M. Geerdes, R. Chaigneau, O. Lingiardi, R. Molenaar, R. Van Opbergen, Y. Sha, J. Warren - Modern Blast Furnace Ironmaking - An Introduction-IOS Press (2020)

Modern
Blast Furnace
Ironmaking
an introduction
This page intentionally left blank
Maarten Geerdes
Rénard Chaigneau
Oscar Lingiardi
Ron Molenaar
Rob van Opbergen
Yongzhi Sha
Peter Warren
Modern
Blast Furnace
Ironmaking
an introduction
Fourth Edition, 2020

© 2020 The authors and IOS Press. All rights reserved.
ISBN 978–1–64368–122–1 (print)

ISBN 978–1–64368–123–8 (online)
DOI 10.3233/STAL9781643681238
Published by IOS Press under the imprint Delft University Press
Publisher
IOS Press BV
Nieuwe Hemweg 6b
1013 BG Amsterdam
The Netherlands
tel: +31–20–688 3355
email: info@iospress.nl
www.iospress.nl
LEGAL NOTICE
The publisher is not responsible for the use which might be made of the following information.
PRINTED IN SLOVENIA
v
Preface
This fourth edition of “Modern Blast Furnace Ironmaking” has been prepared
by an international staff of experts from three continents. We are indebted to
Cor van der Vliet and Hisko Toxopeus†, who prepared the first two editions
of the book and agreed to hand over future editions of the book to the next
generation. Prof. Ivan Kurunov† and John Ricketts made a great contribution
to the third edition. Co–authors of the fourth edition are Prof. Yongzhi Sha
from the China Iron and Steel Research Institute, Ron Molenaar, retired
superintendent of the IJmuiden Blast Furnaces and Peter Warren, presently
working at British Steel in the United Kingdom and Rob van Opbergen,
experienced in casthouse operation and operational improvement.
The objective of the book is to share our insights, that optimization of the
blast furnace is not only based on “best practice transfer”, but also requires
conceptual understanding of why a measure works. In other words, operational
improvement is based not only on know–how, but also on know–why.
Compared to the third edition, we have added operational examples of many

furnaces. The examples are for illustration only and cannot be traced to a
specific furnace. Since the earlier edition, new instrumentation like cameras at
tuyeres and the top are playing a much more prominent role. We have rewritten
the sections on balances and elaborated the section on operational challenges.
We are indebted to the many colleagues we have worked with. We are grateful
to Jennifer Wise–Alexander and Tim Vander, who will find part of their
contributions to the second edition in the book. Edo Engel did the editing, as
in the previous three editions. Danieli Corus supported publication of the book.
For the fourth edition we enjoyed support from many colleagues worldwide.
We are indebted to N. Bleijendaal, J. Borrego and L. Castro (AHMSA, Mexico),
P. Etchevarne (Ternium, Argentina), D. Fisher (British Steel, UK), G.J.
Gravemaker, Dr. B. Nightingale (Australia), T. Spiering (fellow traveler), Yuanyi
Liu (China), Dr. Fuming Zhang (China) and P. Zonneveld.
We learn by sharing our knowledge. We wish the same to our readers.
Maarten Geerdes, Rénard Chaigneau, Oscar Lingiardi, Ron Molenaar,

Rob van Opbergen, Yongzhi Sha, Peter Warren
September 2020
vii
Contents
Preface v
List of Symbols and Abbreviations x
Chapter I Introduction to the Blast Furnace Process 1

1.1 Global steel production 1
1.2 Hot metal production 2
1.3 Blast furnace process 3
1.4 Equipment 5
1.5 Book overview 10
Chapter II Blast Furnace Process and Internal Structure 11

2.1 Gasifying coke and coal 11
2.2 Counter–current reactor 12
2.3 Internal layer structure 13
2.4 How gas flows through the furnace 16
2.5 Processing the ferrous burden 17
2.6 Furnace efficiency 18
2.7 Casting hot metal and slag 19
Chapter III The Ferrous Burden 23

3.1 Introduction 23
3.2 Iron ore 24
3.3 Quality demands for the blast furnace burden 26
3.4 Sinter 30
3.5 Pellets 35
3.6 Lump ore 40
3.7 Metallics charge and briquettes 41
3.8 Interaction of burden components 42
3.9 Chemical control of the burden 44
Chapter IV Coke 47
4.1 Function of coke in the blast furnace 47
4.2 Coal blends for coke making 49
4.3 Coke quality 50
4.4 Coke size distribution 54
4.5 Strength of coke 55
4.6 Coke deadman 58
4.7 Overview of international quality parameters 63
4.8 Coke, nut coke and high PCI 65
viii
Chapter V Injection of Coal, Natural Gas and Oil 67

5.1 Properties of coal, oil and gas 68
5.2 Coal injection 70
5.3 Natural gas injection 82
5.4 Coal–gas co–injection 85
Chapter VI Blast Furnace Gas and Processing the Ferrous Burden 87

6.1 Introduction 87
6.2 Hot metal and slag composition 87
6.3 Top gas composition from a one–stage mass balance 89
6.4 The temperature profile and chemical reactions 91
6.5 Carbon balance: estimation of direct reduction 92
6.6 Gas and burden composition during processing 96
6.7 Permeability 102
6.8 Blast furnace efficiency 109
6.9 Productivity 111
6.10 Process symmetry 112
Chapter VII Burden Distribution 117

7.1 The movement of gas and burden 117
7.2 Burden descent: system of vertical forces 121
7.3 Burden distribution and gas flow control 123
7.4 Coke and ore layer thickness 132
7.5 Finding an optimal burden distribution 135
7.6 Double bell furnaces 136
7.7 Permeability of cohesive zone 137
Chapter VIII Hot Metal and Slag 141

8.1 Processing the ferrous burden above the cohesive zone 141
8.2 Softening 143
8.3 Melting 145
8.4 Formation of hot metal and slag as cast 147
8.5 Slag properties 155
Chapter IX Casting Hot Metal and Slag 161

9.1 Introduction 161
9.2 Liquid level and the blast furnace process 162
9.3 Monitoring casting regimes 167
9.4 The taphole 172
9.5 Single taphole furnaces 177
9.6 Casting problems and remedial actions 178
9.7 Hot metal and slag separation 180
ix
Chapter X Control of Blast Furnace Operationand Burden Selection 183

10.1 Production level and reductant rate 183
10.2 Under control and out of control 185
10.3 What to control in daily operations? 186
10.4 Thermal control 186
10.5 Control of slag basicity 187
10.6 Control of hot blast parameters and injectant at tuyere level 188
10.7 Gas flow control 189
10.8 What should be controlled automatically? 190
10.9 Burden selection 190
Chapter XI Operational Challenges 197

11.1 Raw materials from stockyard: fines and moisture 197
11.2 Burden descent 203
11.3 Channeling 207
11.4 Recirculation of alkali and zinc 208
11.5 Adherences to the furnace wall 211
11.6 Tuyere blockage 212
11.7 Tuyeres: failures and water leakage 213
11.8 Stops and starts 217
11.9 Casthouse challenges 224
11.10 Greenhouse gas emissions 226
Annex I Blast Furnace Instrumentation 229

Annex II Reference Furnace Data 233
Annex III Rules of Thumb 234
Annex IV The Blast Furnace as a Chemical Reactor 235
Annex V Expert Systems and Models 242
Annex VI Coke quality tests 248
Annex VII Rist diagram 250
Annex VIII References 255
Index 259
x
List of Symbols and Abbreviations

Δ Delta (difference)
Èta (utilization)
Ω Omega
AMS Average Mean Size
B2, B3, B4 Basicity, ratio of 2, 3 or 4 components
CRI Coke Reactivity Index
CSR Coke Strength after Reaction
daN deca Newton
HGI Hard Grove Index
HMS Harmonic Mean Size
HOSIM Hoogoven Simulation (Blast Furnace Simulation)
HV High Volatile
IV Inner Volume
ISO International Standardization Organization
JIS Japanese Industrial Standard
LV Low Volatile
MV Mid Volatile
PCI Pulverized Coal Injection
RAFT Raceway Adiabatic Flame Temperature
RR Replacement Ratio
Standard Coke Coke with 87.5 % carbon
STP Standard Temperature and Pressure
tHM tonne of hot metal
VM Volatile Matter
WV Working Volume
I Introduction to the
Blast Furnace Process
1.1 Global steel production
Countless steel products are used in nearly every aspect of our daily life. over
the last 100 years, global crude steel production has risen from 28 million
tonnes in 1900 to 1.869 million tonnes in 2019 an increase of 3.6 % per year, as
shown in Figure 1.1.
2000
1800
1600
Production (Mtpa)
1400
1200
1000
800
600
400
200
0
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 2020
Figure 1.1 World crude steel production (World Steel Association)
Two different process routes are available for the production of steel products:
the blast furnace–oxygen steelmaking and the electric arc steelmaking route.
The routes differ with respect to the type of products that can be made, as well
as the raw materials used.
The blast furnace–oxygen steelmaking route mainly produces flat products,

while electric arc steelmaking is more focused on long products. The former
uses coke and coal as the main reductant (fuel) sources and sinter, pellets and
lump ore as the iron–bearing component, while the latter uses electric energy
to melt scrap and DRI (direct reduced iron). The current trend is for electric arc
furnaces to be capable of producing flat products as well. Nevertheless, the blast
furnace–oxygen steelmaking route is the primary source for worldwide steel
production: oxygen steel is about 75 % of total steel production, as shown in
Figure 1.2.
2 Chapter I
2000
1871
1691 DRI
Crude Steel Production and

Open HBI
Steel
Metallic Charge (Mtpa)

1600 Electric (4.2%)
Hearth Scrap
Steel
Steel (32.8%)
(24.9%)
(0.4%)
1200
Corex/Finex
Hot Metal
(0.4%)
800
Oxygen Blast Furnace
Steel Hot Metal
(74.7%) (62.6%)
400
0
Crude Steel Metallic Charge
Production
Figure 1.2 World crude steel production and metallic charge

(Lüngen and Schmöle, 2018)
In total, for steelmaking the main ferrous sources are hot metal from blast
furnaces (63 %) and scrap (33 %). Minor amounts from Corex/Finex (0.4 %)
and some direct reduced iron and hot briquetted iron (DRI/HBI, 4 %).
In iron ore mines the ferrous material is present as oxides. The oxygen of the
oxides is removed by a reduction process, the most important being the blast
furnace process, which is the subject of this book.
1.2 Hot metal production
Hot metal is produced from sinter, pellets and lump ores as the iron–bearing
components, and coke and coal as the main reductants (fuel). Global hot metal
production from the blast furnace process has risen from about 40 million
tonnes in 1900 to 1278 billion tonnes in 2019. Especially during the last two
decades, mainly due to the contribution from China, global blast furnace hot
metal production has increased dramatically, as shown in Figure 1.3.
1400
Global
1200
China
Production (Mtpa)
1000
800
600
400
200
0
1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020
Figure 1.3 Blast Furnace hot metal production (World Steel Association)
Introduction to the Blast Furnace Process 3
Steel production based on hot metal from blast furnaces contributes

considerably to greenhouse gas emissions: some 1500–2000 kg CO2 is
generated per tonne of steel, since the process route from the mine to the final
steel product is largely based on coal for energy. It is about 6 % of total CO2
emissions worldwide. Producing steel from scrap generates only around 25 % of
the amount of CO2, although the amount and quality of scrap cannot replace
the steel produced from “virgin” iron units. This means that blast furnaces
will continue to produce huge amounts annually, while improving resource
efficiency by applying state–of–the–art technologies.
1.3 Blast furnace process
A blast furnace is filled with alternating layers of coke and iron ore–containing
burden. The inputs and outputs of a blast furnace are shown in Figure 1.4.
– Heat and process gas are generated by gasifying coke and coal by means of
preheated hot blast (up to more than 1250 °C).
– In this process, the oxygen in the blast is transformed into gaseous carbon
monoxide. The resulting gas has a high flame temperature of 1900–2300 °C.
– In the process of generating hot gas, coke is consumed, and the heat of the gas is
used to melt the ferrous burden. Voidage is created in this way, and the burden
descends into this voidage. The gas and ferrous burden movement are illustrated
in Figure 1.5 on the next page.
Figure 1.4 Input and output of a blast furnace

4 Chapter I
Figure 1.5 Voidage is created by gasifying coke at the tuyeres and by melting ferrous
burden, which allows gas to ascend and burden to descend
The very hot gas ascends through the furnace, carrying out a number of vital
functions:
– Heats up the coke in the bosh/belly area.
– Melts the iron ore in the burden, creating voidage.
– Heats up the material in the shaft zone of the furnace.
– Removes the oxygen of the iron ore burden by chemical reactions.
The hot gas travels through the blast furnace in 5–10 seconds. During this
period, its composition changes through chemical reaction and its temperature
decreases from 1900–2300 °C to 100–200 °C.
As a consequence of the gas movement, there are large temperature differences

in the blast furnace in vertical and radial directions. A typical example of the
temperature profile in the blast furnace is shown in Figure 1.6, which shows
that the softening/melting zone is located in an area where temperatures are
between 1200 and 1400 °C. The temperature differences in the furnace are
large. In the example, the temperature gradients are larger in the horizontal
direction than in the vertical direction.
Upon melting, the iron ore produces hot metal and slag, which drips down
through the coke into the hearth, where it is removed by casting through the
taphole.
Figure 1.6 Temperature profile and various zones in a blast furnace
1.4 Equipment
1.4.1 Equipment overview

