (3L
min-1
) was preheated to 350°C followed by blowing into four metal tubes around the
lance. The compositions of the carrier gas and the secondary gas were the same as the
atmospheric air. The base experimental conditions used in this study included the coal
particle size of 100-200 mesh and the reaction temperature of 1200°C. The comparison
of the combustion efficiency among the tested coals was made in terms of the base
conditions. To investigate the effects of reaction temperature and coal particle size on
the coal combustion efficiency, the burning experiments were carried out for the
reaction temperature ranging from 1100 to 1400°C as well as the coal particle size in the
ranges of 60-100, 100-200 and 200-325 meshes. Combustion experiments of binary coal
blends were also performed in this study. As described in Introduction, to increase the
calorific values of coals used for PCI operation, a coal with low fuel ratio can be
blended with a coal with high fuel ratio. This blended fuel can reduce the fuel rate used
in PCI to a certain extent compared to the utilization of the coal with low fuel ratio
alone. Because of this, two coals with high fuel ratio (>5) were individually blended
with a coal with low fuel ratio (<2). The blending ratios for the coals with high fuel
ratio in the coal blends covered 25, 50 and 75 wt%. To ensure the measurement stability
and accuracy, the reaction system was leak tested before the experiments were
performed. Moreover, the R-type thermocouple was calibrated annually by a CSC’s
standard furnace to ensure the accuracy of the measured temperature in this study. In
comparison with the readings given by a standard wire, the errors of the thermocouple
215
were found to be within +/- 3°C in the temperatures ranged from 1000 to 1400°C. The
burnout experiments were carried out at least twice for some tested coals to guarantee
that the results could be reproduced. To provide an analysis of the reliability of the
experiments from the coals covering a wide range of fuel ratio, four different coals,
Coals A, B, D and K, with the fuel ratio in the range of 1.36-6.33 were thus tested. In
other words, the analyzed fuels included the coals with the lowest and the highest fuel
ratios. Figure 8.2 displays the values of burnout of the four coals. As seen in the figure,
the relative differences of the burnout tests were controlled below 5%, revealing that the
quality of the experiments was reliable.
Fuel ratio (-)
Bu
rno
ut
(%)
1 2 3 4 5 6 740
50
60
70
80
90
100
Test 1
Test 2
Coal D
Coal B
Coal A
Coal K
Figure 8.2 Tests of experimental stability of four different coals under the base
experimental conditions.
216
8.3 Results and discussion
In the following discussion, eleven coals (Coal A to Coal K), covering a wide range of
fuel ratio from 1.36 to 6.22, were tested. The proximate analyses, fuel ratios and higher
heating values of the tested coals are summarized in Table 8.1. Meanwhile, this study
also considered the effects of reaction temperature, particle size and blending ratio of
two different coals on the burnout behavior. Some discussion from the viewpoint of
practical operation of PCI based on the obtained results will be addressed.
Table 8.1 Proximate analyses (dry basis), fuel ratios and higher heating values (HHV,
dry basis) of the investigated coals.
8.3.1 Combustion efficiency of individual coals
The combustion efficiency of a coal is often measured in terms of burnout. In this study,
the ash tracer method (Tate, 1993; Osório, et al., 2006] was adopted to determine the
burnout of the tested coals. The burnout can be expressed as
100
1001
(%)
rcuc
rcuc
AshAsh
AshAshBurnout (8.1)
Coal VM % (db) FC % (db) Ash % (db) Fuel ratio
(FC/VM)
HHV
kcal/kg (db)
Coal A 12.65 78.74 8.61 6.22 7,970
Coal B 12.69 78.93 8.38 6.22 7,890
Coal C 13.70 76.85 9.45 5.61 7,861
Coal D 15.74 74.61 9.65 4.74 7,716
Coal E 18.10 71.78 10.12 3.87 7,580
Coal F 23.71 67.39 8.9 2.84 7,638
Coal G 25.88 64.16 9.96 2.48 7,290
Coal H 32.82 58.46 8.72 2.37 7,180
Coal I 34.87 55.90 9.23 1.55 7,385
Coal J 35.79 55.51 8.7 1.54 7,531
Coal K 39.58 53.96 6.46 1.36 7,522
217
where ucAsh and rcAsh stand for ash contents in the unburned char and the raw coal,
respectively. Generally, the ash content analyses of the coal and unburned char were
carried out two times in this study, and the average value was taken if the difference
between two analyses was lower than 0.15%. Physically, by virtue of high reactivity of
volatile matter compared to that of fixed carbon, a coal with low fuel ratio represents
that the reactivity of the coal is high. Alternatively, the combustion efficiency or
burnout of a coal with high fuel ratio is usually lower than that with high fuel ratio. In
the present study, the burnout of the coals with the fuel ratio varying from 1.36 to 6.22
was tested individually under the base experimental conditions. A linear correlation
(R2=0.91) between the burnout and the fuel ratio was apparently exhibited, as shown in
Figure 8.3. This reflects that the burnout was increased with the decrease of the fuel
ratio. However, two of the coals with high fuel ratio, namely, Coal A and Coal C,
deviated from this rule. The fuel ratio (FR) of Coal A (FR=6.22) is much higher than
that of Coal D (FR=4.72), however, the burnout of the former (59%) was fairly close to
that of the latter (61.9%). Alternatively, the fuel ratio of Coal C is close to that of Coal
A, the burnout of the former (45.1%) was by far lower than that of the latter (59%). This
may be attributed to the effects of maceral and mineral contained in the coals rather than
the fuel ratio of the coals (Carpenter and Skorupska, 1993). For example, the maceral
analysis of Coal A, as shown in Table 8.2, indicated that the vitrinite, inertinite and
liptinite were 57.2%, 38.8% and 0.4%, respectively, whereas in Coal B they were
42.3%, 57.7% and 0%, respectively. Because the reactivity of vitrinite was larger than
the other two components and the vitrinite percentage in Coal C was especially low
compared to other coals, this resulted in the low burnout of Coal C, as observed. Seeing
that the combustion of some coals with exceptional performance was found from the
experiments, the evaluation of coal combustion from the DTF can provide useful
information on coal selection, especially for the coals with high fuel ratio.
218
Table 8.2 Maceral analyses of the investigated coals.
Coal Vitrinite, % Inertinite, % Liptinite, %
Coal A 57.2 38.8 4
Coal B 68.0 32.0 0
Coal C 42.3 57.7 0
Coal D 50.6 49.4 0
Coal E 49.0 51.0 0
Coal F 53.4 33.8 12.8
Coal G 66.6 30.8 2.6
Coal H 69.5 22.4 8.1
Coal I 68.0 23.0 9.0
Coal J 68.5 17.7 13.8
Coal K 63.2 23.5 13.3
Fuel ratio (-)
Bu
rno
ut
(%)
0 1 2 3 4 5 6 740
50
60
70
80
90
100
A
B
C
DE
F
G
H
IJ
K
Reaction temperature: 12000C
Particle size: 100-200 mesh
R2=0.91
Figure 8.3 Correlation between burnout and fuel ratio under the standard
combustion conditions.
219
8.3.2 Influences of reaction temperature and particle size
Subsequently, the influence of reaction temperature on the burnout of three selected
coals, Coal C, Coal D and Coal I with the particle size of 100-200 mesh, was examined
in Figure 8.4, where three reaction temperatures of 1000, 1100 and 1200°C were
considered. As seen in Figure 8.4, an increase in the reaction temperature facilitated the
burnout of the tested coals, and the promotion of the burnout on the coals with high fuel
ratio (i.e. Coal C and Coal D) was more pronounced than that on the coal with low fuel
ratio (i.e. Coal I). It follows that increasing blast temperature is an effective
countermeasure to reduce the production of unburned char in the raceway of blast
furnace, especially when the coals with high fuel ratio are blended.
Temperature (0C)
Bu
rno
ut
(%)
1050 1100 1150 1200 1250 1300 135020
30
40
50
60
70
80
90
100
Coal C (FR=5.61)
Coal D (FR=4.74)
Coal I (FR=1.55)
Particle size: 100-200 mesh
Figure 8.4 Distributions of burnout of Coal C, Coal D and Coal I at various
reaction temperatures.
220
For the purpose of increasing the reaction surface of injected pulverized coals for higher
burnout, in the PCI operation worldwide the raw coal is commonly ground to a size
specified by 70-80% of coal particles passing through a 200 mesh sieve (Jaffarullah and
Ghosh, 2005). On the other hand, the injection of granular coal (10-30% minus 200
mesh) is being applied at British Steel for the reduction of coal preparation energy
(Gathergood and Jukes, 1996). To investigate the effect of particle size on coal
combustion, Coal B, Coal E and Coal I were pulverized into three different size regions,
60-100 mesh, 100-200 mesh and 200-325 mesh, respectively. Figure 8.5 demonstrates
the burnout profiles of the three coals where the reaction temperature was 1200oC. The
results presents that the burnout of the tested coals rose when the size of coal particles
was decreased from 60-100 mesh to 100-200 mesh. However, the coal burnout was
slightly decreased when the finer coals with the size of 200-325 mesh were tested. In the
study of Chen et al. (2008) analyzing the particle sizes of unburned char after coal
devolatilization tests, it was reported that when the feed particle size was in the range of
100-200 mesh, the unburned char particles were characterized by swelling, partial
fragmentation and particle agglomeration. Alternatively, as the feed particle size was
200-325 mesh, particle agglomeration was the dominant mechanism of the unburned
char formation. In contrast to the present results, it was presumed that the particle
agglomeration became significant when the fine coals (200-325 mesh) were tested. As a
result, the coal combustion could not be improved further, as observed. This further
implies that the excessive grinding may be avoided in PCI operation, especially when a
high injection rate or dense phase transportation is practiced.
If one further examines the burnout percentages shown in Figures 8.4 and 8.5, it should
be pointed that the burnout difference between different coals tends to be enlarged when
the fuel ratio of coal becomes small, regardless of the variation of reaction temperature
or particle size. Specifically, in Figure 8.4 the burnout difference between Coal I and
Coal D is more pronounced than that between Coal C and Coal D, and in Figure 8.5 the
burnout difference between Coal I and Coal E is more obvious compared to that
221
between Coal E and Coal B. Physically, the lower the fuel ratio of a coal, the higher the
volatile matter contained in the coal. It is known that the reactivity of volatile matter is
generally much higher than that of fixed carbon. A coal with low fuel ratio can be
reacted and depleted easily compared to that with high fuel ratio. This is the reason that
the burnout difference between different coals is amplified as the fuel ratio of a coal
becomes small.
8.3.3 Burnout of blended coals
Upon inspection of Figure 8.3, significant differences in the combustion efficiency
among the coals have been found. It follows that the stability of the lower zone of a
blast furnace may be affected when coal blends are changed. Therefore, the experiments
of coal blend combustion were carried out to figure out the burning behavior of blended
coals. In this study, Coal K, a coal with low fuel ratio, was individually mixed with
Coal A and Coal C, characterized by high fuel ratios, at various blending ratios.
Coal particle size (mesh)
Bu
rno
ut
(%)
20
40
60
80
100
60-100 200-325100-200
Coal I (FR=1.55)
Coal B (FR=6.22)
Reaction temperature: 12000C
Coal E (FR=3.87)
Figure 8.5 Distributions of burnout of Coal B, Coal E and Coal I at various
particle sizes.
222
Figure8.6 depicts that the burnout of the coal blends was linearly predictable from the
combustion performance of individual coals. In other words, no synergistic effect of
combustion (Chen and Wu, 2009) was obtained when two different coals were mixed
and burned. It is thus pointed out that the blending ratio of a PCI coal may be adjusted
from the burnout of individual coals given by the DTF rather than by the fuel ratio. On
the other hand, in examining the distributions shown in Figure 8.6, the coal blends
containing Coal A reached higher burnout than that containing Coal C. Taking the
advantage of the better combustion efficiency from the blends with Coal A, a higher
blending ratio of Coal A can be arranged for the increase of calorific value of a PCI
coal. This implies that the coke replacement ratio in an ironmaking process may be
promoted. Notably, this coincides with the operation experience at CSC’s blast
furnaces. It is thus concluded that the DTF system developed in this study is capable of
provide a practical evaluation for the selection of PCI coals employed in blast furnaces.
Blending ratio (%)
Bu
rno
ut
(%)
0 20 40 60 80 10020
40
60
80
100
Blended with Coal C
Blended with Coal A
Reaction temperature: 12000C
Particle size: 100-200 mesh
Pure Coal K
Figure 8.6 Distributions of burnout with respect to blending ratio for Coal K
individually blended with Coal A and Coal C.
223
8.4 Conclusions
A variety of coals and coal blends have been tested using a developed drop tube furnace
to evaluate the combustion characteristics of the coals applied for PCI in blast furnaces.
The present study also considered the effects of reaction temperature, particle size and
blending ratio of two different coals on fuel burnout. In the individual coal tests, it was
found that a coal with higher fuel ratio had a lower value of burnout. However, one coal
with the fuel ratio of 6.22, denoted by Coal A, was characterized by its superiority in
combustion. On the contrary, the burnout of another coal with the fuel ratio of 5.61,
designated by Coal C, was significant lower than the general trend. This indicates that
the combustion behavior of coal depends not only on the fuel ratio but also on the coal
nature. The burnout of a coal could be enhanced by increasing the reaction temperature
or reducing the particle size of the coal. Nevertheless, when the particle size of the coal
was reduced from 100-200 mesh to 200-325 mesh, the combustion efficiency of the
coal could not be improved any more. It can be explained by particles agglomeration
happened when the particle size was small to a certain extent. The burnout of coal
blends was linearly predictable from the combustion efficiency of individual coals in
the blends. Therefore, the blending ratio of a PCI coal may be adjusted from the burnout
of individual coals given by the DTF rather than by the fuel ratio. From the burnout
tests, the DTF system developed in this study has been proved to be capable of
providing a useful evaluation for PCI coals utilized in blast furnaces.
224
CHAPTER 9
PRETREATMENT OF BIOMASS BY TORREFACTION AND
CARBONIZATION FOR COAL BLEND USED IN PULVERIZED
COAL INJECTION
This chapter presents a fundamental insight into the pre-treatment of biomass and the
combustion characteristics of pulverised biofuels under conditions pertinent to the
raceway of blast furnace.
Du, S. W., Chen, W. H. and Lucas, A. J. (2014), Pretreatment of biomass by
torrefaction and carbonization for coal blend used in pulverized coal injection,
Bioresource Technology, vol. 161, p. 333-339.
225
ABSTRACT
To evaluate the utility potential of pretreated biomass in blast furnaces, the fuel
properties, including fuel ratio, ignition temperature, and burnout, of bamboo, oil palm,
rice husk, sugarcane bagasse, and Madagascar almond undergoing torrefaction and
carbonization in a rotary furnace are analyzed and compared to those of a high-volatile
coal and a low-volatile one used in pulverized coal injection (PCI). The energy densities
of bamboo and Madagascar almond are improved drastically from carbonization,
whereas the increase in the calorific value of rice husk from the pretreatment is not
obvious. Intensifying pretreatment extent significantly increases the fuel ratio and
ignition temperature of biomass, but decreases burnout. The fuel properties of pretreated
biomass materials are superior to those of the low-volatile coal. For biomass torrefied at
300 °C or carbonized at temperatures below 500 °C, the pretreated biomass can be
blended with coals for PCI.
Keywords: Torrefaction and carbonization; Biomass and biochar; Burnout and
ignition; Rotary furnace; Pulverized coal injection (PCI); Blast furnace.
226
9.1 Introduction
Biomass is able to fix atmospheric carbon while it grows; therefore, biomass is regarded
as a carbon-neutral fuel when it is burned. For this reason, using biomass as an
alternative fuel to fossil fuels is considered as an effective countermeasure to reduce
anthropogenic carbon dioxide emissions into the atmosphere and mitigate global
warming (Machado et al., 2010). For example, bioethanol and biodiesel have been
extensively employed for power generation in spark ignition engines and compression
ignition engines, respectively (Gustavo et al., 2013). In addition to the liquid biofuels,
biomass can also be combusted directly to get heat and power. Compared to coals, the
energy density of biomass is low and its moisture content is high (Rousset et al., 2011).
Moreover, more energy will be consumed to comminute biomass due to its
lignocellulosic nature (van der Stelt et al., 2011). These characteristics limit the
applications of biomass in industry.
As far as blast furnaces are concerned, coke, produced from metallurgical coal, is an
essential reducing agent and provides thermal energy for hot metal production (Du and
Chen, 2006). By means of the technique of pulverized coal injection (PCI), non-coking
or weakly coking coals are injected into the raceways of blast furnaces to partially
replace coke (Chen et al., 2007; Du et al., 2007). On account of mass consumption of
coals for cokemaking and PCI in blast furnaces, a large amount of CO2 is emitted from
the ironmaking processes (Wang et al., 2009). Solid biomass is a potential substitute to
coals and can be partially used for PCI without net carbon dioxide emissions into the
atmosphere (Chen and Wu, 2009). However, due to the disadvantages of raw biomass
described earlier, the upgrade of raw biomass is necessary for its application in blast
furnaces.
The upgrade of biomass can be fulfilled via torrefaction and carbonization or pyrolysis
where biomass is thermally degraded in an inert or oxygen-free environment. The
torrefaction temperature is in the range of 200-300 °C (Peng et al., 2013; Lu et al.,
2012; Sabil et al., 2014), whereas carbonization is operated at temperatures of 300-500
227
°C (Abdullah and Wu, 2009). The biomass materials pretreated from torrefaction and
carbonization are called torrefied biomass and biochar, respectively. By virtue of the
partial disruption of lignocellulosic structure in biomass from the two methods, biomass
grindability is improved greatly (Aris et al., 2008). Grinding coal for PCI is an
energy-intensive process. Therefore, the energy for grinding coal can be saved if
torrefied biomass and biochar are used for PCI (Abdullah and Wu, 2009; Phanphanich
and Mani, 2010). Torrefaction and carbonization lead to the release of volatile matter
from biomass and change the hygroscopic material to hydrophobic one. This
transformation improves the reactivity of solid biomass. Bridgeman et al. (2008) studied
raw and torrefied willow exposed to a methane-air flame, and found that the latter was
ignited more quickly than the former. Pimchuai et al. (2010) investigated rice husk
reaction in a spout-fluid bed combustor, and reported that torrefied rice husk ignited
faster and raised the bed temperature to a higher level when compared to raw rice husk.
These ignition observations were very likely due to the low moisture content in the
torrefied willow and rice husk.
When fuel particles are injected into blast furnaces, they proceed from blowpipes,
tuyeres, and then to raceways, and experience rapid heating, devolatilization, gas-phase
combustion, char combustion, and gasification (Hutny et al., 1991; Shen et al., 2009a;
Wijayanta et al., 2014). Devolatilization and gas-phase combustion correspond to the
mass transfer and reactions of volatile matter from fuel particles, while char combustion
and gasification account for the reactions of fixed carbon. Accordingly, particle
reactions are highly related to the volatile matter and fixed carbon contents in the fuels.
The ignition temperature of volatile is much lower than that of char. Therefore, the first
stage of fuel particle reactions is triggered by volatile ignition, while char combustibility
is subject to its residence time in the reactor and the surrounding temperature (Du et al.,
2010). However, after biomass is torrefied or pyrolyzed, part of the volatiles are
liberated from the material and relatively more fixed carbon is retained (Chen et al.,
2012). This may lower the ignition temperature of biomass in the gas phase.
228
Coals with high fuel ratios are frequently blended with low fuel-ratio coals to increase
the flexibility of PCI operation (Du et al., 2010). When biomass is used as an alternative
fuel to coals for PCI, its utility can be evaluated through a number of properties, such as
fuel ratio, ignition temperature, and burnout (Gao and Bian, 2013; Li, et al., 2014). To
the authors’ knowledge, the pretreatment of biomass simultaneously covers torrefaction
and carbonization has not been studied yet. The purposes of the present study are to
examine the fuel properties of biomass pretreated by torrefaction and carbonization and
compare to those of a high-volatile coal and a low-volatile coal. Particular emphasis will
be paid to the applications of upgraded biomass, from the viewpoint of coal blend used
in PCI.
9.2 Experimental
9.2.1 Materials and preparation
Five different biomass materials were studied in the present work; they are bamboo, oil
palm, rice husk, sugarcane bagasse (abbreviated by bagasse), and Madagascar almond.
The bamboo, rice husk, bagasse, and Madagascar almond were obtained in Taiwan. The
oil palm was the fiber fraction left after the nut was removed in a Malaysia oil
extraction mill. Oil palm is an important economic crop in some countries, especially in
Malaysia. Oil palm fibers are abundant wastes from palm oil fruit harvest and oil
extraction processing. The fibers are considered as a potential renewable energy source
due to its high calorific value and quantity (Shuit et al., 2009). Therefore, oil palm fiber
is adopted and studied in the present study.
Meanwhile, a high-volatile bituminous coal (Coal A) and a low-volatile coal (Coal B)
for PCI operation at China Steel Corporation (CSC) were tested for comparison. The
basic properties of the coals and biomass materials, such as proximate, elemental, fiber,
and calorific analyses, are given in Table 9.1. The proximate analysis was performed in
accordance with the standard procedure of the American Society for Testing and
Materials (ASTM E870-82). The elemental analysis was carried out by use of an
229
elemental analyzer (Vario EL III). The fiber contents (hemicellulose, cellulose, and
lignin) in biomass were analyzed through the measurements of neutral detergent fiber,
acid detergent fiber, and ash (Chen, et al., 2010b). The higher heating values (HHVs) of
samples were detected by a bomb calorimeter (IKA C2000 Basic). As shown in the
Table, the higher heating values (HHVs) and fixed carbon values of the five biomass
species are in the range of approximately 17-19 MJ kg-1
and 9-20 wt%, respectively,
which are much lower than those of Coal A (i.e. 23.99 MJ kg-1
and 40.47 wt%) and
Coal B (i.e. 31.01 MJ kg-1
and 76.35 wt%). Bamboo, oil palm, and Madagascar almond
were shredded by a cutting shredder to the particle sizes of 5-10 mesh (i.e. 2-4 mm),
whereas rice husk and bagasse were not treated. The samples were stored in a nitrogen
atmosphere at 75 ºC until biomass pretreatment was performed.
Table 9.1 Proximate, elemental, fiber, and calorific analyses of two coals and raw
biomass materials.
Coal A Coal B Bamboo Oil
palm
Rice
husk
Bagasse Madagascar
almond
Proximate analysis (wt%)
Moisture 13.92 1.09 5.76 7.20 8.00 7.03 10.17
VM 44.09 14.67 78.76 67.25 73.18 75.03 70.38
FC 40.47 76.35 14.40 19.03 9.27 13.61 18.62
Ash 1.52 7.89 1.08 6.52 9.55 4.33 0.83
Elemental analysis (wt%)
C 63.72 83.24 48.64 44.81 43.40 46.38 47.68
H 4.40 3.78 5.64 4.10 4.33 4.68 4.31
N 0.67 1,62 0.52 2.10 0.65 0.50 0.50
S 0.10 0.52 0.03 0.24 0 0 0
O * 29.59 1.86 44.09 42.23 42.07 44.11 46.68
Fiber analysis (wt%)
Hemicellulose 20.38 34.00 21.34 30.59 18.23
Cellulose 39.82 26.78 36.06 45.66 41.86
Lignin 12.16 16.08 21.16 19.38 16.17
Others 27.64 23.14 41.44 5.37 23.74
Higher heating value (MJ kg-1
)
23.99 31.01 18.95 17.12 17.46 18.31 17.32
* By difference
230
9.2.2 Burnout and ignition tests
In addition to the aforementioned basic analyses, the fuel properties of the pretreated
biomass samples were also analyzed through burnout and ignition tests. The burnout of
the samples was tested by a drop tube furnace. The details of the structure and
experimental procedure of the drop tube furnace can be found elsewhere (Chen et al.,
2012). In brief, the pretreated samples were ground into powders by a shedder followed
by sieving using vibrating screens to the particle sizes of 100-200 mesh (i.e. 74-149
μm). The particles were introduced into the drop tube furnace by nitrogen at a flow rate
of 2 L min-1 (25°C), while the secondary gas, namely, air, at a flow rate of 3 L min-1
was preheated to 350°C and blown into the furnace, acting as an oxidizing agent. The
feeding rate of the particles was approximately 5 g h-1 and the reaction temperature in
the furnace was 1000 °C. The combustibility or combustion efficiency of a solid fuel
can be measured in terms of burnout. The ash tracer method (Du et al., 2010; Chen et
al., 2012) was adopted to determine the burnout of a tested fuel. The burnout is
expressed as
010
100
Ash1Ash
AshAsh(%)Burnout
rawuc
rawuc
(9.1)
where Ashuc and Ashraw designate the ash contents in the unburned char and raw fuel,
respectively.