An overview of the major equipment is shown in Figure 1.7. It includes:
– Hot Blast Stoves. Air preheated to temperatures between 1000 and 1250 °C is
produced in the hot blast stoves and is delivered to the furnace via a hot blast
main, bustle pipe, tuyere stocks and finally through the tuyeres. The hot blast
reacts with coke and injectants and forms the raceway in front of the tuyeres.
– Stock house. The burden materials and coke are delivered to a stock house.
The materials are stored, screened and then weighted before final delivery into
the furnace. The stock house is operated automatically. Corrections for coke
moisture are generally made automatically. The burden materials and coke are
brought to the top of the furnace via skip cars or via a conveyor belt, where they
are discharged into the furnace in separate layers of ore and coke.
– Coal or gas auxiliary fuel is injected through the tuyeres.
– Gas cleaning. The top gas leaves the furnace via uptakes and a downcomer. The
hot top gas contains many fine particles, necessitating removal of the particles
and cooling the gas in a gas cleaning system consisting of a dust catcher or
cyclone, wet cleaning of the gas in a scrubber and finally removing water in a
demister. To save cleaning water and aid recovery of sensible heat and pressure
of top gas, the dry cleaning method of blast furnace top gas has been developed
and is widely used in newly built blast furnaces around the world, especially in
China. The method uses a series of chambers filled with bag filters to filter fine
dust in hot top gas.
– Casthouse. The liquid iron and slag collect in the hearth of the furnace, from
where they are tapped via the taphole into the casthouse and to transport ladles.
Depending on the size of the furnace, there may be one to four tapholes and
connected casthouses. The molten slag may be quenched with water to form
granulated slag, which is used for cement manufacturing.
6 Chapter I
Figure 1.7 Blast furnace general arrangement
For the design of the blast furnace proper, it is important to realize that the
blast furnace process can operate at a wide range of setpoints and readily get out
of control. All subsystems of the furnace should be designed to be able to handle
these conditions with minimal interference with the process. All subsystem
requirements from the process will be costly, since production rates and coke
rates can be affected, or major damage can occur. An engineer opting for a more
or less forgiving design has to look seriously into the forgiving design.
The top of the blast furnace is closed, as modern blast furnaces tend to operate
with high top pressure. There are various different systems to distribute the
ferrous burden and coke into the blast furnace. The most common system is the
bell–less top (Figure 1.8).
Top Bin
Rotating Chute
Figure 1.8 Blast furnace top charging by bell–less top

1.4.2 Blast furnace and its development

A blast furnace has the shape of two truncated cones that are joined at their
widest point. The sections from the top down are:
– The throat, where the burden surface is located.
– The stack, where the ores are heated, and reduction reactions start.
– The bosh parallel or belly.
– The bosh, where the reduction is completed, and the ores are melted down.
– The hearth, where the molten material is collected and is cast via the taphole.
The liquid iron and slag are collected in the voids between coke particles in the
hearth.
Blast furnaces grew considerably in size during the 20th century. In the early
years of the century, blast furnaces had a hearth diameter of 4 to 5 meters and
were producing around 100000 tonnes of hot metal per year, mostly from lump
ore and coke. At the end of the 20th century, the biggest blast furnaces had a
hearth diameter of between 14 and 15 meters and were producing 3 to 4 million
tonnes per year.
The size of a blast furnace is often expressed as its hearth diameter, or as its
“working volume” or “inner volume”. The working volume is the volume of the
blast furnace that is available for the process i.e. the volume between the tuyeres
and the burden level. Definitions of working volume and inner volume are
given in Figure 1.9.
Bottom Bottom Bottom of

of bell of bell (vertical)
chute
Zero level burden
1 m below Bottom of 1 m below
bottom movable chute
of bell armour
Working Volume
Inner Volume
Total Volume
Tuyere level
Taphole level
Uppermost brick
bottom layer
Figure 1.9 Definitions of working volume and inner volume

8 Chapter I
Very big furnaces reach hot metal production levels of more than 12000 tonnes
per day. E.g. the Oita blast furnace No. 2 (Nippon Steel Corporation) has a
hearth diameter of 15.6 meters and a production capacity of 13500 tonnes per
day. In Europe, the Schwelgern No. 2 furnace (thyssenkrupp Steel) has a hearth
diameter of 14.9 meters and a daily production of 12000 tonnes per day. In
2013 POSCO commissioned Gwangyang No. 1 blast furnace with an inner
volume of 6000 m³, a hearth diameter of 16.1 meters, and a rated capacity of
5.65 million tonnes per year (~16000 tonnes per day), attaining a very high
productivity of 2,9 t/m³WV/24hrs. It is the largest furnace at the moment.
Currently, there are 32 blast furnaces with inner volumes above 5000 m³ in
operation around the world, most of them in China.
High performance blast furnaces use mainly sinter and pellets as ferrous
burden. The lump ore percentage has generally decreased to 10 to 15 % and
lower. The fuel used for the process has evolved as well. The fuel represents not
only energy but is also the driver of chemical reactions to reduce the oxygen
content of the burden. The term reductants is used for this reason. In the
early days, blast furnaces were operated with all coke operation. Nowadays,
coal injection through the tuyeres is widely used. Presently, about 30 to 40
% of the earlier coke requirements have been replaced by coal injection.
Occasionally, depending on the local situation, natural gas is used. As a result
of new technology to access natural gas deposits, the cost of natural gas has
decreased in some regions, and natural gas is now used for 10 to 15 % of the
total reductant requirement. Currently the co–injection of coal and natural gas
is being optimized in some areas of the world, especially in North America.
1.4.3 Cooling systems

The furnace inside is exposed to severe chemical and physical attack by raw
materials and by gas at very high temperature. In order to maintain the integrity
of the furnace, the inside of the shell is protected by a water–cooled refractory
lining. An extensive discussion of cooling systems and specifications is outside
the scope of this book.
There are two main shell cooling systems in use: stave cooling and horizontal
cooling plates for the bosh of a furnace, as shown in Figure 1.10. Cooling plates
are water cooled, typically produced from copper castings, with a relatively
small distance between each plate location and an insert depth of 40–50 cm.
If required, they can be replaced from outside in a few hours during a blast
furnace stop.
Staves are vertical cooling elements, which can be made from cast iron, steel or
copper, covered with refractories on the hot face. The insert depth is smaller,
which leads to an increase in the working volume of a blast furnace when
cooling plates are replaced by staves in the same shell. Replacing staves in an
operating furnace is more difficult than cooling plate replacement and requires
a furnace stop of at least a few days.
The advantage of staves is that, when the burden descends, the material can
follow the wall more easily. This is an advantage, especially in the stack. The
advantage of cooling plates is that they can work as an anchor for freezing a
protective skull, which is difficult with staves. So, especially in the lower part
of the furnace, cooling plates are more forgiving than staves. For this reason,
hybrid systems are presently being developed and installed in some furnaces.
Figure 1.10 Cooling systems: plate coolers, staves and hybrid
1.4.4 Gas cleaning

The large quantity of gas produced during blast furnace operation contains
10 to 40 g/m³ of dust, which is reduced to below 10 mg/m³. Most of the
dust is removed in a gravity dust catcher or a cyclone dust catcher. The gas is
subsequently cleaned by either a wet cleaning process or a dry cleaning process.
Wet cleaning processes have been used for many years.
A more recent development is to use dry cleaning processes with dry bag–filter
cleaning (DBFC). This is used in many newly built blast furnaces, especially in
China. When blast furnace gas passes through bag filters, more than 99 % of
the fine dust is removed. After cleaning, the dust content of the gas is less than
5 mg/m³. The temperature of the cleaned gas is 80 to 150 °C.
The key point for operational control of a DBFC system is to keep the
blast furnace gas inlet temperature in a suitable temperature range. If the
temperature is too high (above 250 °C), the bag filters would be at risk of
burning. If the temperature is too low (below 80 °C), condensation can
occur, which jeopardizes proper gas cleaning. The advantages of dry gas
cleaning are improved energy recovery from top gas and avoidance of all
sludge related issues. On the other hand, handling the dry dust, which can
ignite spontaneously, requires special precautions. The lower limit of the top
10 Chapter I
temperature can be a concern, since high productivity in a blast furnace can

only be reached at low top temperature. In some cases, measures for the removal
of chlorine compounds in dry cleaned gas have to be taken to prevent corrosion
in the gas grid.
1.5 Book overview
Blast furnace ironmaking can be discussed from three different perspectives:

– The operational approach: discussing the blast furnace with its operational
challenges.
– The chemical technology approach: discussing the process from the perspective
of the technologist who analyzes the progress of chemical reactions and heat
and mass balances.
– The mechanical engineering approach focusing on equipment.
The focus of this book is the “operator’s view”, with the aim of understanding
what is going on inside the blast furnace. To this end the basic principles of
the process are discussed (Chapter II) followed by the demands on burden
quality (Chapter III) and coke and auxiliary reductants (Chapters IV and V).
Processing the ferrous burden by gas is described in Chapter VI. The control of
the process by burden distribution is discussed in Chapter VII. Subsequently,
hot metal and slag quality (Chapter VIII), casthouse operation (Chapter IX),
blast furnace operational control (Chapter X) and special operational situations
like stops and starts, high moisture input or high amounts of fines charged into
the furnace, blow–in and blow–down (Chapter XI) are discussed.
Throughout the present book practical examples are given of operating furnaces,
which are indicated as Examples.
Throughout the book, examples are based on an operating “reference furnace”: a

furnace with a 15 meter hearth diameter and 5000 m³ inner volume, producing
12000 tonnes of hot metal (tHM) per day, a coke rate of 300 kg/tHM and a
coal injection rate of 200 kg/tHM. The details are shown in Chapter II and
Annex II. Examples of operating furnaces are shown in text frames throughout
the book.
II Blast Furnace
Process and Internal Structure
2.1 Gasifying coke and coal
The blast furnace process starts when pre–heated air, or “hot blast”, is blown
into the blast furnace via the tuyeres at a temperature of up to 1250 °C. The
hot blast gasifies coke (and injected coal and gas) in front of the tuyere. The
gasification generates a very hot flame, which is visible through the peepsight
as the “raceway”. The gas contains the heat required for melting the ferrous
burden.
At the same time, oxygen in the blast is transformed into gaseous carbon
monoxide (CO), and hydrogen (H2) is generated from the injected fuels and
from moisture in the hot blast. These gases can reduce the oxygen content of the
burden, hence the name “reductant”.
The raceway has a banana type shape and is typically about 1.8 meters long
(Figure 2.1a). The borders of the raceway are well defined at the inside and
bottom, because small coke particles help create a relatively stable structure. At
the side and upper part of the raceway, the coke is in continuous movement and
no stable structure is formed. In Figure 2.1b on the next page the remains of
the raceway are shown, after a blast furnace has been taken out of operation. It
shows that every tuyere generates its own flame into the furnace.
Figure 2.1a The raceway, horizontal and vertical sections

12 Chapter II
Figure 2.1b The raceway after a blowdown and nitrogen quench

(courtesy John Ricketts)
The raceway gas or bosh gas is formed as follows. For every molecule of oxygen
(O2), two molecules of carbon monoxide (CO) are formed: every cubic meter
(m³ STP) oxygen will generate 2 m³ STP of CO. Air contains 21 % oxygen
and 79 % nitrogen. If the hot blast is enriched to an oxygen level of 27 % (and
thus has 73 % nitrogen), the bosh gas will consist of 57.5 % (i.e. 73/(73+227))
nitrogen and 42.5 % CO gas. For the time being, we neglect the effect of
hydrogen (see section 5.1). In addition, a huge amount of heat is generated in
the raceway from the combustion of coke and injectants (coal, oil, natural gas).
The heat leads to a high flame temperature, which generally is in the range of
1900 to 2300 °C. Since this temperature is higher than the melting temperature
of iron and slag, the heat in the hot gas can be used to melt the burden. Flame
temperature is discussed in more detail in Chapter VI. In the present chapter,
the gas flow through the furnace is analyzed in more detail. The charge consists
of alternating layers of ore burden (sinter, pellets, lump ore) and coke.
2.2 Counter–current reactor
The blast furnace is a counter–current reactor (Figure 2.2 – next page). The
process starts with the hot blast through the tuyeres, which reacts with the
coke and coal or natural gas injection in the raceway (Figure 2.1) generating
hot raceway gas (1900–2300 °C). The hot raceway gas moves upwards, while
heating and melting the charged materials. Consumption of coke and melting
of the ore burden creates voidage inside the furnace, which is filled with
descending burden and coke. The upward movement of gas: gas passes through
a blast furnace in 5–10 seconds, while towards the burden stockline (21 to 25 m
from tuyeres) the gas temperature drops from a flame temperature at the tuyeres
of 1900–2300 °C to a top gas temperature of 100–150 °C. The downward
movement of the burden: burden is charged cold and wet into the top of the
furnace; it is reduced, melted and heated to 1500 °C in 5–8 hrs.
Blast Furnace Process and Internal Structure 13
Figure 2.2 The blast furnace as a counter–current reactor
2.3 Internal layer structure
The internal layer structure of alternating coke and ferrous burden layer is
maintained throughout the furnace until the burden is molten. In the cohesive
zone the burden materials soften and collapse. Upon further heating, the
burden melts and drips down through the active coke zone and collects in
the hearth that is filled with coke and separates into hot metal and slag. Coke
remains solid in the blast furnace. It is consumed by gasification at the tuyeres,
chemical reactions in the furnace and by dissolving in hot metal.
Figure 2.3 Blast furnace zones

14 Chapter II
As indicated in Figure 2.3, an operating blast furnace contains:

– Layers of ore and coke.
– An area where ore starts to soften and melt, known as the softening–melting
zone or cohesive zone.
– An area where there is only coke and liquid iron and slag, called the “active
coke” or dripping zone.
– The deadman or “inactive coke zone”, which is a stable pile of coke in the hearth
and extending into the bosh of the furnace.
The internal structure was derived from studies of quenched furnaces, which are
“frozen in action” using water or nitrogen. The examples in Figure 2.4a clearly
show the layer structure of coke and ore. Figure 2.4b shows a solidified cohesive
zone. Further analysis reveals information about the heating and melting of the
ore, as well as of the progress of chemical reactions.
Figure 2.4a Dissections of quenched blast furnaces Kakogawa 1 and Tsurumi

(Based on Omori et al, 1987)
Figure 2.4b Cohesive zone left after blow–down

Example 2.1 Furnace profile and burden and coke layers

The profile of the reference furnace is shown in Figure 2.5. For the furnace,
the throat diameter is 11 meters, the hearth diameter 15 meters. As a
consequence, the cylinder from throat to hearth is located 2 meters from the
tuyeres, as indicated by the blue cylinder.
This cylinder represents about 50 % of the total working volume of the

furnace. This means that coke and ferrous burden move not only downwards,
but also sideways. The blue cylinder is located at the end of the raceway.
The diameter at the belly is 16.5 meters, which has a surface of 2.3 times the
surface at the throat. The average layer thickness at the belly is approximately
43 % of the thickness at the throat.
In the reference furnace, the coke layer decreases from 57 cm at the throat to
25 cm in the belly, which is 5–6 coke lumps. The ferrous layer decreases from
78 cm to 35 cm, where layer thickness decreases further by softening.
30
Stockline
11 m
25
Height – relative to tuyere level (m)
20
49 %
15
10
16,5 m
5
Tuyeres
0
15 m
Taphole
–5
–10
10 5 0 5 10
Distance from center
Figure 2.5 Dimensions of the reference blast furnace used throughout the
book. The highlighted cylinder represents the volume directly
below the throat, corresponding to 49 % of the volume.
16 Chapter II
2.4 How gas flows through the furnace

The hot gas ascends through the ore and coke layers to the top of the furnace.
If there was only coke in the blast furnace, the chemical composition of the
gas would remain constant, but the temperature of the gas would reduce as
it comes into contact with the colder coke layers high in the furnace. The gas
flowing through a blast furnace filled with coke is shown in Figure 2.6. To the
experienced blast furnace operator, the furnace filled with only coke may seem
a theoretical concept. However, in some practical situations, like the blow–in
of a new furnace or when taking a furnace out of operation for a long time
(banking) the furnace is almost entirely filled with coke.
Figure 2.6 Gas flow in a furnace filled with coke only (left) and in a furnace filled
with alternating layers of coke and ore (right).
In the normal operational situation, the furnace is filled with alternating coke
and ore layers. About 30 to 45 layers of ore separate the coke into layers. It
is important to note that the permeability of coke is much better than the
permeability of ore. This is due to the fact that coke is much coarser than sinter
and pellets, and the void fraction within the coke layer is higher. Consequently,
the burden layers determine how the gas flows through the furnace, while the
coke layers function as gas distributors.
If gas flows from the bosh upwards, what happens to the gas as it gradually
cools down? Firstly, the heat at a temperature in excess of 1400 °C, which
is above melting temperature of hot metal and slag, is transferred to the
layered burden and coke. This melts and heats up the metallic portion. In the
temperature range from 1200 °C to 1400 °C, the burden will soften and stick
together rather than melt. Below the softening and melting zone the remaining
oxygen in the ore burden is removed.
Upon further cooling, the gas is capable of removing oxygen from the ore
burden, while producing carbon dioxide (CO2). The more oxygen that is
removed, the more efficient the furnace is. Below temperatures of 1000 °C the
following takes place:
– Heat is transferred from the gas to the burden.
– CO2 gas is generated from CO gas, while reducing the amount of oxygen of
the ore burden. This is called the gas reduction reaction, and in the literature
is sometimes called “indirect reduction” as opposed to “direct reduction”. No
additional gas is generated during this reaction.
– A similar reaction takes place with hydrogen. Hydrogen can remove oxygen
from the burden to form water (H2O).
Higher in the furnace, the moisture in the burden and coke is vaporized in
the “drying zone” and so is eliminated from the burden before any chemical
reactions take place.
In summary:
– Heat is transferred from the gas to the ore burden, which melts and softens
(above 1200 °C).
– Residual oxygen in the burden is removed and additional CO is generated. This
is known as the direct reduction reaction.
– Heat of gas (< 1000 °C) is transferred to ore burden and coke. CO in the
gas removes oxygen from the burden, and generates CO2, this is called gas
reduction or indirect reduction.
2.5 Processing the ferrous burden
Burden and coke are gradually heated up on their way down the stack, and
moisture is evaporated. At around 500 °C, the removal of oxygen begins. Figure
2.7 shows a simplified diagram of the removal of oxygen from the ore burden.
Figure 2.7 Schematic presentation of reduction of iron oxides and temperature

18 Chapter II
The first step is the reduction of hematite (Fe2O3) to magnetite (Fe3O4). The
reduction reaction generates energy, so it helps to increase the temperature of
the burden. In addition, the reduction reaction creates tension in the crystal
structure of the burden material, which may cause the crystal structure to break
into smaller particles. This property is called Low Temperature Disintegration
(LTD). Several tests are available to quantify the effects (see Chapter III).
The second step in reduction is magnetite (Fe3O4) to wustite (FeO). This

reduction reaction costs energy.
Further down in the furnace the temperature of the burden increases gradually,
and part of the iron oxides are reduced to metallic iron until the burden starts
to soften and to melt in the cohesive zone. The molten iron and slag drip
through the coke layer below the cohesive zone and are collected in the hearth.
We now consider the interaction between the gas and the ore burden. The more
the gas removes oxygen from the ore burden, the more efficient the blast furnace
process. Consequently, intimate contact between the gas and the ore burden
is very important. To optimize this contact the permeability of the ore burden
must be as high as possible. The ratio of the gas flowing through the ore burden
and the amount of oxygen to be removed from the burden must also be in
balance.
Experience has shown that many problems in blast furnaces are the
consequence of low permeability ore layers. The permeability of an ore layer is
largely determined by the quantity of fines (under 5 mm) in the layer. Generally,
the majority of the fines are generated by sinter, or come from lump ores. The
problem with fines in the furnace is that they tend to concentrate in rings in
the furnace. As fines are charged to the furnace they concentrate at the point of
impact where the burden is charged. They are also generated by low temperature
reduction–disintegration. So, it is important to screen the burden materials
well, normally with 5 or 6 mm screens in the stock house, and to control the
low temperature reduction–disintegration characteristics of the burden.
2.6 Furnace efficiency
The process efficiency of the blast furnace, generally considered to be the

reductant (coke plus injectants containing carbon) rate per tonne of hot metal,
is continuously monitored through measurement of the chemical composition
of the gas exiting the furnace top, known as “top gas”. The efficiency is
expressed as the gas utilization formula, which indicates the percentage of CO
gas that is utilized for reduction of the ferrous burden.
CO
CO = CO + CO
2
# 100 %
2
In addition, in modern furnaces the gas composition over the radius is

frequently measured. The latter shows whether or not there is a good balance
between the amount of reduction gas and the amount of ore in the burden
across the radius. The wall zone is especially important, as the coke percentage
in the wall area should not be too low. The wall area is the most difficult area for
melting the burden, because the gas at the wall loses much of its temperature by
heat loss to the cooling system.
The top gas analysis gives a reasonably accurate indication of the efficiency
of the furnace. When comparing different furnaces, one should realize that
hydrogen also takes part in the reduction process.
The gas utilization also depends on the amount of oxygen that must be
removed. Since pellets have about 1.5 atoms of oxygen per atom of Fe (Fe2O3)
and sinter has about 1.45 (mix of Fe2O3 and Fe3O4), the top gas utilization will
be lower when using sinter. It can be calculated as about 2.5 % difference of
the top gas utilization, when comparing an all–pellet burden with an all–sinter
burden.
2.7 Casting hot metal and slag
The liquid hot metal and slag have to be removed from the furnace. This is done
by making a hole (a taphole) of 40–60 mm diameter and length of 2,5–3,5 m
in the furnace. Hot metal and slag separate in two liquid phases. The hot metal
is more dense (~7,2 tonne/m³) than slag (~2,3 tonne/m³). As a consequence,
generally the hot metal comes out of the taphole first, and later hot metal and
slag are drained together. Finally, when the slag level is below the taphole
elevation, some of the gas generated in the raceway blows out of the taphole
indicating the end of the cast. An operator should be aware that the volumes of
slag and iron are quite similar: the hot metal volume and slag volume produced
are about equal at a slag volume of 320 kg/tHM. At higher slag rates, there is
a higher volume of slag to be cast than hot metal. Moreover, since slag is more
viscous and also more variable than hot metal, it is more difficult to remove
from the furnace than hot metal.
20 Chapter II
Example 2.2 Contents of and gas flow through a blast furnace

The contents of a blast furnace and the gas flow can be derived from
operational results. How long does it take to process burden and gas?
Throughout this book a consistent set of calculation examples with the same
starting points is used. The reference furnace has an inner volume of 5000 m³
and a 15 meter hearth diameter. A schematic drawing is shown in Figure 2.5.
The furnace has a daily production of 12000 tonnes of hot metal (tHM) at a
coke rate of 300 kg/tHM and a coal injection rate of 200 kg/t. Yield losses are
neglected. The content can be calculated as below. Additional data are given
in Table 2.1 and Annex II.
Bulk Density Weight percentage

kg/m³ Fe C H
Ferrous Charge kg/tHM 1570 1800 60.2
Coke kg/tHM 300 550 86.0
Coal kg/tHM 200 80.0 4.4

Blast volume m³ STP/min 7500
Oxygen % 27
Hot metal production tHM/d 12000 94.5 4.5
Inner volume m³ 5000
Working volume m³ 4250
Throat diameter m 11.0
Hearth diameter m 15.0
One charge contains tHM 85
t burden 133.5
t coke 25.5
m³ 0
Layer burden at throat m 0.78
Layer coke at throat m 0.57
Voidage in shaft % 30
Table 2.1 Data for calculation of blast furnace contents
How much blast oxygen is used per tonne hot metal?

The oxygen from hot blast comes to 243 m³ STP blast oxygen/tHM as shown
in Table 2.2. This is calculated as 27/100  7500  60  24/12000.
Oxygen % 27
Wind volume m³ STP/min 7500
Production tHM/d 12000

Oxygen use m³ STP/tHM 243
Table 2.2 Oxygen use in reference furnace

In how many hours is the ferrous burden processed?