With regard to the ignition test, the biomass samples were mixed with sodium nitrite
(NaNO2) at a weight ratio of 1:075 and the total amount was 0.2 g in each test. The
sodium nitrite was used as an ignition booster for the ignition test. The mixture was
placed in an ignition analyzer which was heated at a heating rate of 95 °C min-1 under
the open atmosphere. The heating temperature was recorded by a computer, while the
ignition process was shot by a high-speed camera at a frequency of 30 Hz and the
images were stored in the computer. The brightness in the images rose rapidly when an
231
ignition occurred. From the recorded temperature and images, the ignition temperature
of a sample was identified.
To ascertain the analysis quality, the drop tube furnace was leak-tested using nitrogen,
and the axial temperature distribution in the furnace was measured. The measurements
indicated that no leakage occurred in the drop tube furnace, and the temperature
distribution was fairly uniform at the center of the furnace, revealing the spatially
homogeneous gas temperature in the reaction zone. The burnout experiments were
carried out at least twice for some tested samples to ensure the experimental stability
and accuracy. The relative error between each run was controlled within 5%. Moreover,
the ignition temperature of a sample was defined from the average temperature of nine
tests where the deviation of temperature was control within 3 °C.
9.3 Results and discussion
9.3.1 Proximate analysis and van Krevelen diagram
The profiles of the volatile matter (VM), fixed carbon (FC), and fuel ratio (=FC/VM) of
the five biomass species before and after pretreatment are displayed in Figure 9.1. As a
whole, VM linearly decrease with increasing pretreating temperature (Figure 9.1a),
whereas FC linearly increases (Figure 9.1b). Similar behavior has been observed in the
studies of Couhert et al. (2009) and Lu et al. (2012). Because of this, increasing
pretreating temperature has a trend to raise the fuel ratio exponentially (Figure 9.1c).
When the biomass materials are pretreated at 300 °C or higher temperatures, their FC
values and fuel ratios are larger than those of Coal A (i.e. 40.47 wt% and 0.92).
Alternatively, the VM contents in the samples are lower than that in Coal A (=44.09
wt%) when the temperature is as high as 400 °C. In contrast, the VM contents of the
pretreated materials are always higher than that of Coal B (=14.67 wt%). The FC
contents and fuel ratios of the pretreated materials are lower than those of Coal B (i.e.
76.35 and 5.20), except for bamboo carbonized at 500 °C. From the viewpoint of
proximate analysis, the obtained biomass materials from the pretreatment at
232
temperatures of 300-500 °C are between a high-volatile bituminous coal and a
low-volatile bituminous coal. FC in a fuel gives a higher contribution on heat release
than VM when the fuel is burned (Parikh et el., 2005). This implies, in turn, that the
higher the fuel ratio in biomass, the higher the calorific value of the biomass.
Accordingly, the thermal pretreatment should be operated at temperatures equal to or
higher than 300 °C, from the viewpoint of coal replaced by biomass.
The profiles of atomic H/C ratio versus atomic O/C ratio, namely, the van Krevelen
diagram, of the investigated materials are plotted in Figure 9.2. The linear regression of
the data is also shown in the figure. The coefficient of determination (R2) of the linear
regression is 0.9312. This indicates that there exists a strong linear correlation between
the atomic H/C ratio and the atomic O/C ratio, whether the biomass is torrefied or
carbonized. The slope of the regression line is 1.681, revealing that the impact of the
thermal pretreatment on the H/C ratio is higher than on the O/C ratio. In other words,
more hydrogen is depleted from the thermal degradation when compared to oxygen.
This behavior is different from the observations of Rousset et al. (2011) and Chen et al.
(2012) where only torrefaction was practiced. It follows that the influence of higher
pretreating temperatures (i.e. carbonization) on hydrogen is more than on oxygen. In
light of the profiles shown in Figure 9.2, the van Krevelen diagram can be partitioned
into three different regions. For the raw biomass materials, the atomic O/C and H/C
ratios are larger than 0.53 and 1.17, respectively. After undergoing torrefaction, the O/C
and H/C ratios of the materials are in the ranges of approximately 0.16-0.53 and
0.65-1.17, respectively. When the samples are carbonized, the atomic O/C and H/C
ratios are smaller than 0.16 and 0.65, respectively. The atomic H/C and O/C ratios of
Coal A are 0.83 and 0.31, respectively, which are located in the torrefaction regime.
Alternatively, the H/C and O/C ratios of Coal B are 0.54 and 0.02, respectively,
situating in the carbonization regime.
233
Temperature (oC)
Fix
ed
ca
rbo
n(w
t%)
0
20
40
60
80
100
Raw 250 300 400 450 500
(b)
Temperature (oC)
Vo
latile
ma
tte
r(w
t%)
0
20
40
60
80
100
BambooOil palmRice huskBagasseMadagascar almond
Raw 250 300 400 450 500
(a)
Coal B
Coal A
Temperature (oC)
Fu
elra
tio
(=F
C/V
M)
0
1
2
3
4
5
6
Raw 250 300 400 450 500
(c)
Figure 9.1 (a) Volatile matter, (b) fixed carbon, and (c) fuel ratio values of raw and
pretreated biomass materials.
234
Atomic O/C ratio
Ato
mic
H/C
ratio
0 0.1 0.2 0.3 0.4 0.5 0.6 0.70
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
Bamboo
Oil palm
Rice husk
Bagasse
Madagascar almond
Coal A
Coal B
y = 1.681x + 0.305R
2= 0.9312
Raw
Carbonization
Torrefaction
9.3.2 Solid yield and energy yield
The profiles of the HHVs of the coals and biomass are displayed in Figure 9.3a. When
the biomass is torrefied, the HHV is intensified obviously. In contrast, the increment in
the HHV of the biomass under carbonization is not as significant as that under
torrefaction, especially for rice husk. The HHV of pretreated rice husk is always lower
than that of Coal A; hence, rice husk is not recommended to replace coals used in blast
furnaces. The ash contents of the pretreated biomass samples are given in Table 9.2.
When rice husk is thermally pretreated from 250 to 500 °C, the ash content goes up
from 13.54 to 26.60 wt%. In view of the high ash contents in the raw and pretreated rice
husks, their calorific values are always lower than that of Coal A. Alternatively, the ash
contents in bamboo and Madagascar almond merely increase from 1.08 to 3.71 wt% and
from 0.83 to 3.78 wt%, respectively. It has been addressed that the thermal degradation
of fibrous biomass (i.e. oil palm, rice husk, and bagasse) is inherently different from
Figure 9.2 Atomic H/C versus O/C ratio (van Krevelen diagram) of raw and
pretreated biomass materials.
235
that of ligneous biomass (i.e. bamboo and Madagascar almond) (Chen et al., 2013). On
the other hand, hemicellulose is the most active constituent in biomass (Chen and Kuo,
2010a) which will be thermally decomposed drastically from the torrefaction and
carbonization. Table 9.1 suggests that the hemicellulose contents in raw bamboo and
Madagascar almond are 20.38 and 18.22 wt%, which are relatively lower than the other
species. This is the reason that carbonization is still able to effectively intensify the
calorific values of bamboo and Madagascar almond. Meanwhile, it can be seen that the
HHVs of the two species pyrolyzed at 500 °C are almost equivalent to that of Coal B.
The intention of the present study is to evaluate the potential of torrefied or carbonized
biomass for the replacement of coals consumed in blast furnaces. A physical scale of the
replacement ratio is defined as the HHV ratio between pretreated biomass and a coal.
Consequently, the calorific value of a fuel is higher than that of a coal when the ratio is
larger than unity. Overall, the replacement ratios based on Coal A and Coal B are in the
ranges of 0.71-1.32 (Figure 9.3a) and 0.55-1.02 (Figure 9.3b), respectively. When the
five biomass materials are torrefied at 250 °C, their replacement ratios based on Coal A
are always lower than unity. The replacement ratios of bamboo and Madagascar almond
are larger than unity when they are torrefied at 300 °C, whereas the ratios of the other
three species are smaller than unity. After undergoing carbonization, the replacement
ratios of the samples are higher than unity, with rice husk giving an exception. From the
calorific view of point, the biomass materials except for rice husk can be employed to
replace high-volatile bituminous coals when they are carbonized. Alternatively, only
bamboo and Madagascar almond carbonized at 500 °C can be utilized to replace Coal B
or low-volatile bituminous coals.
The enhancement factors of the HHVs of the five biomass materials are tabulated in
Table 9.3. The factor stands for the HHV ratio between the pretreated biomass and its
parent biomass. Rice husk has the lowest enhancement factor among the five species, as
a result of high ash content (Table 9.2). Alternatively, the enhancement factors of
bamboo and Madagascar almond are relatively high, especially for the latter, which is
236
lifted up to 1.79 when the pretreating temperature is 500 °C. This can be explained by
the low ask contents in the two torrefied materials (Table 9.2). The profiles of the solid
yield and energy yield of the biomass at various pretreating temperatures are
demonstrated in Figure 9.4. The former and the latter are the weight ratio and the energy
ratio between the pretreated biomass and the raw one, and the energy yield is the
multiplication of the solid yield and the enhancement factor of HHV (Lu et al., 2012).
The solid yield is lower than 50% when the biomass is carbonized, regardless of which
sample is pretreated. This reflects that over 50 wt% of raw biomass is consumed from
biochar production at temperatures higher than 400 °C. Though the HHV of
Madagascar almond is promoted markedly from carbonization (Figure 9.3a), it is
noteworthy that its solid yield is lessened obviously, ranging from 22 to 25 wt%. This
behavior is similar to the pyrolysis of mallee wood (Abdullah and Wu, 2009). As a
consequence, only around 40% of energy is retained in the biochar carbonized from
Madagascar almond. The energy yield of rice husk is also lower than 50% from its
carbonization. In contrast, bagasse has the highest energy yield at temperatures of
300-450 °C. This can be explained by the highest cellulose content in bagasse (Table
9.1) among the five raw biomass species.
Table 9.2 Ash contents in pretreated biomass materials.
Temperature (°C) Bamboo Oil palm Rice husk Bagasse Madagascar
almond
250 1.48 10.35 13.54 5.56 1.21
300 2.02 13.12 23.68 6.33 1.79
400 2.69 14.36 24.68 8.61 3.35
450 3.11 15.18 25.41 9.19 3.54
500 3.71 17.85 26.60 13.25 3.78
237
Temperature (oC)
HH
V(
MJ
/kg
)
15
20
25
30
35
Bamboo
Oil palm
Rice husk
Bagasse
Madagascar almond
Raw 250 300 400 450 500
Coal B
(a)
Coal A
Temperature (oC)
Re
pla
ce
me
nt
ratio
0.4
0.6
0.8
1
1.2
1.4
1.6
(b)
Coal A
Raw 250 300 400 450 500
Temperature (oC)
Re
pla
ce
me
nt
ratio
0.4
0.6
0.8
1
1.2
1.4
1.6
(c)
Coal B
Raw 250 300 400 450 500
Figure 9.3 (a) HHVs and replacement factors of biomass materials based on (b)
Coal A and (c) Coal B.
238
9.3.3 Ignition and burnout
The profiles of the ignition temperatures of the coals and raw/pretreated biomass
samples are shown in Figure 9.5. VM is the main factor triggering biomass ignition due
to the gas-phase reaction in a high temperature environment. Without the consideration
of bagasse, the ignition temperatures of the raw biomass materials are in the range of
261-271 °C which are lower than that of Coal A (302 °C) to a certain extent and by far
lower than that Coal B (418 °C), attributing to the high VM contents in the raw biomass
Temperature (oC)
So
lid
yie
ld(%
)
200 250 300 350 400 450 500 5500
20
40
60
80
100
Bamboo
Oil palm
Rice husk
Bagasse
Madagascar almond
(a)
Temperature (oC)
En
erg
yyie
ld(%
)
200 250 300 350 400 450 500 5500
20
40
60
80
100
(b)
Figure 9.4 (a) Solid yield and (b) energy yield of pretreated biomasses materials.