To produce one tonne of hot metal, the furnace is charged with 300 kg
coke (300/550 = 0.55 m³) and 1.570 kg sinter/pellets (1570/1800 = 0.87
m³). In total, 1.42 m³ of material is charged per tonne of hot metal. This
is put through a furnace with a working volume of 4250 m³. The burden
is processed in 5.68 hrs. The estimation is shown in Table 2.3, taking into
account a typical compression of 10 %. Processing will be faster towards the
center and slower in the wall area.
Coke kg/tHM 300 m³ 0.55
Ferrous charge kg/tHM 1570 m³ 0.87
Tonne hot metal m³ 1.42

Hot metal production tHM/d 12000
Coke charge t/d 3600 m³/d 6545
Ferrous charge t/d 18840 m³/d 10467
Total m³/d 17012
Compression % 10 m³/d 15311
Minus 1 m at throat m³ –95
Minus 3 m active coke zone m³ –530
Available m³ 3625
Processed m³/d 15311 m³/h 638
In available volume m³ 3625
Processing time h 5.68
Table 2.3 Ferrous burden processing time
How many charges are present above the active coke zone?
A charge consists of a layer of ore and coke. The number of charges depends
on the weight of one layer and coke rate. In the reference furnace a layer
contains 134 tonnes of ferrous burden, corresponding to 85 tonnes of hot
metal. The number of charges is 33, as calculated in Table 2.4, where 10 %
compression is considered.
Available volume m³ 3625

Volume per tonne hot metal m³/tHM 1.42
In available volume tHM 2557
Tonne hot metal per charge tHM 85
Number of charges 30
Compression % 10
Charges considering compression 33
Table 2.4 Number of charges in the reference furnace

22 Chapter II
Time required for the gas to travel through the furnace

The blast volume is 7500 m³ STP/min with 27 % oxygen. For every m³
of oxygen, two m³ of CO are produced. This gas has a high temperature,
the furnace is operated at a high pressure, and extra gas is formed by the
direct reduction reaction. It contains hydrogen as well. If all these effects are
neglected, the exercise is straightforward: Suppose the void fraction in the
burden is 30 %, then the open volume in the furnace is the working volume
times voidage (4250 m³, voidage 30 %). Through this volume more than
7500 m³ STP gas is blown per minute. So the residence time of the gas is
approximately 10 seconds: (42500.3)/(7500/60). This is a crude estimate.
It is possible to make the corrections for temperature, pressure, additional

CO from direct reduction and hydrogen from coal. Assume an average
temperature of the gas of 900 °C and an average pressure of 3.2 bar. The
estimation shown in Table 2.5 results in 6.3 seconds for the reference
furnace. Note that the gas velocity above the stockline is an average of only
approximately 1 m/s. However, in the center the gas speed maybe 10 times
the average. In general, as soon as the hot blast has entered through the
tuyeres, gas velocities drop down to low or very low values.
Blast volume m³ STP/min 7500
Oxygen % 27
Bosh gas (H2 neglected) m³ STP/min 9525

Gas from direct reduction
Carbon used for DR kg/tHM 96
Hydrogen from coal kg/tHM 8.8
Extra gas m³ STP/tHM 278
Total m³ STP/min 2315
Total gas volume m³ STP/min 11840
Gas volume in m³
Higher temperature (est.) °C 900
Blast pressure bar 4.3
Top pressure bar 2.5
Pressure inside (est.) bar 3.2
Real gas volume m³/min 12112
Flows through
Voidage % 30
Available volume m³ 1275
Time in furnace min 0.105
sec 6.3
Table 2.5 Time gas is in the furnace

III The Ferrous Burden
3.1 Introduction
In the early days of commercial ironmaking, blast furnaces were often located
close to iron ore mines. In those days, blast furnaces were using local ore and
charcoal, later replaced by coke. In most of the industrial areas of the time,
the 19th century, there were many blast furnaces operating in Germany, Great
Britain and the United States. After the application of the steam engine for
ships and transportation, the center of industrial activity moved from the ore
bodies to the major rivers, such as the river Rhine, and later from the rivers
to the coastal ports with deep sea harbors. This trend, supported by seaborne
trade of higher quality ores may appear clear at present, but has only a recent
history. In 1960 there were sixty operating blast furnaces in Belgium and
Luxembourg. Since 2009, only two have been operating; both have a favorable
coastal location. A similar trend, with a much faster pace can now be observed
in China.
The trend towards fewer but larger furnaces has increasingly made a rich iron
burden a requirement. A rich iron burden translates into a high Fe content,
frequently obtained after a physical beneficiation or enrichment process of the
ore at the mine. This has consequently made more fine material available, as
opposed to a rich and good lump ore, which in turn has become more scarce.
These fine ores are too impermeable to gas flow to be charged directly to the
blast furnace. Hence sintering and pelletizing as an agglomeration process is
favored. Sinter and pellets are used in combination with lump ore as the ore
burden. Briquetting of in–plant revert materials is an option, as recycling of
some of these waste materials through a sinter plant becomes more restricted
or, in the absence of a sinter plant, a prerequisite. Boosting blast furnace
productivity through charging of metallic iron (scrap, DRI or HBI) in
combination with reducing greenhouse gas emissions can make up the total
ferrous burden
A well prepared blast furnace burden consists, for the major part, of sinter and/
or pellets and can be topped off with sized lump ore (Figure 3.1). Sinter burdens
are prevalent in Europe and Asia, while pellet only burdens are used more
commonly in North America and Scandinavia. In China, ferrous burden now
consists of an estimated 70 % sinter, 15–20 % pellets and 10–15 % lump. Many
other companies use sinter as well as pellets, although the ratios vary widely.
24 Chapter III
Sinter Pellets Lump

90 % < 25 mm 9.5–16 mm 8–40 mm
Figure 3.1 Burden materials
Most mines generate fines only or have a lump yield below 30 %, with a few
exceptions (South Africa, Canada) reporting over 70 % yield. Lump being a
natural agglomerate, its properties are harder to control and it can have a large
effect on the blast furnace process. For this reason, it is mainly used as a lower
cost replacement for acid pellets. For high productivity, low coke rate blast
furnace operations, lump ore is generally maximized in the range of 10 to 15 %.
The achievable rate depends on lump ore quality and its proper use.Hence the
successful use of higher percentages is documented. The present chapter deals
with the ferrous burden quality.
3.2 Iron ore
Iron is the fourth most abundant element in the earth’s crust, making up
approximately 5 % of the total. However, mining of iron (as an oxide) is only
economically viable where substantial concentration has occurred, and only
then can it be referred to as iron ore. More than 3 billion years ago, through the
generation of Banded Iron Formation, the first concentration occurred.
Figure 3.2 Banded Iron Formation (Natural History Museum, courtesy Rio Tinto)
The Ferrous Burden 25
The conventional concept is that several billion years ago, the banded iron layers
were formed at the bottom of ancient shallow seas, as the result of an increase
in oxygen in the water. These formed insoluble iron oxides that precipitated
out of the water, alternating with mud, later forming cherts and silicate layers.
Subsequently, metamorphoses or leaching out of the cherts and silicates resulted
in a concentration of the iron oxide. Through further geological processes such
as (de)hydration, inversion leaching, deformation and sedimentation, a wide
variety of iron ore deposits were created all around the world. These total over
300 billion tonnes at an average Fe content of 47 %.
A minor fraction of these deposits is currently commercially mined as iron ore

with Fe contents ranging from below 30 % to up to 68 % (pure iron oxide as
hematite, Fe2O3, contains 70 % Fe). As mentioned before, an efficient blast
furnace process requires a rich Fe burden, preferably in excess of 58 % Fe. With
the conventional mining process of drilling, blasting and crushing, a wide size
range of particles is obtained. Through screening, lump ore and sinter fines
are separated. However, an increasing proportion of the iron ore needs further
beneficiation and processing prior to becoming a usable material for the blast
furnace. A vast amount of equipment has been developed to upgrade the iron
ore to a suitable product economically. These processes will not be described
here, but most of them are based on liberating the iron oxide from the gangue
minerals and then making use of the differences in density, magnetic properties
or surface properties between the minerals to separate them physically.
Sometimes large amounts of quartz (SiO2) need to be removed, and desliming
of undesirable clay, or minor amounts of impurities (such as phosphorus in the
mineral apatite) is carried out. Depending on the specific requirements, these
processes can be easily achieved, or they can be difficult or even impossible.
These processes result in a wide variety of beneficiated iron ores with varying
grades, impurities and sizes being available on the market. Silica content
can vary from 0.6 % to above 10 %, and phosphorus from below 0.005 % to
above 1 %. Similar variations apply for other components such as the oxides of
aluminum, calcium, magnesium, manganese, titanium and alkalis. With tighter
environmental control over the whole process chain, tramp elements at minute
levels are starting to play a more dominant role. From sulfur, zinc and copper
to mercury, arsenic and vanadium. The importance of each of these elements
greatly depends on the applied process and process conditions, environmental
measures and locally applicable legislation.
Due to the physical beneficiation, more and finer iron ores have been generated.
Though the designation is not consistent, the ore in the size range for sintering
(indicative between 6 mm and 150 μm) is called sinter feed, while smaller
and more narrow sized is denoted concentrate, and finally there is pellet feed
which is indicative < 150 μm (90 % minus 200 mesh) and suitable mainly for
pelletizing. The original lump ore is typically between 8 mm and 40 mm. Most
common ferrous ores are hematite (Fe2O3), magnetite (Fe3O4) and goethite
(Fe2O3xH2O).
26 Chapter III
3.3 Quality demands for the blast furnace burden

The demands for the blast furnace burden extend to the chemical composition
and the physical durability of the burden materials. The chemical composition
must be such that after the reduction and melting processes the desired iron
and slag compositions are produced. This will be determined by the chemical
composition of all the materials charged in the furnace. The physical and
metallurgical aspects of the quality demands are related to the properties in
both the cold and the hot state and to assure that the gas can flow through the
burden and react with the ferrous components. Both aspects are discussed in
depth in this chapter.
3.3.1 Generation of fines, reducibility, softening and melting

In the shaft zone of the blast furnace, the permeability of the burden is
determined by the sizing of the burden components and specifically the amount
of fines (see Figure 3.3). Fines may be defined as the fraction of the material less
than 5 mm, since the burden components have a general range of 5 to 25 mm.
If there are too many fines and especially fines below 5 mm, the void fraction
will decrease and the resistance for gas will increase (Hartig et al, 2000). There
are two sources for fines, those that are directly charged into the furnace, and
those that are generated in the shaft by the process.
Figure 3.3 Permeability for gas flow depends on void fraction, which depends on the
ratio of smaller and larger particles. Example of two types of spherical
particles, large (V l) and small (Vs). The x–axis gives the fraction of the
large particles: V l/(V l+Vs). The ratio V l/Vs itself strongly influences the
absolute void fraction. This example assumes a high ratio between two
sizes of particles; in reality a size range will always exist.
Fines before charging can be controlled by proper screening in the stockhouse,

just before charging to the blast furnace. The target should be less than 3 %
below 5 mm, after screening in the stockhouse. Measurement of the percentage
of fines after screening in the stockhouse can give an indication whether or not
excessive fines are charged into the furnace. Material from the stockyard will
have varying levels of fines and moisture, and thus screening efficiency will be
affected accordingly. In some cases, iron ore is pre–screened in the stockyard to
remove fines that cannot be removed effectively in the stockhouse, like piggy–
back fines.
Once in the furnace, during the first reduction step from hematite to magnetite,
the structure of the burden materials breaks down, and fines are generated.
Sinter and lump ore are especially prone to this effect, known as reduction–
disintegration or reduction–degradation. The reduction–disintegration depends
on the strength of the bonds between the particles of ore fines in sinter and
lump ore. Generally speaking, the reduction–disintegration is dependent on:
– The FeO percentage, especially in the sinter. The more magnetite (Fe3O4,
which corresponds to FeO.Fe2O3) is present, the stronger the sinter and the
less reduction–disintegration can take place at low temperature, caused by the
change in crystal structure from hematite to magnetite. In the sinter process,
the FeO percentage in the sinter can be increased by cooling sinter with air that
is low in oxygen or increased by adding more fuel (coke breeze) to the sinter
blend.
– The chemical and mineralogical composition of the lump ore and sinter. For
sinter, basicity, Al2O3 and MgO content play an important role. Higher basicity
improves sinter strength while higher alumina and magnesia can have a negative
impact on sinter strength. Lump ore strength is dictated by mineralogy in
combination with compactness/porosity.
– The heating and reduction rate in the furnace. The slower the progress of
heating and reduction, the more time is given for disintegration of sinter and
lump ore.
– The amount of hydrogen in the reducing gas. More hydrogen in the reducing
gas leads to lower reduction–disintegration.
A major requirement for the blast furnace ore burden is to limit the quantity of
fines within the furnace to as low as possible. This can be achieved by:
– Proper screening of burden materials before charging. Screens with 5 mm
aperture are normal operational practice. Dry sinter and pellets screen more
efficiently than wet pellets, lump and stocked sinter from the yard.
– Good reduction–disintegration properties.
During charging, fines in the burden material tend to concentrate at the point
of impact on the burden surface. The level of reduction–disintegration increases
in areas where the material is heated and reduced slowly. A charged ring of
burden with a high concentration of fines will impede gas flow, experience the
slower warm–up and so result in a higher level of reduction–disintegration.
Hence this negative process sustains itself.
28 Chapter III
The reducibility of the burden is controlled by the contact between gas and
the burden particles as a whole, as well as the gas diffusion into the particles.
Whether or not good reduction is obtained in the blast furnace is governed
by the layer structure of the burden and the permeability of the layers, which
determines the blast furnace internal gas flow. The intrinsic reducibility of the
burden components will be of less importance, since reduction to wustite takes
the same time for materials with different reducibility. See operational results in
Example 2.
As soon as burden material starts softening and melting, the permeability

for gas is greatly reduced. For this reason, the burden materials should start
melting at relatively high temperatures, and the difference between softening
and melting temperatures should be as narrow as possible, so that they do not
impede gas flow while they are still high up in the stack. Furthermore, the
amount of slag is also important for furnace permeability. Melting properties of
burden materials are determined by the slag composition. Melting of acid pellets
and some lump ore can start at temperatures as low as 1000 to 1100 °C, while
fluxed pellets and basic sinter generally starts melting at higher temperatures.
See also Section 8.7 on how iron ore melts.
3.3.2 Ore burden quality tests

Ore burden material is characterized by the following.
– Chemical composition.
– Physical characteristics:
– Size distribution, which is important for the permeability of ore burden layers
in the furnace.
– Cold strength, which is used to characterize the degradation of ore burden
materials during transport and handling.
– Metallurgical properties:
– Reduction–disintegration, which characterizes the effect of the first reduction
step and is relevant in the stack zone of the furnace.
– Reducibility, which characterizes the capability for the material to improve
the extent of reduction reactions.
– Softening and melting properties, which are important for the formation of
the cohesive and melting zone in the furnace.
Most of the tests are ISO standardized; some of the tests are generic for
sinter, pellets and lump but some are specific. For pellets for example, a cold
compression strength and a swelling test are applicable. The latter is to ensure
that the volume increase during reduction does not exceed a set maximum.
Specifically, lump ores are tested for decrepitation: fracturing of the lump due to
thermal decomposition of crystalline water during the initial stage of heating in
the top of the blast furnace.
Table 3.1 summarizes these characteristics with an indicative range. The authors
consider the ranges suitable for high performance blast furnace operation,
which is a productivity above 2,4 tHM per m³ inner volume per 24 hours and a
coke rate below 300 kg/tHM.
A short description of generic tests used for characterization of materials is given

below with the objective being to understand the terminology.
Figure 3.4 Principle of tumble test
Optimum Range
What is measured? Results Sinter Pellets Reference
Mean Size Size distribution Average size, mm ISO 4701
% 6.3–16 mm > 95 %
% < 0.5 mm <2% <2%
Cold Strength Size distribution % > 6.3 mm > 70 % > 90 % ISO 3271
after tumbling % < 0.5 mm <5%
Compression daN/p > 150 ISO 4700
Strength after Size distribution % > 6.3 mm > 80 % ISO 4696
reduction after reduction % < 3.15 mm < 20 %
LTD (Low Temp. and tumbling % < 0.5 mm < 10 %
Disintegration)
Reducibility Weight decrease %/min > 0.7 % > 0.5 % ISO 4695
during reduction
Table 3.1 Characterization of ore burden, indicated range is suitable for high
performance blast furnace operation.
3.3.2.1 Tests for cold strength

Cold strength is mostly characterized by a tumbler test. For this test an
amount of material is tumbled in a rotating drum for a specified time interval.
Afterwards the amount of fines is measured. The size distribution after
tumbling is determined and used as a quality indicator (Figure 3.4). This
process simulates the physical transport and handling of all ferrous burden
components. Pellets are also tested for Cold Compression Strength to simulate
the weight of the burden on the pellets inside the furnace and amount of
breaking under a load.
30 Chapter III
3.3.2.2 Tests for reduction–disintegration

The reduction–disintegration tests are carried out by heating a sample of the
burden to at least 500 °C and reducing the sample with gas containing CO (and
sometimes H2). After the test, the sample is cooled, tumbled and the amount of
fines is measured. The quoted result is the percentage of particles smaller than
3.15 mm. Since pellets generate a minimum of fines, sometimes a dynamic test
is used to better distinguish the difference between pellet qualities. In this case
the pellets are reduced while being tumbled in a small rotating retort.
3.3.2.3 Reducibility
Reducibility is measured with two different ISO tests in a gas containing CO,
at a temperature of 900 °C or 950 °C. During the test period, the weight of the
sample is continuously measured. The weight loss is contributed to oxygen loss,
and either a figure for reducibility (dO/dt) or final reduction degree after 180
minutes is obtained.
Sometimes, non–ISO standardized tests are applied with generally the aim to
simulate the actual blast furnace conditions. HOSIM is such an example of a
blast furnace simulation test where the sample is reduced to the endpoint of
gas–reduction in a furnace. After the test, the sample is then tumbled. The
results are the reducibility, defined by the time required to reduce the sample to
the endpoint of gas reduction. The reduction–disintegration is represented by
the percentage of fines (under 3.15 mm) after tumbling.
Although all these tests are relevant for the upper part of the blast furnace
process, the ISO tests are excellent for having an idea on burden quality,
comparing qualities or controlling shipment to shipment quality. The more
advanced simulation tests provide a more realistic description of the effects
of that burden in the blast furnace. If test conditions are continued to higher
temperatures and the direct reduction (i.e. simulation of the lower part of
the furnace) is included, tests are referred to as softening and/or melting
tests. Pressure drop over the ferrous burden sample, softening and melting
trajectories and amount of material dripped out of the sample characterizes
the burden. Examples of softening tests are RUL (Reduction Under Load)
and tests including melting are for example REAS (Reduktion, Erweichung,
AbSchmelzen), ASAM (Advanced Softening And Melting) and multiple other
non–standardized softening and melting tests developed in–house.
3.4 Sinter
Sinter plants are almost always located in the vicinity of the blast furnace to
process a mixture of different iron ore fines, recycled ironmaking products and
fluxes into sinter. With the aid of solid fuel (coke breeze/anthracite), the aim
is to obtain a sinter with suitable burden characteristics (chemical, physical,
and metallurgical). This is achieved by a correct combination of the nature and
composition of each component of the mixture and the sinter process conditions
applied. Sinter characterization not only includes chemical and granulometric
analysis, with sometimes determination of the mineral phases in its structure,
but also assessing a series of quality indices that include cold strength, low
temperature degradation and reducibility. Within these constraints, sinter plants
will be operated at their maximum productivity, as it generally is the lowest cost
agglomerate.
3.4.1 Sinter quality

Sinter is made in three different types: acid (CaO/SiO2 ratio below 1.0), fluxed
(CaO/SiO2 ratio between 1.0 and 2.5) and super–fluxed sinter (CaO/SiO2 ratio
above 2.5) sinter. Fluxed sinter is the most common type. Since sinter properties
vary considerably with the blend type and chemical composition, only some
qualitative remarks can be made.
The sinter quality is defined by:

– Size distribution: sinter size ranges from 5 mm to 40 mm with a mean size
ranging from 15 to 25 mm as measured after the sinter plant. The more basic
the sinter, the smaller the average size. Sinter degrades during transport and
handling, so sinter has to be re–screened at the blast furnaces to remove the
generated fines. Sinter from stockyard may have different properties from
freshly produced sinter directly from the sinter plant. If stock sinter must be
used in the blast furnace, it should be charged in a controlled fashion, and
diluted with as much fresh sinter as is possible, such as by using a dedicated bin
in the stockhouse to stock sinter or pre–screened in the stockyard before going
to the stockhouse.
– Cold strength: normally measured with the tumble test. The more energy that
is used in the sinter process, and the more gangue components such as silica, the
stronger the sinter. The cold strength influences the sinter plant productivity
because a low cold strength results in a high–fines recycle rate.
– Reduction–disintegration (reduction–degradation index or RDI used in Japan)
properties. The reduction from hematite to magnetite generates internal stresses
within a sinter particle. The stronger the sinter, the better the resistance to these
stresses. The reduction–disintegration properties improve with denser sinter
structure, i.e. when the sinter is made with more coke breeze. As a consequence
of the higher coke breeze usage, the FeO content of the sinter will increase.
From experimental correlations, it is well known that, for a given sinter type,
reduction–disintegration improves with FeO content. For an operator this is the
fastest way to improve sinter quality, at cost. However, reducibility properties
are adversely affected.
The driving force of low temperature reduction–disintegration of sinter is the

changeover of the crystal structure from hematite to magnetite, which causes
internal stress in the iron oxide crystal structure. So, reduction–disintegration
of sinter is related to the fraction of hematite in the sinter. Most sinters contain
both primary hematite and secondary hematite (precipitated from the melt).
32 Chapter III
Particularly the latter causes reduction–disintegration, since it is more easily

reduced in the upper part of the furnace than primary hematite (see Figure 3.5).
Figure 3.5 Cracking of calcium ferrites (SFCA=silico ferrites of calcium and

alumina) due to reduction of primary (left) and secondary (right)
hematite (H) into magnetite (M). Pores appear black. (Chaigneau, 1994)
The higher the secondary hematite percentage in the sinter, the more the sinter
is prone to reduction–disintegration effects. This can also be said in reverse:
that is, there is a strong relationship between the FeO content of the sinter and
the reduction–disintegration. The higher the FeO content, the less reduction–
disintegration will take place. The FeO content of sinter can be increased by
adding more fuel to the sinter blend, which is normally done in the form of
coke breeze. However, the precise relationship between the FeO content of the
sinter and the sinter quality depends on the ore blend used and is plant–specific.
The reduction–disintegration properties depend on the type of FeO present in

the crystal structure. To illustrate this by example; a high fraction of magnetite
in the sinter blend will give sinter with a high (primary) magnetite fraction.
Moreover, in the presence of sufficient SiO2 fayalite structures, (2FeO.SiO2)
can be formed. These structures are chemically very stable and can only be
reduced at high temperatures by direct reduction reactions (see Section 8.2.1).
Alternatively, in the presence of MgO, spinel structures containing large
amounts of FeO can be formed. These spinel structures are relatively easy to
reduce. Low basicity sinter (CaO/SiO2 ratio below 1.1) can be produced, but
has to be formed at high temperatures and will therefore contain glass–like
structures where the FeO is relatively difficult to reduce.
Sinter plants in Europe use for cooling an air–gas mix with a reduced oxygen
percentage (12 to 14 %), which suppresses formation of secondary hematite.
This results in a relatively high FeO content of the sinter, because less secondary
hematite is formed. It has a major benefit for the reduction–disintegration
properties of this type of sinter. In addition, the calorific value of the blast
furnace top gas increases, as less oxygen has been removed from the ore burden,
giving an economic advantage.
During the sintering process there is a major difference between the use of
CaO and MgO as fluxes. Both materials are normally added as the carbonate,
using limestone as CaCO3 or dolomite as Ca.Mg(CO3)2. The carbonates are
decomposed on the sinter strand, requiring a large energy input. However, the
melts containing substantial amounts of CaO have low liquidus temperatures,
such as 1100 °C for mixtures of 20 to 27 % CaO and iron oxides. For the
melts containing MgO, the spinel structures mentioned above, the melting
temperatures are much higher. For this reason, it is easier to form slag–bonds
in the sinter using CaO than with MgO. And generally, making sinter with
CaO can be done at lower temperature. But sinter with high MgO is more
resistant to reduction–disintegration. MgO content can be increased by adding
olivine or serpentine to the sinter blend. However, there is an upper limit for
MgO content where sinter strength is negatively impacted. Again all these
relationships and their limits are plant specific based on raw materials and sinter
equipment.
Alumina, present in almost all iron ores in varying quantities, is an undesired

element, since at higher levels it increases the sinter RDI. European sinter
averages between 0.6 and 1.8 % alumina, whilst RDI is kept around 30 %. Each
0.1 % additional alumina can increase sinter RDI by 2 %. Though alumina
promotes SFCA (Silico–Ferrites of Calcium and Aluminum) formation, it also
influences the SFCA structure. Higher alumina promotes columnar SFCA
against the acicular and dendritic structures, whilst it also increases the viscosity
of the primary melt. All this leads to a more brittle matrix structure. As with
the effect of most elements in sinter, one must bear in mind that the origin of
the alumina (Koalinite, Gibsite, Bauxite) plays a major role in the fusion process
and hence the effect it has on the final sinter RDI. Titanium is also an element
reported to have a negative influence on sinter RDI.
For the final result of the produced sinter, it is important to note that the sinter
blend prior to sintering is far from homogeneous. It contains various types of
material, and locally there are widely varying compositions and sizes present.
Ore particles can be larger than 5 mm, coke breeze up to 3 mm and limestone
and dolomite up to 2.5 mm. All types of chemical compositions are present on
the micro–scale, where the sintering takes place. Types of materials used, size
distribution of the various materials, the blending of the sinter mix, the amount
of slag–bonds forming materials in the blend, as well as the amount of fuel used
for the sintering, all have specific disadvantages for good sinter quality. This
makes optimization of sinter–quality a plant–specific technological challenge.
3.4.2 Sinter basicity and sinter plant productivity

During sinter production, the vast majority of the mix is actually melted, and
slag bonding with its associated mineralogy is critical for the sinter properties.
Sinter basicity heavily imparts the formation of different mineral phases. As acid
sinter has a high percentage of brittle silicate glass phases as slag bonding and
lacks significant volumes of SFCA, it is weaker than fluxed/super–fluxed sinter.
34 Chapter III
Strength
0 1.0 2.0 3.0