239
samples (Table 9.1). The higher ignition temperature of raw bagasse (315 °C) might be
due to the high cellulose content in the biomass. An increase in pretreating temperature
almost linearly increases the ignition temperature, with the exception of bagasse. As
long as the biomass materials are carbonized, their ignition temperatures are higher than
that of Coal A. This implies that more traveling time for carbonized biomass particles in
raceways is required to activate combustion reaction when compared to Coal A.
Nevertheless, the pretreated biomass can be ignited more easily than that of Coal B.
The profiles of the burnout of the biomass materials versus pretreating temperature are
presented in Figure 9.6a. The burnout has a trend to decrease linearly with increasing
pretreatment temperature. For biomass carbonized under the same temperature, their
burnout values are close to each other, and the values are lower than that of Coal A but
substantially higher than that of Coal B. To keep a constant fuel rate for hot metal
production and reasonable combustion efficiency in the raceways of blast furnaces, the
carbonized biomass examined in this study can be used as the injection fuel, except that
from rice husk. The profiles of burnout versus fuel ratio are shown in Figure 9.6b. In a
previous study (Chen, et al, 2012), a linear correlation between the burnout and the fuel
ratio was exhibited when biomass was torrefied. This implies, in turn, that the
combusting behavior of biomass could be predicted from its fuel ratio, whether the
biomass is torrefied or not. However, when torrefaction and carbonization are
simultaneously considered, as seen in Figure 9.1c, the fuel ratio grows rapidly when the
pretreating temperature increases. The reactivity of VM in an oxidizing environment is
much higher than that of FC, stemming from the gas-phase reaction. Therefore, the
higher the fuel ratio in a fuel, the lower the reactivity or burnout of the fuel is. By virtue
of the significant growth in the fuel ratio with increasing pretreatment severity, the
burnout substantially decays and no linear correlation is observed.
The study of Basu et al. (2013) suggested that mass and energy yields from biomass
torrefaction increased with increasing particle length but decreased with particle
diameter. The analysis of Peng et al. (2012) indicated that the torrefaction rate was
240
affected by the particle size, especially at high temperatures. Accordingly, the particle
size distribution has a significant effect on the torrefaction and pyrolysis processes, even
the ignition and burnout of pretreated biomass. This arises from the fact that heat
transfer and decomposition behavior are related to the particle surface and the total
particle surface depends strongly on the particle size. In the present study, bamboo, oil
palm, and Madagascar almond were shredded by a cutting shredder to the particle sizes
of 5-10 mesh. Though rice husk and bagasse were not treated, their particle sizes, in
fact, are close to those of bamboo, oil palm, and Madagascar almond in the study.
Accordingly, it should be illustrated that the effect of particle size on in this study not
considered. Alternatively, in reviewing recent studies concerning economic analysis, the
overall cost, including the production, transportation and logistics costs, for torrefied
pellets has been evaluated and compared to regular pellets (Bergman, 2005; Peng et al.,
2010). It was reported that torrefied pellets were cheaper than regular pellets. However,
the economic analysis of biomass torrefaction alone is absent so far. Therefore, the
comparison in cost between torrefaction and pyrolysis remains unknown. The topics of
the effect of particle size on torrefaction and carbonization and economic analysis
deserve further investigation in the future.
Temperature (oC)
Ign
itio
nte
mp
era
ture
(oC
)
250
300
350
400
450
BambooOil palmRice huskBagasseMaragascar almond
Coal A
Raw 250 300 400 450 500
Coal B
Figure 9.5 Ignition temperatures of raw and pretreated biomass materials.
241
9.4 Conclusions
The utility potential of torrefied and carbonized biomass materials for PCI in blast
furnaces has been evaluated. A strong linear correlation between the atomic H/C ratio
and O/C ratio in biomass over the operations of torrefaction and carbonization is
exhibited. The fuel properties, such as fuel ratio, burnout, and ignition temperature, of
biomass torrefied at 300 °C or pyrolyzed between 400 and 500 °C, are between a
Temperature (oC)
Bu
rno
ut
(%)
0
20
40
60
80
100
Bamboo
Oil palm
Rice husk
Bagasse
Madagascar almond
Raw 250 300 400 450 500
Coal B
(a)
Coal A
Fuel ratio
Bu
rno
ut
(%)
0 1 2 3 4 5 620
40
60
80
100
Increasingpretretmentseverity
(b)
Coal A
Coal B
Figure 9.6 Burnout versus (a) pretreated temperature and (b) fuel ratio.
242
high-volatile bituminous coal and a low-volatile one. Therefore, the pretreated biomass
can partially replace the coals consumed for PCI and blends with coals to keep
reasonable burnout in raceways.
243
CHAPTER 10
CONCLUSIONS AND RECOMMENDATIONS
This chapter summarises the main achievements of the work and their applications in
the PCI operation at CSC. Besides, the further works for the improvement of PCI are
also recommended.
244
10.1 Introduction
The goal of this thesis is to advance knowledge on the coal combustion in the regions of
blowpipe, tuyere and tuyere of blast furnace through modelling and experimental
studies. The next section summarises the main achievements and conclusions of the
studies. In addition, the applications of the studies in the blast furnaces of China Steel
Corporation are also briefly talked. Further works for improving PCI operation are
recommended in the subsequent section.
10.2 Achievements and conclusions
10.2.1 Modelling
The 3-D CFD coal combustion model has been established in this work to find the
factors influencing the coal combustion for PCI operation and to develop methods to
stabilise the operation of blast furnace raceway. The development of the model
comprised 4 phases:
(1) validation of kinetic parameters of the coal devolatilisation model (Section 2.3.1.2
and Chapters 3 and 4) ;
(2) modelling of coal burning characteristics in the regions of blowpipe and tuyere at
different operation conditions and injection patterns (Chapters 3 and 4);
(3) examination of coal blend combustion in the regions of blowpipe, tuyere and
simplified raceway (Chapter 5); and
(4) development of a comprehensive coal combustion in the blowpipe, tuyere and
raceway featured by a porosity contour of 0.4 in a packed bed (Section 2.4 and
Chapter 6).
To validate the kinetic parameters used in the single and two competing devolatilisation
models, the coal combustion behaviours within the combustion rig (Burgess et al.,
1985) s simulated in the first phase. The RNG k-ε model (gas phase turbulent),
Lagrangian approach (particle trajectory) and the mixture fraction probability density
function model (turbulent combustion) were incorporated into the model. By comparing
245
the predicted temperature distribution with the experimental data of Burgess et al.,
(1985), it is found that the kinetic parameters proposed by Ubhayakar et al., (1976) for
the two competing devolatilisation model is able to sufficiently reflect the pulverised
coal burning characteristics. According to the calculated gas temperature, the PCI coal
combustion can be partitioned into two stages: a very fast step of volatile combustion
followed by a much slower step of char combustion. It suggests that the coal burnout
within the region of blowpipe-tuyere-raceway is largely contributed by the volatile
release rather than by char combustion.
In the second phase, the sensitivity analysis of the operation conditions on the coal
burning in the regions of blowpipe and tuyere was carried out. The simulation suggests
that either increasing the hot blast gas temperature or decreasing the mass flow rate of
carrier gas is able to promote the coal burning within the tuyere. Besides, the
application of the tuyere embedded with ceramic sleeve can be a countermeasure for
enhancing coal burnout. When tracking the trajectories of coal particles, it is observed
that the mixing of the coal particles and hot blast gas played a crucial role in heating,
devolatilisation and combustion of the injected coal particles. In other words, the coal
burnout can be increased when the mixing of coal particles with the hot blast is
improved. From the simulation, earlier ignition was achieved with the operation of
double lance in comparison with the single one. Accordingly, the double air-cooled
coaxial lance system was developed at No3 blast furnace of CSC in 2001. The average
PCI rate of No3 blast furnace was promoted from 110 to 153 kg/tHM within 6 months in
2001. Following successful plant trials, the single lance injection was replaced by the
double air-cooled coaxial lance system in 2002 at the blast furnaces of CSC.
The region of calculation was extended to the jet zone of raceway in the third phase. In
the calculation, the influence of volatile matter content in PCI coal on combustion was
especially examined. Practically, the reduction of volatile matter content in PCI coal can
be made by blending of low volatile coals into high ones. It was found that when the
coal volatile matter reduced from 35% to 25%, the coal burnout decreased from 81.5%
246
to 63%, leading to more unburnt char generated within the combustion region. The
increase of unburnt char might cause drop in gas permeability in the lower zone of the
blast furnace. On the other hand, the calculation also indicated that the pressure loss
across the combustion region could be effectively abated when the coal bend injection
was employed. With the coal blend injection, the performance of the blast furnace may
be improved as long as the consumption rate of unburnt char exceeds its accumulation
in the lower zone of blast furnace. As confirmed during the plant trials in CSC’s No3
blast furnace, the permeability resistance in the lower zone of the furnace was decreased
with the increase of low volatile coals in the coal blend from 30 to 50%. With this
advantage, the flow rate of hot blast could be increased for promoting its hot metal
production. Accordingly, the PCI operation of CSC’s blast furnaces has been shifted
from high volatile coal injection into coal blend since 2003. At present, the volatile
content of coal blend for PCI operation in CSC’s blast furnaces is kept in the range of
19 to 21%. Further improvement in PCI operation may be achieved by late ignition of
coal within the raceway while maintaining a high coal burnout.
In the last phase, the multi Eulerian-Eulerian multi–fluid model was used to determine
the raceway shape, which was featured by the voidage contour of 0.4. The calculated
raceway shape and gas composition distribution within the raceway agreed well with
the measurements reported by Nogami et al. (2004). Taking the operation conditions of
CSC’s No3 blast furnace as the boundary conditions, the burning characteristics of
pulverised coal within the regions of blowpipe, tuyere and raceway were simulated. It
was found that the coal plume was surrounded by enriched oxygen when oxy-coal lance
was used. Fuel ignition was delayed, as a consequence of the cooling effect of the
enriched oxygen. Therefore, less pressure loss across the combustion region is exhibited
in comparison with single and double lance injections. Most importantly, the coal
combustion was intensified by the enriched oxygen at the downstream of the coal
plume. As a result, the coal burnout was as high as that given by double lance injection.
247
It is concluded the oxy-coal injection could be an effective countermeasure at CSC to
optimise its blast furnaces operating.
10.2.2 Coal combustion experiments
In this research, a drop tube furnace was established to evaluate the combustion
performance of PCI coals and pretreated biomass in an environment with high heating
rate (>104
K/s). The drop tube furnace was employed to study:
(1) volatile release and particle formation characteristics of injected coal under
conditions simulating the PCI operation environments within the raceway (Chapter
7);
(2) the combustion behaviours of the PCI coals used in CSC’s blast furnaces,
considering the effects of fuel ratio (the weight ratio of fixed carbon to volatile
matter), reaction temperature, particle size and blending ratio of two different coals
on fuel burnout (Chapter 8); and
(3) the utility potential of pretreated biomass in the PCI operation in the blast furnaces
of CSC (Chapter 9).
In the first research with the drop tube furnace, the experiments were performed by
dropping coal particles into the drop tube furnace by nitrogen at temperatures between
1100 to 1400oC. An examination of the R-factor (volatile release ratio under rapid
heating condition to ASTM test) reveals that either decreasing the feed coal particle size
or increasing the reaction temperature can efficiently enhance volatiles liberation. With
rapid heating, significant agglomeration of the residual char particles is observed when
the coal particle size was reduced from 100-200 mesh to 200-325 mesh for both high
and low volatile coals tested in the experiments, implying that the reaction time of the
char particles will be elongated. Particle swelling during the coal devolatilisation is
found in all testes, regardless of coal rank and size. Notably, only the residual char
particles given by the coarser low volatile coal are characterised by fragmentation. The
formation of tiny soot particles and tar droplets converted from the release volatile
248
matters is found for the two tested coals. The experimental results indicate that their
production is highly sensitive to the volatile matter in the coal. Comparing the reactivity
of the soot to that of the unburnt char, the former is always lower than the latter. It
implies the utilisation of the injected fuels can be improved when the oxygen level
within the coal plume can be enriched.