Basicity CaO/SiO2
Figure 3.6 Sinter showing poor strength in the basicity range 1–1.4 and superior
strength for super–fluxed sinter (after Fernández–González et al, 2017)
The SFCA phase has been shown to be the preferred phase for improving the
quality of sinter i.e. strength. Figure 3.6 shows the correlation of sinter strength
with basicity. Moreover, flux addition often results in an increase in sinter
productivity as the average permeability of the sinter mix improves (especially
with burnt lime) and melting phases at lower temperature and viscosity are
formed, requiring less time and air volume for sintering. Tumbler index (TI)
and low temperature degradation index (LTD) of super–fluxed sinters are
comparable to lower basicity sinter. Softening and melting tests indicate that
sinter with a basicity of 3.5 has higher softening temperatures compared to
lower basicity sinter (basicity below 2.5).
In the above sections the importance of reduction–disintegration of sinter is

stressed. The lower the reduction–disintegration, the poorer the reducibility
of the sinter. Needle–like structures of calcium ferrites have a relatively open
structure and are easily accessible for reduction gas in the blast furnace. In cold
conditions the sinter is strong (i.e. good tumbler test results), the degradation
during transportation is also good, but the relatively fast reduction in the
blast furnace makes the sinter very prone to reduction–disintegration. More
solid structures in the sinter have better properties in this respect. Reduction–
disintegration leads to poorer permeability of the ore layers in the furnace and
impedes proper further reduction of the iron oxides in the blast furnace.
Sinter typically has a basicity (B2, CaO/SiO2) above 1.6. The range of B2
between 1.6 and 1.9 is very common, higher basicities are being used in specific
plants which use a low sinter percentage (below 50 %). Super–fluxed sinter has
a basicity above 2 and operation with a basicity of up to 4 has been reported. At
increasing basicity, production of the sinter plant (in tonne sinter/m2) increases
as shown in Figure 3.7 with data from two sinter plants in operation. The effect
is large at basicity of 1.6 but decreases at higher basicities. Higher basicity sinter
has a lower Fe content, which is compensated by increased production. At a
basicity of around 2.5, sinter plant productivity still increases, but no longer
compensates the decrease of Fe content, thus bringing less Fe units to the blast
furnace. Precise effects are site–dependent and depend among others on burnt
lime usage. So, an operator has to review his own historic data.
55 35
45 30
(t sinter/m².d)
Productivity
Productivity
(t Fe/m².d)
35 25
25 20
15 15
1.50 1.75 2.00 2.25 2.50
Basicity CaO/SiO2
Figure 3.7 Sinter basicity (B2) and sinter plant productivity, as tonne sinter/m².d
(drawn line) and tonne Fe in t/m².d (dotted). Data from two different
sinter plants.
3.5 Pellets
Similar to sinter, it is the interaction between the iron ore and the pelletizing
process that dictates whether high–quality pellets can be produced with high
productivity. Unlike sinter, most pellets are produced at the mine site, though
especially in China this is changing with some coastal located pellet plants.
3.5.1 Pellet quality

With the correct chemical composition and induration conditions, pellets can
be easily transported from mine to blast furnace. Pellets can be stocked and
generally remain intact in the blast furnace. Therefore, when judging pellets,
the main issues are:
– Chemical composition. Pellet chemical composition in principle can be made
to order. However not every ore body lends itself for to a low silica level (below
2.5% SiO2),
– Cold strength, measured as compression strength and the fines generated
through tumbling. Low figures for compression and tumble index indicate bad
or lean firing, which results in a pellet with a hard outer shell but an unfired
core.
– Pellet sizing. A narrow sizing with limited small pellets enhances overall pellet
quality.
– The reduction–disintegration properties. These properties are of less of a
concern with pellets than with sinter and lump ore.
– The swelling properties. With incorrect slag composition pellets tend to have
extreme swelling properties. Since the phenomenon is well known, it normally
does not happen with commercially available pellets.
– The softening and melting. Acid pellets tend to melt at lower temperatures than
fluxed sinter and fluxed pellets. More details on softening and melting can be
found in Section 10.9.
36 Chapter III
3.5.2 Pellet chemical composition

Alongside proper induration, the slag volume and composition and the bonding
forces mainly determine the quality of pellets. The three main pellet types are:
– Acid pellets (CaO/SiO2 ratio below 0.5)
– Basic or Fluxed pellets using limestone or dolomite as additives (CaO/SiO2 ratio
between 0.9 and 1.3)
– Olivine pellets, which are acid pellets with a high MgO content produced with
olivine as the fluxing additive.
Typical properties of the three types of pellets are shown in Table 3.2.
Pellet Type Compression Reducibility Swelling

Acid ++ – +/–
Basic + + +
Olivine + + +
Compression
Pellet Type Fe % SiO2 % CaO % MgO % (daN/pellet)
Acid 66–67 2–5 < 0.5 0.1–0.6 > 270
Basic 63–66 1.5–4 0.8–1.1 0.1–1.5 > 240
Olivine 64–67 2–4 < 0.5 1.3–1.8 > 200
Table 3.2 Overview pellet properties (daN = Decanewton, around 1.02 kgf)
Acid pellets are strong, but have moderate metallurgical properties. They have
good compression strength (over 250 kg/pellet), but may show relatively poor
reducibility indices. The softening and melting temperatures are low compared
to fluxed pellets or sinter. In addition, acid pellets are very sensitive to the CaO
content with respect to swelling. At a CaO/SiO2 ratio above 0.25, some pellets
have a strong tendency to swell, which might jeopardize proper blast furnace
operation.
Basic and fluxed pellets have good metallurgical properties for blast furnace
operation. By adding limestone or dolomite to the pellet blend, the energy
requirement of the firing/induration increases because of the decarbonization
reaction. For this reason, production capacity of a pellet plant can sometimes be
10 to 15 % lower when producing basic pellets compared with acid. However,
the fluxed pellet reducibility, softening temperature and melting temperature
are higher than acid pellets.
Olivine pellets contain MgO in place of CaO, which is added to the blend as
olivine or serpentine. The pellets are somewhat weaker when tested for cold
compression strength but softening and melting properties are improved.
An important attribute of pellets is their ability to continue to yield a superior

agglomerate at minimum silica content. When the ore body and upgrading
lends itself, pellets with silica below 1 % can be produced. Low silica pellets
would have in average less than 2.5 % silica and in combination with a sinter
burden, this offers the possibility to lower blast furnace slag volume and
enhanced productivity through less slag and consequently less coke.
3.5.3 Pellet sizing

Pellets represent the narrowest sized agglomerate in the burden. Though
generally over 90 % is quoted to be within the 9–16 mm range, there is an
argument for a size range as narrow as possible. The size of the pellets is
determined in the green balling stage and subsequent roller screening of these
green balls. Upon induration, each pellet receives the same heat treatment,
irrespective of its size. Under such conditions, size variations will result in
unequal induration and hence in quality variations due to under or over firing.
A pellet with a size of 16 mm is already 50 % larger in volume than one of
14 mm. Some pellet suppliers report pellet sizing as narrow as 94 % between
9.5–14 mm.
Large pellets promote a higher burden permeability while smaller pellets reduce
more easily. A wide distribution will negatively impact burden permeability,
although if pellets are part of a sinter/lump burden, this negative effect will
diminish.
3.5.4 Cold compression strength

The difference in compression strength might seem large. However, in the blast
furnace the pellets are reduced, and the difference diminishes during reduction.
After the first reduction step to Fe3O4, the cold compression strength drops as
low as 45–50 daN for acid pellets and to 35–45 daN for olivine pellets. So, a
little lower average compression strength has no drawback for the blast furnace
process.
However, the average low compression strength can occasionally be the

consequence of an increased percentage of very weak pellets (< 60 daN/pellet).
That fraction is a good indicator for the pelletizing/induration process: the more
pellets that collapse at low compression strength, the poorer the pellets have
been fired. Therefore, pellet quality can be influenced by the production rate:
the slower the grate moves, the stronger the firing can be, so the induration
period increases and the pellets become stronger.
Though it is sometimes claimed that pellets need to be strong to support the

burden on top, actual calculations show that the theoretical strength of a pellet
needs to be minimum: with a ferrous burden density of 1800 kg/m³, the weight
per cm2 (roughly the cross–sectional surface area of a single pellet of 11 mm) is
0.18 kg. A column of 20 m height of ferrous burden imposes a static “pressure”
of 3.6 kg on this single pellet only. Factors like burden movement, blast pressure
counteracting the downward pressure, bridging and coke will offset this
calculated pressure.
38 Chapter III
3.5.5 Swelling
As mentioned above, pellets, in contrast to sinter and lump ore, can have the
tendency to swell during reduction. Generally, a volume increase of up to 20 %,
measured according to ISO 4698, is seen as acceptable. Abnormal swelling can
easily go over 100 % volume increase. The effect of swelling, however, depends
on the percentage of pellets used in the burden. Abnormal swelling may occur
during the transformation of wustite into iron. Like any transformation,
this is a balance between iron nucleation and growth of these nuclei. During
swelling, limited nucleation occurs and these nuclei grow like needles causing a
volume increase which is seen as swelling, see Figures 3.8. These needles, called
whiskers, are difficult to observe; a microscopic image of the phenomenon is
shown in Figure 3.9. Under certain conditions, for example in the presence
of alkalis in the blast furnace, the swelling can become excessive and a
cauliflower–structure develops with cracks (Figure 3.8). This coincides with a
low compression strength of this structure, with the risk of generating fines.
Figure 3.8 Balance between iron nucleation and nuclei growth. Limited swelling
accompanied by the formation of an iron shell (left). Limited iron
nucleation followed by strong needle growth of the nuclei with as a result
excessive swelling of the pellet (right).
Figure 3.9 A swollen pellet and its Whisker formation
Main factors influencing pellet swelling are basicity and gangue (non–iron
compounds) content. Figure 3.10 shows how swelling depends on pellet basicity.
Pellets with a basicity between 0.2 and 0.7 are more prone to swelling.
Volume Increase B2 Basicity:

Maximum
Tolerated CaO
SiO2
0.2 0.7
B2 Basicity
Figure 3.10 Volume increase of pellet by swelling with increasing basicity.
Swelling is mitigated by proper induration. In the blast furnace local process

conditions like temperature and gas composition greatly influence the swelling
behavior. At higher reduction degrees also, swollen pellets will shrink. The
Japanese measure this shrinkage in a Contraction Test and have empirical data
showing that pellet contraction over 10% in a large blast furnace has negative
impact on productivity and reductant rate. As the phenomenon of swelling is
well known, it is normally under control with commercially available pellets,
but always requires a check because it could have a severe impact on the
regularity of the blast furnace process.
3.5.6 What are good pellets?

An overview of pellet properties is shown in Table 3.3. As pellets can be
produced as acid and fluxed pellets, whereas sinter is generally fluxed and lump
ore always acid, a difference in LTD and reducibility indices becomes apparent.
What is measured? Results Acceptable Reference

Range
Mean Size Size distribution % 6.3–16 mm > 95 % ISO 4701
% < 6.3 mm <2%
Cold Strength Compression Average kg/p > 150 daN/p ISO 4700
Strength % < 60 kg/p <5%
Tumbling Strength % > 6.3 mm > 90 % ISO 3271
and Abrasion % < 0.5 mm <5%
LTB (Low Temp. Size distribution after % > 6.3 mm > 80 % ISO 4696
Breakdown) static reduction and
tumbling
Reducibility Weight decrease %/min (dR/dt) 40 > 0.5 %/min ISO 4695
during reduction
Table 3.3 Characterization of acid and fluxed pellets

40 Chapter III
This difference is mainly driven by the ISO test conditions themselves,

immediately showing some of the limitations of ISO tests. Since an acid pellet
does not readily reduce at low temperatures, it sustains a higher intrinsic
strength (high LTD) against the more reducible fluxed pellet. However, this
does not imply that an acid pellet does not reduce or disintegrate. It will happen
only at more elevated temperatures in the furnace.
3.6 Lump ore
Lump ores are natural iron–rich materials, which are used directly from the
mine after the crushing and screening operations. Because the lump ores are
screened out at the mine, such operations generally produce lump ore as well as
(sinter) fines. Major lump ore deposits are present in Australia (Pilbara region),
South America (Iron Ore Quadrangle), South Africa (Sishen) and Canada
(Baffinland). In many other places limited amounts of lump ores are produced.
Good lump ores are becoming more and more scarce.
The lump ores demand generally a lower premium than pellets. For this reason,
in many blast furnaces, high amounts of lump ore are being considered. The
lower cost of the lump ore compared with pellets as a manufactured burden
component is offset by the poorer physical and metallurgical properties.
Generally speaking, in comparison with pellets, lump ores, which will also be
screened before transport at the dispatch port:
– Have a yield penalty where more fines need to be screened before charging to
the furnace, ranging from hardly any to over 25 % in case of severe handling
prior to charging.
– Some lump ores have a high moisture input into the furnace, either from
“intrinsic” moisture or because the lump was washed to remove fines.
– May have “Loss On Ignition” (LOI). The weight loss comes mainly from
crystalline water in goethite lump ores that is released as the lump ore is heated
in the furnace shaft. As a consequence of this crystalline water driven off at high
temperature, lump ores may show degradation, called decrepitation. The level
of decrepitation depends on the amount of crystalline water and the density of
the lump. A low LOI in a dense lump ore may show higher decrepitation than a
high LOI in a porous lump, where the crystalline water can more easily escape.
The disintegration effect is tested with the decrepitation test.
– May have poorer reduction degradation properties and or may have poorer
reducibility properties. Dense lump ores are generally stronger, but harder to
reduce. Poor reducibility is only translated into a coke penalty when the direct
reduction increases.
– Will have a lower melting temperature when they exhibit a higher natural acid
gangue content.
– Have greater diversity in physical properties due to being naturally occurring
and a broader sizing.
Consequently, lump ores may be penalized by an increased coke demand. Lump

ore is used in an appropriate size fraction, such as 8–40 mm. For blast furnace
operation at high productivity and high coal injection levels, lump ore may not
be the preferred burden material, but lump ore is a natural material, and its
properties will differ from type to type. Certain types of lump ores can compete
favorably with sinter, and for example the Ternium Siderar blast furnaces in
Argentina have operated successfully with up to 40 % in the burden using a
Brazilian lump ore at high furnace productivity.
3.7 Metallics charge and briquettes
Two other components with ferrous content can be charged to the blast
furnace:
– Metallics like Hot Briquetted Iron (HBI, Figure 3.11), Direct Reduced Iron
(DRI) and steel scrap to increase the productivity and efficiency of the blast
furnace whilst limiting the greenhouse gas exposure.
– In–plant revert as briquettes. Especially if an on–site sinter plant is not
available, valuable revert materials containing iron, manganese, carbon and
fluxes can be charged with the burden. The phosphorus level is the limiting
factor in many cases.
Figure 3.11 Hot Briquetted Iron (HBI)
Direct reduced iron (DRI) is seen more and more as a feedstock for blast
furnaces. After hot briquetting in the form of HBI it can be charged with the
other burden components, and with this HBI addition the hot metal production
is increased for situations where a mill is hot metal short. An advantage may be
secured in this way over charging these metallic directly into the converter.
Typically, HBI contains over 90 % Fe and has a metallization degree in excess

of 90 %. Since the metallics only need to be melted and hardly reduced, a coke
rate decrease per tonne of hot metal is achieved, and this helps to lower the
carbon dioxide generation at an integrated steel plant. Table 3.4 on the next
page provides typical compositions of HBI in comparison with typical sinter
and pellets.
42 Chapter III
Fe FeO Met.Fe SiO2 CaO Al2O3 MgO C

HBI 92 15 80 2.0 0.9 0.8 0.3 1
Sinter 58 7 0 4.5 9.2 1.7 1.1 0
Pellet 65 0 0 4.5 1.0 0.4 0.4 0
Table 3.4 Typical chemical composition of HBI versus sinter and pellets
In our experience, if 100 kg HBI/DRI/Steel Scrap is used per tonne of hot

metal, the productivity increases by 4–6 % and the coke rate can be decreased
by 30 kg/tHM.
Cold briquetting of in–plant reverts, such as blast furnace dust, BOF sludge
and mill scale, is a way to recycle these materials. The preferred way is to recycle
them through the sinter process. In North America, some sinter plants only
operate on these waste materials, whilst other locations lack such sinter capacity
and need to landfill them, sell them or recycle them via the briquetting route.
These cold bonded briquettes often have cement as a bonding agent resulting in
high strength cold properties but poorer reduction disintegration characteristics.
Cold briquettes offer the option of incorporating carbon, which improves
reduction properties.
3.8 Interaction of burden components
The results of burden tests on the total burden can differ greatly from results
on sinter, pellets and lump ore alone. An example is provided in Figure 3.12. A
relatively poor quality of lump ore is blended with good sinter. It is shown that
the behavior of the blend is better than expected from the arithmetic mean of
the data.
3.0
sinter
(mm water gauge/column height)
60/40 blend
2.5
Pressure difference Δp
lump
weighted average
2.0
1.5
1.0
0.5
0
900 1000 1100 1200
Temperature (°C)
Figure 3.12 Softening behavior of sinter, lump ore and a 60/40 blend, softening is
indicated by an increase in the pressure difference
(from Chaigneau – 5t ECIC, Stockholm, 2005)
Example 3.1 Effect of blending sinter and pellets in a blast furnace

In a blast furnace (1880 m³ inner volume) the pellets were blended over the
full length of the conveyor belt. Prior to dedicated efforts to make the ferrous
blend consistent there were no strict requirements on mixing ratio of burdens
on belt. In that situation gas flow was unstable, especially when lump rate
was increased. It was impossible to reach a lump percentage above 15 %. The
ferrous blend was made more consistent by mixing pellets and lump over a
much larger length of the sinter charge on the main conveyor belt. Pellets and
lump were no longer charged in the center area. The blast furnace showed the
following operational results.
(1) Central gas flow was more stable after the change. This is indicated by Z value
in Figure 3.13 (Z is central temperature divided by average temperature).
12.0
10.0
8.0
6.0
Z
4.0
Improved blending
2.0
0
Jan 2012
Jul 2012
Jan 2013
Jul 2013
Jan 2014
Jul 2014
Jan 2015
Jul 2015
Jan 2016
Jul 2016
Figure 3.13 Effect of improved blending on central gas flow stability
(2) Lump rate was increased gradually, max up to 21 % in January 2016, as

shown in Figure 3.14.
25
20
Lump ore rate (%)
15
10
0
Jan 2015
Apr 2015
Jul 2015
Oct 2015
Jan 2016
Apr 2016
Figure 3.14 Lump ore rate, January 2015 to March 2016

44 Chapter III
Generally speaking, blending materials dilutes the disadvantages of a certain

material. For this reason, the blast furnace burden components have to be
properly blended when charged to the furnace. More details on softening and
melting can be found in Chapter X.
In this respect, burden components should be chosen to be either comparable in

composition, like fluxed sinter with fluxed pellets to yield one desired softening
and melting outcome, or one should choose compatibility with for example a
very distinct super–fluxed sinter in combination with acid pellets. Super–fluxed
sinter has a high softening temperature and keeps the cohesive zone permeable
for gas to high temperatures. In Section 10.9.6 scenarios for using pellets at high
pellet operation are discussed.
The permeability of the cohesive zone for gas can also be improved by adding
different types of material to the ferrous burden. Nut coke remains solid and is
used by most steel plants blended into the ferrous burden; a similar effect is also
generated by using scrap and/or HBI, since these materials melt as well at high
temperatures.
3.9 Chemical control of the burden
Though the aim of a blast furnace is to make iron, the importance of all other
components coming with the burden as discussed in Section 3.2 cannot be
neglected and a certain burden composition is required to achieve a balance
between these components. The vast majority will be tapped as slag during a hot
metal tap, though some will partition between slag and hot metal, such as silica,
manganese and sulfur and others, like phosphorus, will completely revert to the
hot metal phosphorus.
Silica, together with alumina represents in quantity the bulk components

originating from the ore gangue and coal and coke ash. They need to be fluxed
with CaO and some MgO to achieve the slag chemistry for good liquid slag
and appropriate desulfurization. A low slag volume is desired but depends on
the quality of the raw materials available. Beneficiation of the raw materials
to a desired low gangue and ash levels at the mine site is not always possible
or comes with increasing cost and yield losses. Furthermore, some gangue is
required to attain a proper sinter quality. Operators will not go below a 3.5 %
silica level in their sinter, with 4.5–5.5 % being more common.
With pellets, a good quality can still be obtained with low gangue levels, and
consequently blast furnaces operating with 100 % pellets can achieve the lowest
slag volumes per tonne of hot metal. This is one of the merits of low silica pellets
in lowering slag volume and enhancing stable and high productivity with low
coke rates and high PCI levels
As an example: what does it mean when the burden contains 1 % more

silica (SiO2)? Since we charge about 1600 kg burden per tonne of hot metal,
1 % additional silica means 16 kg of silica contribution, which for such an
addition all ends up in the slag. Since slag contains 35–40 % SiO2, the slag
volume increases by a factor 2.5–3 or 40–50 kg/tHM, as also extra fluxes have
to be charged. So, the slag volume increases by 40–50 kg/tHM, coke breeze
consumption in the sinter process increases by 1 kg per tonne of hot metal and
coke consumption in the blast furnace increases by 2–2.5 kg/tHM, while blast
furnace productivity will decrease.
The amount of alumina in slag is generally limited to levels below 18–20 %,

though some operators have learned to exceed these boundaries. If the burden
composition does not allow for lower alumina levels, the only solution is to
increase the total slag volume with a siliceous component and dilute the high
level of alumina to acceptable levels. In this case the slag volume is dictated by
alumina rather than based on silica in the ores. Low alumina levels below 8 %
generate a difficult slag as well.
Slag MgO levels vary in the range of 5–12 %. The furnace process is more
forgiving if a range of 8–11 % MgO is maintained, if the increase in weight
percentage of MgO is compensated by a decrease in weight percentage of CaO.
When a desirable burden composition has been established, fine tuning can be
achieved by charging small amounts of high siliceous ore, quartzite, or direct
charge fluxes such as BOF slag, limestone or dolomite to the burden. The latter
two are carbonates, which require a considerable amount of heat in the blast
furnace for decomposition, which preferably should take place in a sinter plant
with coke breeze. Daily operational control is further discussed in Chapter X.
IV Coke
4.1 Function of coke in the blast furnace
Coke is a strong, non–melting material which forms lumps based on a structure

of carbonaceous material internally glued together (Figure 4.1). The average
size of the coke particles is much larger than that of the ore burden materials;
average size of coke is 45–55 mm and of ferrous burden materials 12–22 mm.
Coke does not melt and is a solid and permeable material up to very high
temperatures (above 2000 °C). This is of particular importance in the hearth
and melting and softening zone. Because of its large size coke is permeable for
gas and liquids.
Figure 4.1 Coke
For blast furnace ironmaking the most important functions of coke are:
– To provide the structure, through which gas can ascend and be distributed
through the burden supporting the total weight of the blast furnace content in
and above the cohesive zone allowing slag and iron to flow downwards to the
taphole.
– To generate heat to melt the burden.
– To generate reducing gases to remove the oxygen attached to iron in the ferrous
burden.
– To provide the carbon for carburization of the hot metal (also called dissolution
of carbon)
– To act as a filter for soot and dust.
48 Chapter IV
The permanent efforts aimed at reducing the costs of ironmaking have led
to an increasing portion of substitute reduction materials for coke, which
has mainly been coal injected through the tuyeres and recently natural gas
in North America. Nowadays, blast furnaces with total coal injection rates
in excess of 200 kg/tHM are operated with coke consumptions of less than
300 kg/tHM. At these high coal injection rates, coke is subjected to more
rigorous conditions in the blast furnace. Dissection of furnaces taken out of
operation, along with probing and sampling through the tuyeres of furnaces
in operation, have allowed assessment of the extent of coke degradation in the
furnace. Coke degradation is controlled by the properties of feed coke, i.e.
mechanical stabilization, resistance to chemical attack (solution loss, alkalis,
and graphitization) and by the blast furnace operating conditions.
In Figure 4.2 we show the residence time of coke in various zones. In the
“drying” zone, before the burden starts melting, the coke is exposed for 3–6
hours in the blast furnace gas flow. In the cohesive zone and the active coke
zone, the coke is present for 1–3 hours and is subjected to high temperature and
aggressive attack by chemical reaction (direct reduction reactions, alkali). In the
hearth the coke residence time is much longer, especially in the more central
part called “deadman”. There the residence time is from a week to two months.
As soon as coke reaches the area below the tuyeres, there is very little “chemical”
consumption of coke left nor is there a large dissolution of carbon in the hot
metal.
Figure 4.2 Residence time and quality parameters for coke in various zones in the
blast furnace
In this chapter we will discuss coke quality parameters, test methods,

degradation processes of the coke in the blast furnace, and finally the range of
coke qualities targeted by blast furnaces that are currently at, or are aiming to
operate at, the highest production levels, so are more demanding in terms of
coke quality.
Coke 49
4.2 Coal blends for coke making

The coal selected to make coke is the most important variable that controls
the coke properties. The rank and type of coal selected impacts coke strength,
while coal chemistry largely determines coke chemistry. In general, bituminous
coals are selected for blending to make blast furnace coke of high strength with
acceptable reactivity and at competitive cost. For the conventional recovery
coking process, the blend must contract or shrink sufficiently for easy removal
from the oven, and oven wall pressure must be acceptable to prevent coke oven
damage. For the heat–recovery method of coke making, these constraints are
not valid, which leads to an increase of usable coal types in this type of process.
Good coking coals are expensive and for this reason creating a coal blend that
gives good coke at minimum cost is specialist work.
Table 4.1 shows the typical chemical composition of coke that may be
considered to be of good quality.
Typical Coke Analysis % (db)

Coke Analysis Fixed Carbon 87–92
Nitrogen 1.2–1.5
Ash 8–12
Sulphur 0.6–0.8
Volatile Matter 0.2–0.5

(for well carbonised coke)
Hydrogen 0.2
Ash Analysis Silica SiO2 52.0
Alumina Al2O3 31.0

Iron Fe 7.0
Lime CaO 2.5
Potassium K 2O 1.8
Magnesia MgO 1.2
Sodium Na2O 0.7
Phosphorous P 0.3
Manganese Mn 0.1
Table 4.1 Coke chemistry for good coke quality
Fixed carbon content of coke Cfix can be estimated as follows (all percentages):
Cfix = 100 – ash – S – VM –N
VM is volatile matter, N is nitrogen, typically 1.3 % and often not taken into
account.
50 Chapter IV
Ash directly replaces carbon, so higher ash results in higher slag rate and lower
energy value. The increased amount of slag requires energy to melt and more
fluxes to provide a liquid slag. Ash, sulfur, phosphorus, alkalis and zinc can be
best controlled by careful selection of all coal, coke and burden materials. The
financial repercussions of ash, sulfur and phosphorus may be assessed by value–
in–use calculations for PCI coal, coking coal blends and burden materials.
Alkalis and zinc should remain below certain threshold levels (Chapter VI).
4.3 Coke quality
How to characterize coke quality? How to define and measure coke properties?
In other words, how to establish a target for coke manufacturing based on
determined coke properties in line with the needs of the blast furnace process.
From the above discussion, the following parameters should be considered to
limit the coke degradation and maintain suitable coke behavior in the blast
furnace, especially at high coal injection rates. Qualitatively the coke should:
– Be made up of large, stabilized particles within a narrow size distribution band.
– Have a high resistance against volume breakage.
– Have a high resistance against abrasion.
– Have a high resistance against chemical attack (CO2, alkali).
– Have a high residual strength after chemical attack.
– Have sufficient carburization properties (the dissolution of carbon in hot metal).
A more detailed description of coke quality tests is added in Annex VI,

summarized in Section 4.5.2.
4.3.1 Coke degradation mechanisms in the blast furnace