In second step, a variety of coals with the fuel ratio (the weight ratio of fixed carbon to
volatile matter in coal ranging from 1.36 to 6.22) are tested by using the drop tube
furnace to evaluate the combustion characteristics of the coals, considering the effects
of fuel ratio, reaction temperature, particle size and blending ratio of two different coals
on fuel burnout. It is found the burnout of tested coals is increased with the decrease of
the fuel ratio. However, a coal with high fuel ratio deviates from this rule, showing
higher burnout than the trend. It is concluded to be contributed by its higher vitrinite
content in the maceral. Accordingly, this coal has become the primary low volatile PCI
coal of the blast furnaces of CSC. The experimental results show higher fuel burnout
can be achieved when the particle size of coal is reduced from 60-100 to 100-200 mesh.
However, once the size of the tested coals is in the range of 200 and 325 mesh, the
burnout can not be improved further, resulting from the agglomeration of fine particles.
According to this result, the coal quantity passing through 200 mesh has been reduced
from 80 to 60% in the CSC’s PCI operation for less energy consumed in coal grinding.
Notably, this test has become a standard procedure at CSC to evaluate the coal
combustion using the drop tube furnace before the coal is injected into the blast
furnaces.
In the third step, the fuel properties, including fuel ratio, ignition temperature, and
burnout, of five torrefied and carbonised biomasses were analyzed and compared to
those of a high volatile coal and a low volatile one used in pulverised coal injection at
CSC. A strong linear correlation between the atomic H/C ratio and O/C ratio in biomass
over the operations of torrefaction and carbonization is exhibited. The fuel ratio,
burnout, and ignition temperature, of biomass torrefied at 300 °C or pyrolysed between
249
400 and 500 °C, are close to those of the high volatile coal. Therefore, the pretreated
biomass can partially replace the high volatile coals in the coal blend for PCI operation.
Although it has been emphasised in many studies that high burnout of pulverised coal in
the regions of blowpipe, tuyere and raceway is required, as revealed in this work,
moderate pressure loss in the raceway is also essential for achieving high injection rate
or high productivity in a blast furnace.
In this work, some results obtained from the comprehensive experimental and numerical
studies have been taken into PCI operation at CSC. This seems as a limitation of the
current study, but it may have a wide range of applications for the improvement of PCI
operation.
10.3 Recommendations
10.3.1 Raceway control
To improve the performance of PCI operation by controlling the raceway, further
research works are recommended below:
(1) Investigation into raceway phenomena by considering the formation of bird’s nest.
In this study, the multi Eulerian-Eulerian multi–fluid model has been used in this
research to determine the raceway shape. As matter of fact, the structure of raceway
is more complicated than that used in the calculation. To investigate the raceway
structure and coke degradation, the sampling of coke at tuyere level has been
performed at CSC’s blast furnaces during scheduled time of shutdowns. Figure 10.1
indicates the coke particles taken from CSC’s No3 in 2013 during scheduled
shutdown. The sampled cokes could be characterised into three distinctive zones in
terms of coke size; bosh coke (0 to 55 cm in depth from wall), raceway coke (105 to
135 cm) and bird’s nest (135 to 160 cm), which makes the raceway boundary tight
and impervious. The penetration of hot blast into the deadman zone of blast furnace
could be adversely affected with an accentuated growth of bird’s nest. As a result,
250
the deadman will become inactive. For a stable operation of blast furnace, research
works related to the structure of raceway are required.
Experimental study on the raceway phenomena is undergoing at CSC using a cold
2-D model. To investigate the formation of raceway cavity and bird’s nest, the
model was filled by mung beans (breakable materials) rather than by rigid particles
in the first stage of experiments (Du and Tsai, 2014). Figure 10.2 shows the
relationship between the raceway shape and the thickness of bird’s nest. As observed
in the experiments, the bean particles circulate along the raceway boundary at a high
velocity, resulting in the generation of particle fines. In Figure 10.2c (elapsed time
of 15 minutes), a dense shell layer is found due to the accumulation of bean debris in
160 cm 135 cm 105 cm 80 cm
80 cm 55 cm 30 cm 0 (Wall)
Figure 10.1 Zonal structures in a drill core.
251
the boundary of the raceway, while the cavity volume increases. At elapsed time of
35 minutes (Figure 10.2d), the raceway is bended upwards, indicating the gas flow
within the raceway is driven towards the wall. It will increase the gas flow in the
periphery, resulted in higher heat load. Obviously, the shell formation and its
thickness play important roles for determining the permeability and gas distribution
in the lower zone of blast furnace. To preciously predict the pressure loss and the
raceway shape, the coke breakage caused by circulation and the combustion of coke
fines trapped in the bird’s nest should be taken into account in the calculation model.
This may be achieved by coupling DEM (discrete element method) with CFD code.
In DEM, the coke particle can be considered as an agglomerate of 2-3 sub particles,
and the coke breakage occurs when the impact force is higher than the bonding
strength among the sub particles.
(2) Modelling of co-injection of oxygen and pulverised coal in double lance operation.
(a) t = 0 (b) t = 2 min
(c) t = 15 min (d) t = 35 min
Injection nozzle
bird nest
Circulating fines
Figure 10.2 Formation of cavity and bird’s nest in a pack bed.
252
For the reduction the thickness of bird’s nest, individually injecting pure oxygen
into the raceway may be an effective way to improve the permeability of the
raceway.
(3) Further case studies on torrefied or carbonised bio-fuel injection in the raceway.
10.3.2 Improvement of permeability by charging pattern
The ore/coke (O/C) of burden in blast furnace increases with the increase of pulverised
coal injection rate due to the relative increase of the ferrous than the coke (Ishii, 2000).
With high O/C, the gas permeability resistance of blast furnace is increased, and the
stability of blast furnace may turn worse. Therefore, proper charging patterns are
required to improve the ascending gas distribution within the furnace. Practically, the
burden profile is considered as essential information for the evaluation of the charging
patterns and the status of the furnace. To measure the burden profile during the
operation, an online monitoring system has been successfully developed and applied at
CSC’s No3 blast furnace (Lu and Du, 2010). Accordingly, the measured profiles have
provided useful basis for determining the charging patterns of the furnace. As indicated
in Figure 10.3, the burden profile with a terrace in the vicinity of wall was found after
changing the first charging angle from 45 to 42.6o (Du et al., 2012). In general, the
burden profile with side terrace is preferred at CSC for better gas distribution and stable
burden movement in the furnace. As a result, the PCI rates of the furnaces were
gradually promoted from 140 to 170 kg/tHM, hitting the monthly highest injection
record (180.6kg/tHM) of CSC in August 2010. To increase PCI rate and fuel efficiency
of blast furnace, technologies on the on-line measurement and control of burden profile
are recommended.
253
Figure 10.3 The measured burden profiles and calculated descending rate at No3
blast furnace.
furnace
254
References
Abdullah, H. and Wu, H. W. (2009), Biochar as a fuel: properties and grindability of
biochars produced from the pyrolysis of mallee wood under slow-heating
conditions, Energy Fuel, vol. 23, p. 4174-81.
Alonso, M. J., Borrego, A .G., Alvarez, D., Kalkreuth, W. and Menendez, R. (2001),
Physicochemical transformations of coal particles during pyrolysis and combustion
Fuel, vol. 80, p. 1857-70.
Anthony, D. B. and Howard, J. B. (1976), Coal devolatilization and hydrogasification.
AIChE Journal, vol. 22, p. 625-656.
Aoki, H., Nogami, H., Suge, T., Miura, T. and Furukawa, T. (1993), Simulation of
transport phenomena around the raceway zone in the blast furnace with and
without pulverized coal injection, ISIJ International, vol. 33, p. 646-654.
Arias, B., Pevida, C., Fermoso, J., Plaza, M. G., Rubiera, F. and Pis, J. J. (2008),
Influence of torrefaction on the grindability and reactivity of woody biomass, Fuel
Processing Technology, vol. 89, p. 169-175.
Ariyama, T, Sato, M., Yamakawa, Y., Yamada, Y. and Suzuki. M. (1994), Combustion
behaviors of pulverized coal in tuyere zone of blast furnace and influence of
injection lance arrangement on combustibility, ISIJ International, vol. 34, p.
476-483.
Ariyama, T., Ueda, S., Natsui, S., Inoue, R. and Sato, M., (2007), Current technology
and future aspect on CO2 mitigation in Japanese steel industry, Proceedings of the
5th International Congress on the Science and Technology of Ironmaking,
Shanghai, China, p. 57-64.
Austin, P. R., Chew, S. J., Maldonado, D., Mathieson, J. G., Pinson, D. J. Rogers, H.
and Walsh, M. (2011), PC Injection studies at BlueScope Steel, Proceedings of the
METEC InSteel Conference, Düsseldorf, Germany, Section 27, p. 1-10.
Babich, A., Senk, D. and Fernandez, M. (2010), Charcoal behaviour by its injection into
the modern blast furnace, ISIJ International, vol. 50, p. 81-88.
Babich, A., Yaroshevskii, A. Formoso, A. Isidro, A., Ferreira, S. Cores, A. and Garcia,
L. (1996), Increase of pulverized coal use efficiency in blast furnace, ISIJ
International, vol. 36, p. 1250-58.
255
Babich, A., Gudenau, H. W., Mavrommatis, K. T., Froechling, C., Formoso, A., Cores,
A. and Garcia, L, (2002), Choice of technological regimes of a blast furnace
operation with injection of hot reducing gases, Revista de Metalurgia, vol. 33, p.
288-305
Badzioch, S. and Hawksley, P. G. W. (1970), Kinetics of thermal decomposition of
pulverized coal particles, Industrial and Engineering Chemistry Process Design
and Development, vol. 9, p. 520-530.
Basu, P., Roa, S. and Dhungana, A. (2010), An investigation into the effect of biomass
particle size on its torrefaction, The Canadian Journal of Chemical Engineering,
vol. 91, p. 466-474.
Bergman, P.C.A. (2005), Combined torrefaction and pelletization the TOP process,
Report ECN-C-05-073, The Netherlands: ECN.
Biswas, A. K. (1981), Principles of blast furnace ironmaking, Cootha Publishing House,
Brisbane, Australia.
Biswas G and Eswaran V. (2002), Turbulent flows: fundamentals, experiments and
modelling, Alpha Science.
Bortz, S. and Flament, G. (1983), Experiments on pulverized-coal combustion under
conditions simulating blast-furnace environments, Ironmaking and Steelmaking,
vol. 10, p. 222-229.
Boubel, R. W., Fox, D. L., Turner, D. B. and Stern, A. C. (1994), Fundamentals of air
pollution, Academic Press, San Diego.
Bridgeman, T.G., Jones, J.M., Shield, I. and Williams, P.T. (2008), Torrefaction of reed
canary grass, wheat straw and willow to enhance solid fuel qualities and
combustion properties, Fuel, vol. 87, p. 844-856.
Burgess, J. M., Jamaluddin, A. S., McCarthy, M. J., Mathieson, J. G., Nomura, M.
Truelove, J. S. and Wall, T. F. (1983), Pulverised coal ignition and combustion in
the blast furnace tuyere zone, Proceedings of Joint Symposium of ISIJ and AIMM,
Tokyo, Japan, p. 129–141
Buss, W. E., Toll, H. and Worberg, R. (2000), Developments in Cokemaking in Europe,
AISE Steel Technology, vol. 77, p. 47-52.
Card, J. B. A., and Jones, A. R. (1995), A drop tube furnace study of coal combustion
and unburned carbon content using optical techniques, Combustion and Flame, vol.
101, p. 539-547.
256
Carpenter, A. M. and Skorupska, N. M. (1993), Coal combustion – analysis and testing,
IEA Coal Research, London.
Chattopadhyay, K., Isac, M. and Guthrie, R. I. L. (2010), Review: Applications of
Computational Fluid Dynamics (CFD) in iron- and steelmaking: Part 1,
Ironmaking and Steelmaking, vol. 37, p. 554-561.
Chen, C., Horio, M., Kojima, T. (2000a), Numerical simulation of entrained flow coal
gasifiers. Part I: modeling of coal gasification in an entrained flow gasifier,
Chemical Engineering Science, vol. 55, p. 3861-74.
Chen, C., Horio, M. and Kojima, T. (2000b), Numerical simulation of entrained flow
coal gasifiers. Part II: effects of operating conditions on gasifier performance,
Chemical Engineering Science, vol. 55, p. 3875-83.
Chen W. H., Du, S. W., Yang, H. H. and Wu, J. S. (2008), Formation characteristics of
aerosol particles from pulverized coal pyrolysis in high-temperature environments,
Journal of Air and Waste Management Association, vol. 58, p. 702-710.
Chen, W. H., Du, S. W., Tsai, C. H., and Wang, Z. Y. (2012), Torrefied biomasses in a
drop tube furnace to evaluate their utility in blast furnaces, Bioresource
Technology, vol. 111, p. 433-438.
Chen W. H., Du, S. W. and Yang, T. H. (2007), Volatile release and particle formation
characteristics of injected pulverized coal in blast furnace, Energy Conversion and
Management, vol. 48, p. 2025-33.
Chen, W. H., Du, S. W., Yang, H. H., Wu, J. S. (2008), Formation characteristics of
aerosol particles from pulverized coal pyrolysis in high-temperature environments,
Journal of the Air & Waste Management Association, vol. 58, p. 702-710.
Chen, W. H. and Jiang T. L. (2000), Double, triple, and tetra-brachial flame structures
around a pair of droplets in tandem, Combustion Science and Technology, vol.
151, p. 105-132.
Chen, W. H. and Kuo, P. C. (2010), A study on torrefaction of various biomass
materials and its impact on lignocellulosic structure simulated by a
thermogravimetry, Energy, vol. 35, p. 2580-2586.
Chen. W. H., Lin, M. R., Leu, T. S. and Du, S. W. (2011), An evaluation of hydrogen
production from the perspective of using blast furnace gas and coke oven gas as
feedstocks, International Journal of Hydrogen Energy, vol. 36, p. 11727-37.
257
Chen, W. H., Lu, K. M., Lee, W. J., Liu, S. H. and Lin, T.C. (2013), Biomass
torrefaction characteristics in inert and oxidative atmospheres at various superficial
velocities. Bioresource Technology, vol. 146, p. 152-160.
Chen, W. H., Tu, Y.J. and Sheng, H. K. (2010), Impact of dilute acid pretreatment on
the structure of bagasse for producing bioethanol, International Journal of Energy
Research, vol. 34, p. 265-274.
Chen, W. H., and Wu, J. S. (2009), An evaluation on rice husk and pulverized coal
blends using a drop tube furnace and a thermogravimetric analyzer for application
to a blast furnace, Energy, vol. 34, p. 1458-1466.
Choi, YC., Li, X.Y., Park, T.J., Kim, J.H. and Lee, J.G. (2001), Numerical study on the
coal gasification characteristics in an entrained flow coal gasifier, Fuel, vol. 80, p.
2193-2201.
Chung, J. K. and Hur, N. S. (1997), Tuyere level coke characteristics in blast furnace
with pulverized coal injection. ISIJ International, vol. 37, p. 119-125.
Couhert, C., Salvador, S. and Commandre, J. M. (2009), Impact of torrefaction on
syngas production from wood, Fuel, vol. 88, p. 2286-2290.
de Lassat de Pressigny, Y., Picard, M., Prado, G., Aleboyeh, H. and Simonin, O. (1990),
Study of coal combustion with respect to blast furnace injection, Proceedings of
Ironmaking Conference, Iron and Steel Society, Detroit, USA, p. 473-480.
Du, S. W. (2001), Development of 3D coal combustion model for PCI operation, CSC’s
internal research report, PJ-90043.
Du, S. W. (2011), Modelling of raceway shape prediction and oxy-coal lance
evaluation, CSC’s internal research report, PJ-100050.
Du, S. W. and Chen, W. H. (2006), Numerical prediction and practical improvement of
pulverized coal combustion in blast furnace, International Communications in Heat
and Mass Transfer, vol. 33, p. 327-334.
Du, S. W., Chen, W. H. and Lucas, A. J. (2007), Performances of pulverized coal
injection in blowpipe and tuyere at various operational conditions, Energy
Conversion and Management, vol. 48, p. 2969-78.
Du, S. W., Chen, W. H. and Lucas, A. J. (2010), Pulverized coal burnout in blast
furnace simulated by a drop tube furnace, Energy, vol. 35, p. 576-581.
258
Du, S. W., Chen, W. H. and Lucas, A. J. (2014), Pretreatment of biomass by
torrefaction and carbonization for coal blend used in pulverized coal injection,
Bioresouce Technology, vol. 161, p. 333-339.
Du, S. W., Ho, C. K., Tsai, S. T. and Yeh, C. M. (2001), Development of pulverised
coal injection lance at China Steel Corporation, World Coal, vol. 10, p. 39-42.
Du, S. W., Lu, C. Y., C. Y. and Chou, C. S. (2012), Development and application of
blast furnace burden profile measurement system at CSC, Proceedings of 4th
Scanmet: International Conference on Process Development in Iron and
Steelmaking, Luleå, Sweden, p. 320-329
Du, S. W. and Tsai, C. H. (2014), Investigation into the formation of the raceway of
blast furnace, CSC’s internal research report, RE-102018.
Du, S. W., Yeh, C. M., Ho, C. K. and Yang, M. K. (2004), Practice of high productivity
at No.3 blast furnace of China Steel Corporation, Proceedings of AISTech
Conference, Tennessee, USA, p. 195-204.
Du, S. W., Yeh, C. P., Chen, W. H., Tsai, C. H. and Lucas, J. A. (2015), Burning
characteristics of pulverized coal within blast furnace raceway at various injection
operations and ways of oxygen enrichment, Fuel, vol. 143, p. 98-106.
Eghlimi, A. and Sahajwalla, V. (1997), Pulverized coal combustion in a 2D
furnace-generalised finite rate chemistry versus mixture fraction/PDF model,
Proceedings of International Conference Fluid and Thermal Energy Conversion,
Indonesia, p. 25-30.
Ergun, S. (1952), Fluid flow through packed columns, Chemical Engineering and
Processing, vol. 48, p. 89-94.
Field, M. A. (1969), Rate of combustion of size–graded fractions of char from a low
rank coal between 1200 K – 2000 K, Combustion and Flame, vol. 13, p. 237-252.
Fletcher, T. H., Ma, J., Rigby, J. R., Brown, A. L. and Webb, B. W. (1997), Soot in coal
combustion system, Progress in Energy and Combustion Science, vol. 23, p.
283-301.
Gao, Y. H. and Bian, L. T. (2013), A evaluation on pulverized coal combustion
properties using a thermogravimetric analyze method, Applied Mechanics and
Materials, p. 423-426, p. 609-613.
259
Gathergood, D. S. and Jukes, M. H. (1996), Blast furnace injection of granular coal,
Proceedings of Injection Technology in Ironmaking and Steelmaking, Brussels,
Belgium, p. 89-107.
Geerds, M., van Laarm R. and Vaynshteyn, R. (2011), Low-cost hot metal: the future of
blast furnace of blast furnace ironmaking, Iron and Steel Technology, vol. 8, p.
51-57.
Gibb, J. (1985), Combustion of residual char remaining after devolatilization. Lecture at
Course of Pulverised Coal Combustion, Imperial College, London.
Gibbins, J. R., Man, C. K. and Pendlebury, K. J. (1993), Determination of rapid heating
volatile matter contents as a routine test, Combustion Science and Technology, vol.
93, p. 349-361.
Gidaspow, D. (1994), Multiphase flow and fluidization: continuum and kinetic theory
descriptions, Academic Press, New York.
Goto, K., Murai, R. Murai, A., Murao, A. Sato, M. Asanuma M. and Ariyama, T.
(2002), Massive combustion technology of solid fuels injected into blast furnace,
Proceedings of International BF Lower Zone Symposium, Wollongong, Australia,
p. 1-10.
Greuel, M., Hillnhütter, F. W., Kister, H. and Kruger, B. (1974), Investigation on
movement mechanisms in front of the tuyere of an industrial blast furnace using an
endoscope, Stahl u. Eisen, vol. 94, p. 533-538.
Gu M. Y., Zhang M. H., Selvarasu, N. K. C., Zhao, Y. F. and Zhou, C. Q. (2008),
Numerical analysis of pulverized coal combustion inside tuyere and raceway, Steel
Research International, vol. 79, p. 17-24.
Gudenau, H. W. and Kiesler, R. (1991), Iron and steel making, especially pulverized
coal injection into the blast furnace, Proceedings of the Symposium on Raceway
Control for Optimum Blast Furnace Performance, McMaster University, Canada,
p. 1-23
Gudenau, H. W., Peters, M. and Joksch, M. (1994), Meβtechnische untersuchungen zur
kohleeinblasung am hochofen, Stahl and Eisen, vol. 114, p.81-86.
Gunn, D. J. (1978), Transfer of heat or mass to particles in fixed and fluidized beds,
International Journal of Heat and Mass Transfer, vol. 21, p. 467–476.
260
Guo B. Y., Zulli P., Rogers, H., Mathieson J. and Yu A. B. (2005), Three-dimensional
simulation of flow and combustion for pulverised coal injection, ISIJ International,
vol. 45, p. 1272-1281.
Guo, H. W., Su, B. X., Zhang, J. L., Shao, J. G., Zuo, H. B. and Ren, S. (2012), Energy
conservation for granular coal injection into a blast furnace, JOM, vol. 64, p.
1002-10.
Guo, X. Z., Wei, G, Zhou, Y. P., Ma, Z. F., Ding, Z. M. and Shen, F. M. (2011), Effect
of MgO addition on burnout rate of pulverized coal for blast furnace injection, Iron
and Steel, vol. 46, p. 7-10.
Guo, Y. C., Chan, C. K. and Lau, K. S. (2003), Numerical studies of pulverized coal
combustion in a tubular coal combustor with slanted oxygen jet, Fuel, vol. 82, p.
893-907.
Gustavo B. Leite, G. B., Abdelaziz, A. E. M., Patrick C. and Hallenbeck, P.C. (2013),
Algal biofuels: challenges and opportunities, Bioresource Technology, vol. 145, p.
134-141.
Hammond, G. P. and Norman, J. B. (2014), Heat recovery opportunities in UK industry,
Applied Energy, vol. 116, p. 387–397.
Hartig, W., Zewe, H., Leyser, P., Cano, S. and Cortina, C., Mahowald, P. and Muller, B.
(2011), Increase of pulverized coal injection at ROGESA, Proceedings of AISTech,
Warrendale, USA, p. 475-486.
Haywood, R. J., Truelove, J. S. and McCarthy, M. J. (1994) Modelling of pulverized
coal injection and combustion in blast furnaces, Proceedings of Ironmaking
Conference, Iron and Steel Society, Chicago, USA, p. 437-442.
He, J. C., Kuwabara, M. and Muchi, I. (1986), Analysis of combustion zone in raceway
under operation of pulverized coal injection, Tetsu to Hagane, vol. 72, p. 1847-54.
Hinrichs, R.A. and Kleinbach, M. (2002), Energy: Its use and the environment (3rd
Ed.),
Harcourt, Inc., Philadelphia.
Ho, C. K. (2000), Development of AE sensor system for measuring burden falling point
in blast furnace, CSC’s internal research report, PJ-89020.
Ho, C. Y. and Du, S. W. (2008), Development of tuyere monitoring system, CSC’s
internal research report, PJ-97075.
261
Hsieh, L. H., Kao, T. S., Liu, K. C. and Wang, G. C. (2002), Investigation of low fluxes
in iron ore sintering and its application, Proceedings of Ironmaking Conference,
Tennessee, USA, p. 721-730.
Hurt, R. and Davis, K. (1999), Percolative fragmentation and spontaneous
agglomeration, Combustion and Flame, vol. 116, p. 662-670.
Hutny, W. P., Lee, G. K. LEE and Price, J. T. (1991), Fundamentals of coal combustion
during injection into a blast furnace, Progress in Energy and Combustion Science,
vol. 17, p. 373-395
Inatani, T., Fukutake, T. and Okabe, K. (1973), Investigations on the coke combustion
and coke size in the raceway, Der Hochofenprozess Hersg., Verein Deutscher
Eisenhüttenletute, Düsserdorf, p. 114-121.
Inatani, T., Okabe, K., Nishiyama, T., Serizawa, Y., Takahashi, H. and Saino M. (1976),
Heavy oil combustion in blowpipe and tuyere, Tetsu to Hagane, vol. 62, p.
514-524.
Ishii, K. (2000), Advanced pulverised coal injection technology and blast furnace
operation, Oxford, UK, Elsevier.
Iwanaga, Y. (1991), Gasification rate analysis of unburnt pulverized coal in blast
furnace, ISIJ International, vol. 31, p. 500-504.
Jaffarullah, R. and Ghosh, B. (2005), Alternate fuel in blast furnace to reduce coke
consumption, I.E (I) Journal-MM, vol. 86, p. 16-23.
Jamaluddin, A. S., Wall T. F., and Truelove, J. S. (1986), Mathematical modeling of
combustion in blast furnace raceways, including injection of pulverized coal,
ironmaking and steelmaking, vol. 13, p. 91-99.
Jamaluddin, A.S., Wall T.F., Truelove, J.S. (1987), Modelling of high intensity
combustion of pulverized coal in a tubular combustor, Combustion Science and
Technology, vol. 55, p. 89-113.
Jianwei, Y., Guolong, S., Cunjiang, K., Tianjun, Y. (2003), Oxygen blast furnace and
combined cycle (OBF-CC) - an efficient iron-making and power generation
process, Energy, vol. 28, p. 825-835.
Jones, W.P. and Whitelaw, J.H. (1982), Calculation methods for reacting turbulent
flows: A review. Combustion and Flame, vol. 48, p. 1-26.
262
Kalkreuth, W., Borrego, A. G., Alvarez, D., Menendez, R., Osorio, E., Ribas, M.,
Vilela, A., and Cardozo Alves, T. (2005), Exploring the possibilities of using
Brazilian subbituminous coals for blast furnace pulverized fuel injection, Fuel, vol.
84, p. 763-772
Khalil, E.E. (1982), Modelling of Furnace and Combustor, Abacus Press.
Kim, S. M., Chung, J. K. and Cho, C. M. (1996), Improvement coal combustion
efficiency for high injection rate and its effect on raceway behavior, POSCO
Technical Report, vol. 1, p. 110-114.
Kobayashi, H., Howard, J. B. and Sarofim, A. F. (1977), Coal devolatilization at high
temperatures, Proceeding of 16th Symposium (International) on Combustion, The
Combustion Institute, Pittsburgh, PA, p. 411-425.
Kuo, K. K. (1986), Principles of combustion, John Wiley and Sons, New York.
Kurose, R., Ikeda, M., Makino, H., Kimoto, M. and Miyazaki, T. (2004), Pulverized
coal combustion characteristics of high-fuel-ratio coals, Fuel, vol. 83, p. 1777-85.
Kuwabara, M., Hsieh, Y. S., Isobe, K. and Muchi, I. (1981), Mathematical modelling of
the tuyere combustion zone of the blast furnace, Proceedings of International Blast
Furnace Hearth and Raceway Symposium, Newcastle, Australia, P. 7.1-7.7.
Lai, F. C., Chang, D. Y. and Chen, E. K. (1994), Promotion of PCI rate in China Steel
Corporation’s No. 1 blast furnace, Proceedings of the First International Congress
Science and Technology of Ironmaking, p. 541-546.
Lee, J. G., Kim, J. H., Lee, H. J., Park, T. J. and Kim, S. D., 1996; Characteristics of
entrained flow coal gasification in a drop tube reactor, Fuel, vol. 75, p. 1035-42.
Li, H. Y., Elliott, L. Rogers, H. and Wall, T. (2014), Comparative study on the
combustion performance of coals on a pilot-scale test rig simulating blast furnace
pulverized coal injection and a lab-scale drop-tube furnace, Energy and Fuels, vol.
28, p. 363-368.
Li, R. R., Zhu, J. M. and Li, Y. Q. (2007), Practice of raising Baosteel No. 4 BF’s coal
injection rate rapidly, Baosteel Technology, No. 1, p. 11-14.
263
Lu, K. M., Lee, W. J., Chen, W. H. and Lin, T. C. (2013), Thermogravimetric analysis
and kinetics of co-pyrolysis of torrefied wood and coal blends. Applied Energy,
vol. 105, p. 57–65.
Lu, C. Y. and Du, S. W. (2010), Development of an online blast furnace burden profile
measuring system, China Steel Technical Report, no. 23, p. 25-30.
Lu, K. M., Lee, W. J., Chen, W. H., Liu, S. H. and Lin, T. C. (2012), Torrefaction and
low temperature carbonization of oil palm fiber and eucalyptus under nitrogen and
air atmospheres, Bioresource Technology, vol. 123, p. 98-105.
Machado, J. G. M. S., Osorio, E., Vilela, A. C. F., Babich, A., Senk, D. and Gudenau,
H. W. (2010), Reactivity and conversion behaviour of Brazilian and imported
coals, charcoal and blends in view of their injection into blast furnaces, Steel
Research International, vol. 81, p. 9-16.
Magnussen, B. F. and Hjertager, B. W. (1976), On mathematical modelling of turbulent
combustion with special emphasis on soot formation and combustion, Proceedings
of 16th
Symposium (International) on Combustion. Pittsburgh, The combustion
institute.
Malgarini, M. G. (1991), Direct use of coal in ironmaking, Proceedings of World Coal
Institute Conference and Exhibition: Coal in the Environment, London, p. 367-368.
Maloney D. J. and Jekins, R. G. (1984), Coupled heat and mass transport and chemical
kinetic rate limitations during coal rapid pyrolysis, Proceedings of 20th
Symposium
(International) on Combustion, The Combustion Institute, Pittsburgh, PA, p.
1435-43.
Maki, A., Sakai, A., Takagaki, N., Mori, K., Ariyama, T., Sato, M. and Murai, R.
(1996), High rate coal injection of 218 kg/t at Fukuyama No. 4 blast furnace, ISIJ
International, vol. 36, p. 650-657.
Mathieson, J. G., Rogers, H. Somerville, M. A. and Jahanshahi, S. (2012), Reducing net
CO2 emissions using charcoal as a blast furnace tuyere injectant, ISIJ international,
vol. 52, p. 1489-96.
Mathieson, J. G., Truelove, J. and Rogers, H. (2005) Towards an understanding of coal
combustion in blast furnace tuyere injection, Fuel, vol. p. 1229–37.
McCarthy, M. J. (1986), Replacement of oil by coal injection at the blast furnaces, BHP
Research Report, NERDDP/EG/86/602.
264
Mondal, S. S., Som, S. K. and Dash, S. K. (2005), Numerical predictions on the
influences of the air blast velocity, initial bed porosity and bed height on the shape
and size of raceway zone in a blast furnace, Journal of Physics D–Applied Physics,
vol. 38, p.1301–07.
Morsi, S. A. and Alexander A. J. (1972), An investigation of particle trajectories in
two–phase flow system, Journal of Fluid Mechanics, vol. 55, p. 193–208.
Myers, G.E. (1971), Analytical Methods in Conduction Heat Transfer, McGraw-Hill,
New York.
Nakamura, N., Togino, Y. and Tateoka, M. (1978), Behavior of coke in large blast
furnaces, Ironmaking and Steelmaking, vol. 5, p. 1–17.
Narita, K. Maekawa, M. Kanayama, H. Seki, Y. And Saito, T. (1982), Combustion of
pulverised coal in experimental furnace, Tetsu-to-Hagane, vol. 68, p. 2385-92.
Niksa, S. Heyd, L. E., Russel, W. G. and Saville, D. (1984), On the role of heating rate
in rapid coal devolatilization, Proceedings of 20th
Symposium (International) on
Combustion. The Combustion Institute, Pittsburgh, PA, p. 1445-53.
Niksa, S. and Lau, C. W. (1993), Global rates of devolatilization for various coal types,
Combustion and Flame, vol. 94, p. 293-307.
Nogami, H., Miura, T. and Furukawa, T. (1992), Simulation of transport phenomena
around raceway zone in the lower part of blast furnace, Tetsu-to-Hagane, vol. 78,
p. 1222-29.
Nogami, H., Yamaoka G. and Takatani, K. (2004), Raceway design for the innovative
blast furnace, ISIJ International, vol. 44, p. 2150-58
Nolde, H. D. Peters, C. and Wagner, E. D. (1996), Blast furnace coal injection,
Proceedings Blast Furnace Injection Symposium, Ohio, USA, p. 33-46
Ohno, Y., Furukawa, T. and Matsu-Ura, M. (1994), Combustion behaviour of
pulverized coal in a raceway cavity of blast furnace and its application to a large
amount injection. ISIJ International, vol. 34, p. 641-648.
Osório, E., Gomes, M. D. I., Vilela, A. C. F., Kalkreuth, W., de Almeida, M. A. A.,
Borrego, A. G., Alvarez, D. (2006), Evaluation of petrology and reactivity of coal
blends for use in pulverized coal injection (PCI), International Journal of Coal
Geology, vol. 68, p. 14–29.
265
Ostrowski, E. J. (1983), Factors influencing optimization of blast-furnace coal injection,
Ironmaking and Steelmaking, vol. 10, p. 215-221.
Parikh, J., Channiwala, S. A. and Ghosal, G. H. (2005), A correlation for calculating
HHV from proximate analysis of solid fuels. Fuel, vol. 84, p. 487–494.
Peng, J. H., Bi, X. T., Lim, C. J. and Sokhansanj, S. (2013). Study on density, hardness,
and moisture uptake of torrefied wood pellets, Energy Fuel, vol. 27, p. 967–974.
Peng, J. H., Bi, X. T., Sokhansanj, S. and Lim, C. J. (2012), A study of particle size
effect on biomass torrefaction and densification, Energy Fuel, vol. 26, p.
3826-3839.
Peng, J. H., Bi, X. T., Sokhansanj, S., Lim, C. J. and Melin S. (2010), An economical
and market analysis of Canadian wood pellets, International Journal of Green
Energy, vol. 7, p. 128-142.
Perlov, N. I. (1987), Technological approaches to energy saving in blast-furnace
operations in the iron and steel industry of the U.S.S.R, Energy, vol. 12, p.
1177-81.
Peters, M. and Lüngen H.B. (2009), Iron Making in Western Europe, Proceedings of the
5th International Congress on the Science and Technology of Ironmaking,
Shanghai, China, p. 22-28.
Peters, K. H., Beppler, E., Korthas, B. and Peters, M. (1991), Effect of high levels of
coal injection on blast furnace operation, Proceedings of 2nd European Ironmaking
Congress, Glasgow, U.K, p. 247-262.
Phanphanich, M.and Mani, S. (2010), Impact of torrefaction on the grindability and fuel
characteristics of forest biomass, Bioresource Technology, vol. 102, p. 1246-1253.
Picard, M. (2001), Pulverized coal combustion in the raceway, using a 3D numerical
simulation, Proceedings of Ironmaking Conference, Maryland, US, p. 229-239.
Picard, M., Bolsigner, P. J., Succurro, A., Eymond, J. L. and Valdan, G. (2000),
Development of injection lances in order to improve coal ignition inside the tuyere,
La Revue de Metallurgie-CIT, 36, p. 21-28.
Pimchuai, A., Dutta, A. and Basu, P. (2010), Torrefaction of agriculture residue to
enhance combustible properties, Energy Fuels, vol. 24, p. 4638-4645.
Poos, A. and Ponghis, N. (1990), Potentials and problems of high coal injection rates,
Proceedings of Ironmaking Conference, Michigan, USA, p. 443-453.
266
Porzio, G. F., Fornai, B., Amato, A., Matarese, N., Vannucci, M., Chiappelli, L., and
Colla, V. (2013), Reducing the energy consumption and CO2 emissions of energy
intensive industries through decision support systems – an example of application
to the steel industry, Applied Energy, vol. 112, p. 818–33.
Ranz, W. E. and Marshall, W. R. (1952a), Evaporation from drops, part I, Chemical
Engineering Process, vol. 48, p. 141–146.
Ranz, W. E. and Marshall, W. R. (1952b), Evaporation from drops, part II, Chemical
Engineering Process, vol. 48, p. 173-180.
Rousset, P., Aguiar, C., Labbe, N. and Commandre, J. M. (2011), Enhancing the
combustible properties of bamboo by torrefaction, Bioresource Technology, vol.
102, p. 8225-8231.
Sabil, K.M., Aziz, M.A., Lal, B. and Uemura, Y. (2014), Synthetic indicator on the
severity of torrefaction of oil palm biomass residues through mass loss
measurement, Applied Energy, vol. 114, p. 104-113.
Sami, M., Annamalai, M., and Wooldridge, M. (2001), Co-firing of coal and biomass
blends, Progress in Energy and Combustion Science, vol. 27, p. 171-214.
Sato, M., Ariyama, T., Yamakawa, Y. and Suzuki, M. (1994), Combustion mechanism
of pulverized coal injected into the blast furnace and its mathematical simulation,
Proceedings of ICSTI ’94, Iron and Steel Institute of Japan, Sendai, p. 511-516.
Sato, M., Murai, R. and Ariyama, T. (1996), Development of one-dimensional
mathematical model for pulverized coal combustion considering particle
dispersion, Tetsu to Hagane, vol. 82, p. 13-18.
Sato, M., Murai, R, Ariyama, T., Maki, A., Shimomura, A., and Mori, K. (1998),
Development of injection lance with high combustibility for high rate coal
injection, Tetsu to Hagane, vol. 84, p. 37-42.
Scaife, P. H., Mathieson, J. G., Rogers, H. and Nomura, S. (1983), Replacement of oil
by coal injection at the blast furnace, BHP Research Report, NERDDP/EG/84/268.
Scaife, P. H., Mathieson, J. G., Rogers, H. and Nomura, S. (1983), Replacement of oil
by coal injection at the blast furnace, BHP Research Report, NERDDP/EG/84/268
267
Schott, R. (2012), State-of-the-art PCI technology for blast furnace ensured by
continuous technological and economical improvement, Proceedings of the Iron
and Steel Technology Conference, Georgia, USA, P. 589-604.
Seeker, W. R., Samuelsen, G. S., Heap M. P. and Trolinger, J. D. (1981), The thermal
decomposition of pulverized coal particles, Proceedings of 18th
Symposium
(International) on Combustion, The Combustion Institute, Pittsburgh, PA, p.
1213-26.
Selvarasu, N. K., Huang, D., Chen, Z. M., Gu, M. Y., Chaubal, P. and Zhou, C. Q,
(2006), Prediction of raceway in a blast furnace, Proceedings of ASME 2006
International Mechanical Engineering Congress and Exposition, Illinois, USA, p.
297-303.
Shampine, R. W., Cohen, R. D., Bayazitoglu, Y. and Anderson, C. F. (1995), Eeffect of
agglomeration on pulverized-coal combustion, Combustion and Flame, vol. 101, p.
185-191.
Shen, Y. S., Guo, B. Y., Yu, A. B., Austin, P. R. and Zulli, P. (2011), Three
dimensional modelling of in-furnace coal/coke combustion in a blast furnace, Fuel,
vol. 90, p. 728-738.
Shen, Y. S., Guo, B. Y., Yu, A. B., Maldonado, D., Austin, P. and Zulli, P. (2008),
Three-dimensional modelling of coal combustion in blast furnace, ISIJ
International, vol. 48, p.777-786.
Shen, Y. S., Guo, B. Y., Yu, A. B. and Zulli, P. (2009a), A three-dimensional numerical
study of the combustion of coal blends in blast, Fuel, vol. 88, p. 255-263.
Shen, Y. S., Guo, B. Y., Yu, A. B. and Zulli, P. (2009b), Model study of the effects of
coal properties and blast conditions on pulverized coal combustion, ISIJ
International, vol. 49, p. 819-826.
Shen, Y. S., Maldonado, D., Guo, B. Y., Yu, A. B., Austin, P. and Zulli, P (2009c),
Computational fluid dynamics study of pulverized coal combustion in blast furnace
raceway, Industrial and Engineering Chemistry Research, vol. 48, p. 10314–23.
Shuit, S. H., Tan, K. T., Lee, K. T. and Kamaruddin, A.H. (2009), Oil palm biomass as
a sustainable energy source: A Malaysian case study, Energy, vol. 34, p.
1225-1235.
Singer, S. (1984), Pulverized coal combustion: recent developments, Noyes
Publications, New Jersey.
268
Sivathanu, Y. R. and Faeth, G. M. (1990), Generalized state relationships for scalar
properties in non-premixed hydrocarbon/air flames, Combustion and Flame, vol.
82, p. 211-230.
Smith, I. W. (1982), The combustion rates of coal char: A review, Proceedings of
Nineteenth Symposium (International) on combustion, Pittsburgh, The combustion
institute, p. 1045-65.
Smoot, L. D. and Pratt, D. T. (1979), Pulverized coal combustion and gasification, New
York, London, Plenum.
Smoot, L. D. and Smith P. J. (1985) Coal combustion and gasification, New York and
London, Plenum Press.
Spalding, D. B. (1971), Mixing and chemical reaction in steady confined turbulent
flames, Proceeding of 13th Symposium (International) on Combustion, Pittsburgh,
The Combustion Institute, p. 649-657.
Steeghs, A. G. S. (1992), Determination of coal combustion under simulated blast
furnace raceway conditions, Technical Steel Research Report of European
Commission.
Steeghs A. G. S., Schoone E. E. and Toxopeus H. L. (1996), High injection rate of coal
into the blast furnaces of Hoogonvens IJmuiden, Proceedings of Injection
Technology in Ironmaking and Steelmaking, Brussels, Belgium, p. 75-87.
Steiler J. M., Lao D. and Lebonvallet J. L. (1996), Development of coal injection in the
blast furnace at Usinor Sacilor, Proceedings of Injection Technology in Ironmaking
and Steelmaking, Brussels, Belgium, p.15-32.
Suopajärvi, H., Pongrácz, E. and Fabritius, T. (2014), Bioreducer use in Finnish blast
furnace ironmaking – Analysis of CO2 emission reduction potential and mitigation
cost, Applied Energy, vol. 124, p. 82-93.
Suzuki, T., Hirose, R. Morimoto, K. and Abe, T. (1984), High intensity combustion of
coal for application to a blast furnace, Proceedings of 20th
Symposium
(International) on Combustion, University of Michigan, Michigan, p. 1419-25.
Suzuki, T., Smoot, L. D., Fletcher, T. H. and Smith, P. J. (1986), Prediction of
high-Intensity pulverized coal combustion, Combustion Science and Technology,
vol. 45, p. 167-183.
269
Suzuki, T, Uehara, T. and Akedo, H. (1990), Combustion characteristics of pulverized
coal for blast furnace coal injection, Proceedings of Ironmaking Conference, Iron
and Steel Society, Warrendale, PA, USA, p. 465-471.
Tate, A. (1993), The Variation of Char Reactivity during the Combustion of Pulverized
Coal, Ph.D. Thesis, The University of Newcastle.
Takeda, K. (1994), Mathematical modelling of pulverised coal combustion in a blast
furnace, Ph.D thesis, Department of Mechanical Engineering, The University of
London, England.
Takahashi, H., Kawai, H. and Suzuki, Y. (2002), Analysis of stress and buoyancy for
solids flow in the low part of a blast furnace, Chemical Engineering Science, vol.
57, p. 215-226.
Takeda, K. and Lockwood F. C. (1997), Integrated mathematical model of pulverised
coal combustion in a blast furnace, ISIJ International, 1997, vol. 37, p. 432-440.
Takeda, K., Miyagawa, S. and Taguchi, S. (1990), Effect of coal properties on
combustibility of coal injected to blast furnace, Proceedings of Ironmaking
Conference, Iron and Steel Society, Detroit, USA, p. 455-465.
Toporov, D. (2014), Combustion of pulverised coal in a mixture of oxygen and recycled
flue gas, Esvier Science, London.
Toyoda, S. (1983), Changes of the use of energy in Japanese steel industry,
Transactions of ISIJ, vol. 23, p. 1-13
Ubhayakar, K., Stickler, D. B, Rosenberg, C. W. V. and Ganon, R. (1976), Rapid
devolatilization of pulverized coal in hot combustion gases, Proceedings of 18th
Symposium (International) on Combustion, The Combustion Institute, Pittsburgh,
p. 427–436.
Ueno, H., Yamaguchi, K and Tamura, K. (1993), Coal combustion in the raceway and
tuyere of a blast furnace, ISIJ international, vol. 33, p. 640-645.
Vamvuka, D., Schwanekamp, G. and Gudenau, W. (1996), Combustion of pulverized
coal with additives under conditions simulating blast furnace injection, Fuel, vol.
75, p. 1145-50.
van der Stelt, M. J. C., Gerhauser, H., Kiel, J. H. A. and Ptasinski, K. J. (2011), Biomass
upgrading torrefaction for the production of biofuels: a review. Biomass and
Bioenergy, vol. 35, p. 3748-62.
270
Wall, T. F., Liu, G. S., Wu, H. W., Roberts, D. G., Benfell, K. E., Gupta, S, Lucas, J. A.
and Harris, D. J. (2002), The effects of pressure on coal reactions during pulverised
coal combustion and gasification, 'Progress in Energy and Combustion Science,
vol. 28, p. 405-433.
Wakimoto, H. Sato, K., Fujiura, M. and Hara, H. (1983), Combustion characteristics of
pulverized coal in pressurized state, Tetsu-to-Hagane, vol. 69, p. S105.
Wang, C., Ryman, C. and Dahl, J. (2009), Potential CO2 emission reduction for
BF-BOF steelmaking based on optimised use of ferrous burden materials,
International Journal of Greenhouse Gas Control, vol. 3, p. 29-38.
Wei, S. M. and Qi, Z. D. (1983), Coal-powder injection on blast furnace of Shoudu Iron
and Steel Co., Ironmaking and Steelmaking, vol. 10, p. 109-113.
Wijayanta, A. T., Alam, M. S., Nakaso, K., Fukai, J., Kunitomo, K. and Shimizu, M.
(2014), Combustibility of biochar injected into the raceway of a blast furnace, Fuel
Processing Technology, vol. 117, p. 53-59.
Wikström, J. O., Sköld, B. E. and Karsrud, K. (1996), SSAB/MEFOS oxy-coal system –
3 Years of industrial experience, Proceedings of Ironmaking, Iron and Steel
Society, Pittsburgh, p. 24-27.
Willmer, R. R. (1989), The effect of blast furnace coal injection upon bosh coke
properties, coke combustion and furnace permeability, Proceedings of Ironmaking
Conference, Chicago, USA, p. 359-401.
Wu, J. B. and Hsieh, J. S., (2003), The development and result of one bit opening
method at CSC blast furnace”, Mining and Metallurgy, vol. 47, No. 3, p.62-67.
Yamagata, C, Suyama S., Horisaka, S., Takatani, K., Kajiwara, Y., Komatsu, S.,
Shibuta, H. and Aminaga, Y. (1992), Fundamental study on pulverized coal
injected into coke bed at high rate, ISIJ International, vol. 32, p. 725-732
Yamaguchi, K., Ueno, H. and Tamura K. (1992), Maximum injection rate of pulverized
coal into blast furnace through tuyeres with consideration of unburnt char, ISIJ
International, vol. 32, p. 716-724.
Yan, B. H., Cao, C. X., Cheng, Y., Jin, Y and Cheng, Y. (2014), Experimental
investigation on coal devolatilization at high temperatures with different heating
rates, Fuel, vol. 117, p. 1215-22.
271
Yeh, C. M., Ho, C. K., Du, S. W., Chou, C. S. and Hsu, C. H. (2002), Development of
pulverized coal injection lance with air cooling in No. 3 blast furnace of CSC,
Proceedings of 61st Ironmaking Conference, Tennessee, USA, p. 267-276.
Yeh, C. P. Du, S. W., Tsai, C. H. and Yang, R. J. (2012), Numerical analysis of flow
and combustion behavior in tuyere and raceway of blast furnace Fueled with
pulverized coal and recycled top gas”, Energy, vol. 42, p. 233-240.
Yu, J. L., Lucas, J., Strezov, V. and Wall, T. (2003), Swelling and char structures from
density fractions of pulverized coal, Fuel, vol. 82, p. 1977-87.
Yu, Q. Y. (1999), Experimental research on the combustion efficiency of blended coal
injection into BF, Baosteel Technical Report, vol. 1, p. 33-36.
Zhang, W. D., Chen, H., Ma, Z. J. Song, J. L. and Wu, J. L. (2013), Technological
development and competitiveness analysis of large capacity blast furnace,
Proceedings of the Fifth Baosteel Biannual Academic Conference, Shanghai,
China, p. 235-240.
Zhou, L. X. (1993), Theory and numerical modelling of turbulent gas-particle flows and
combustion, CRC Press, Florida, USA.
Zhu, M. and Xu H. Y. (2014), Production practice of keeping No.1 1880m3 BF long
term stable regular working in Laiwu Steel, Shandong Metallurgy, vol. 36, p. 4-6.
Ziebik, A. and Stanek, W. (2001), Forecasting of the energy effects of injecting plastic
wastes into the blast furnace in comparison with other auxiliary fuels, Energy, vol.
26, p. 1159-73.