The basic concepts of coke degradation in the blast furnace, according to the
interconnected thermal, physical, and chemical conditions coke is subjected to
in the furnace, are described in Figure 4.3.
At the stockline, the coke is generally well stabilized due to handling to get it
to the furnace top. The effect of gasification on strength is controlled by the
mechanisms of the heterogeneous reaction. In general, diffusion is the limiting
step and the reaction is located at the surface of the lumps, the core remaining
quite unaffected. As gasification and abrasion proceed simultaneously, a peeling
of coke particles occurs (3–5 mm size reduction), leaving an exposed unreacted
core and fines.
Beyond gasification, coke reacts with alkali vapors when passing through the
alkali circulating zone, and the structure is penetrated by alkalis. This reaction
reduces the strength of the coke, making it more susceptible to size reduction
by breakage from mechanical action. Coke that has been already weakened,
on arriving in the high temperature zone of the raceway, loses its alkalis by
gasification. High temperature, mechanical action and graphitization bring
about severe degradation, decrease of size and the formation of fines.
Coke 51
Figure 4.3 Basic concepts of coke degradation in a blast furnace
The coke travelling to the deadman is exposed to moderate temperatures,

high alkalis during long periods of time, along with additional reactions
(reduction of slag, carburization) that mostly affect the surface of the coke
lumps. Deadman coke, sampled by core drilling, corresponds more or less to
the unreacted core of the initial lumps, and it is not surprising that it exhibits
similar strength to the coke that is charged at the top.
Example 4.1 Harmonic mean size and average mean size

The average mean size (AMS) is the weighted average of the size fractions, the
harmonic mean size (HMS) is the weighted average of 1 divided by the size
fractions and consequently places emphasis on smaller particles.
This is shown in Table 4.2. So, the average mean size correlates with volume
(weight) of a mixture, while the harmonic mean size correlates with the
average surface of a mixture.
Size (mm) 25 35 50
Fraction 0.2 0.3 0.5
Average Mean Size 40.5 0.225+0.335+0.550
Harmonic Mean Size 37.6 1/(0.21/25+0.31/35+0.51/50)
Table 4.2 Calculation of AMS and HMS

52 Chapter IV
4.3.2 Degradation of coke during its descent in the blast furnace

During its descent through the blast furnace coke, is subjected to the process
conditions and gradually degrades, as shown by the evolution of coke size
in Figure 4.4. Research has shown that the harmonic mean size (HMS), of
the coke mass yields the highest correlation with the resistance to flow of gas
passing through the coke bed.
Figure 4.4 Coke size in the blast furnace throughout the journey from the top to the
bottom of the furnace. HMS and AMS: Harmonic and Average Mean
Size. Data for temperature and ηCO for wall area (Chaigneau, 2001), coke
sizes as measured for operating furnaces. Dots represent feed coke.
Coke is exposed to stresses and (chemical) attack during its journey downwards.
1. Charging zone: Due to the fall of the coke onto the stockline, some breakage
and abrasion will occur during charging.
2. Granular zone: In this region the coke and ore remain as discrete particles
within their separate layers. Drying occurs, and recirculating elements such
as zinc, sulfur and alkalis, deposit on the burden materials as they descend to
the bottom of the granular zone. From a temperature of 900 °C coke starts to
oxidize with CO2, continuing to do so as the temperature increases to over 1000
°C. In this zone coke degradation (mostly abrasion) occurs due to mechanical
load and mild gasification.
3. Cohesive zone: This zone starts where ore agglomerates, begins to soften and
deform, creating a mass of agglomerated particles sticking together. This mass is
barely permeable, and the rising gas can only pass through the remaining coke
layers. Coke gasification with CO2 becomes significant due to increased reaction
rates at the higher temperature level (1000 – 1300 °C). The contact between
the softened or molten materials and the coke lumps becomes more intensive,
leading to increased mechanical wear on the outer surface of the coke particle.
The residence time within the cohesive zone is rather short (30 to 90 minutes)
depending on productivity and softening properties of the agglomerates.
Coke 53
4. Active Coke or Dripping zone: This is a packed bed of coke through which
liquid iron and slag percolate towards the furnace hearth. The coke particles
play an active role in further reducing the remaining iron oxides and increasing
the carbon content of the iron through dissolution of carbon from the coke
into the iron. The bulk of the coke arriving in this zone (also referred to as bosh
coke) flows towards the raceway region. The remaining part will move into the
deadman (also called inactive coke zone). The residence time varies from 1 to 3
hours. The temperature increases gradually from 1200 to 1500 °C.
5. Raceway: Hot blast containing oxygen is introduced through the tuyeres. The
kinetic energy of the blast creates a raceway (cavity) in front of each tuyere.
Coke particles circulate at very high velocity in this semi–void region while
being gasified together with injectants such as coal, oil and natural gas. A part
of the coke and injected reductants is not burnt completely. Soot is produced
during injection of coal and natural gas. Soot and dust are transported upwards
by the gas stream. They cover coke particles and react later following solution
loss reaction. They decrease the reactivity of coke and cause an increase in
apparent viscosity of liquid phases. The temperature increases rapidly to over
2000 °C due to the exothermic oxidation of coke and injectants. Coke fines
and injectant fines that are generated in the raceway either completely gasify
or get blown out of the raceway into the coke bed. Coke and coal fines may
accumulate directly behind the raceway, forming an almost impermeable zone
called the bird’s nest. Observations of the raceway were made in blast furnaces
in operation by inserting an endoscope through a tuyere. These observations
showed that in this zone the coke is subjected to very severe conditions.
6. The Hearth: Since the rate of coke consumption is highest in the ring of the
raceway, an almost stagnant zone (not directly feeding the raceway) develops
in the furnace center. This zone is called the deadman, and is thought to have
a conical shape and a relatively dense skin structure. Molten iron and slag
accumulate throughout the structure before being tapped through the tapholes.
Tracer experiments in a German furnace gave values in the range of 10 to 14
days, but residence times of 60 days are also mentioned in the literature for the
deadman coke.
The above mentioned processes are summarized in Table 4.3 on the next page.
At high coal injection rates the amount of coke present in the furnace decreases
and the remaining coke is subjected to more vigorous mechanical and chemical
conditions: increased mechanical load as the ore/coke ratio becomes higher;
increased residence time at high temperatures; increased solution loss reaction
(CO2, liquid oxides); and alkali attack. More severe coke degradation during its
descent from the furnace stock line into the hearth can therefore be expected at
high coal rates.
54 Chapter IV
Blast Furnace Function of Coke Coke Degradation Coke Requirements

Zone Mechanism
Charging Zone – Impact Stress – Size Distribution
– Abrasion – Resistance to
Breakage
– Abrasion Resistance
Granular Zone – Gas permeability – Alkali Deposition – Size & Stability
– Mechanical Stress – Mechanical Strength
– Abrasion – Abrasion Resistance
Cohesive Zone – Burden support – Gasification by CO2 – Size Distribution
– Gas permeability – Abrasion – Low Reactivity to CO2
– Iron and slag – High Strength after
drainage Abrasion
Active Zone – Burden support – Gasification by CO2 – Size Distribution
– Gas permeability – Abrasion – Low Reactivity to CO2
– Iron and slag – Alkali attack and – Abrasion Resistance
drainage ash reactions
Raceway Zone – Generation of CO – Combustion – Strength against
– Thermal Shock Thermal Shock and
– Graphitisation Mechanical Stress
– Impact Stress and – Abrasion Resistance
Abrasion
Hearth Zone – Burden support – Graphitisation – Size Distribution
– Iron and slag – Dissolution into – Mechanical Strength
drainage hot metal – Abrasion Resistance
– Carburisation of iron – Mechanical Stress – Carbon Solution
Table 4.3 Coke functions, degradation mechamisms and requirements
However, high coal injection rates can also affect the direct reduction reactions.
1. Coal injection increases hydrogen content and at elevated temperatures (800–
1100 °C), hydrogen is a very effective agent in gas reduction of iron oxides.
2. The unburnt coal char remaining after the raceway is more reactive than coke
and used for direct reduction in preference to coke.
3. The alkali cycle is reduced as a consequence of the elimination of alkali through
the hot furnace center.
So, at high coal injection rates the attack of coke by direct reduction reactions
and by alkalis may also decrease. This is beneficial for coke integrity in the
lower part of the furnace.
4.4 Coke size distribution
The shape of the coke particles and the size distribution of the particles are the
decisive factors for the permeability of the coke bed, for ascending gas as well as
for the descending liquids.
The lowest flow resistance is obtained when large coke is being used of high
uniformity. Fines in particular have a strong decreasing effect on the harmonic
mean size, which increases the bulk resistance of the coke. Coke screen aperture
size is generally between square holes of 24 and 35 mm, although on some
plants this is as high as 40 mm. The size selected depends on the screening
Coke 55
efficiency and how well the coke has been screened before arriving at the
stockhouse. The objective is to remove the fine coke. Typical specifications for
coke after screening is maximum 2 % smaller than 25 mm or maximum 3 %
smaller than 30 mm. The undersize after screening is partly used in ferrous layer
as nut coke, where it replaces the coarser coke. Details in Section 4.8.
Once the bulk coke has been classified by screening and crushing (see also
Figure 4.4), the aim is to have a resultant coke with high mechanical strength
under the blast furnace conditions. This is to prevent an excessive formation of
coke fines during its descent in the blast furnace.
4.5 Strength of coke
4.5.1 Coke particle size during the blast furnace process

During carbonization in a coke oven, fissures in the coke are generated due to
stresses that arise from the differential contraction rates in adjacent layers of
coke, which are at different temperatures. They are typically longitudinal, that
is perpendicular to the oven walls. Additionally, many transverse fissures are
formed during pushing. These fissures determine the size distribution of the
product coke by breakage along their lines during subsequent handling. But not
all the fissures lead to breakage at this early stage, and a number of them remain
in the coke particles. The initial coke distribution is a function of the coal blend
and the coking conditions. A significant number of internal fissures remain
present and cause further degradation under mechanical loads during transport
and charging of the blast furnace. This process of coke degradation is called
stabilization. Stabilization lowers the mean size of the coke, but the resulting
particles are less prone to further breakage. For blast furnace performance,
it is not only important to have large, stabilized and narrow size distribution
coke charged into the furnace, but it is even more important to have the
same qualities present during its descent through the furnace as well. With
mechanical handling, coke particles will degrade due to breakage and abrasion.
Breakage is the degradation of coke by impact due to fissures already present
in the coke. Abrasion is the degradation of the surface by relatively low impact
processes (rolling and sliding against other particles or the furnace walls). It is
one of the main mechanical processes for decreasing the coke size below the
stock line, next to breakage in the raceway area. Abrasion causes the formation
of fines which may hamper blast furnace permeability.
The resistance to abrasion will deteriorate in the blast furnace, due to reactions
such as graphitization, gasification and carburization of the iron. Graphitization
results in a more crystalline form of carbon in the coke that is more brittle. In
Figure 4.5 the typical development of the HMS of coke from the coke wharf to
the tuyeres is presented.
56 Chapter IV
60
Belt coke average size (as HMS, mm)

50 Wharf
CP screen
BF screen
40
BF top
30
Tuyere
20
0 50 100 150 200
Cumulative drop (m)
Figure 4.5 Development of Harmonic Mean Size after mechanical handling in the
form of drops between conveyors and screens
In the transport route presented, the coke is screened at 35 mm (square) at the

coke plant and at 24 mm (square) at the blast furnace. The increase in HMS of
the sample after screening is due to the removal of the undersized coke.
4.5.2 Coke cold strength simulation tests

Although it is known that coke degrades more rapidly at high temperatures,
there is no test in practical use that is performed at high temperatures. Not only
because of the complexity and high costs, but also because it has been shown
that coke with poor low–temperature strength also exhibits poor strength
at high temperatures. For this reason, most tests in practical use are done at
ambient temperature.
Coke strength is traditionally measured by empirical tumble indices. During

mechanical handling, coke size degradation takes place via two independent
processes, these being breakage into smaller lumps along fissures and cracks
still present in the lumps, and abrasion at the coke surface resulting in small
particles (smaller than 10 mm). So, it is common to measure a ‘strength’ index
related to degradation by volume breakage, for example, I40, M40, stability index;
and an ‘abrasion’ index, for example, I10, M10, D150. These empirical indices
cannot be directly related to fundamental coke properties.
Figure 4.6 shows a schematic representation of particle motion in a tumble

drum. As lifter bars inside the drum rotate, a portion of the coke is lifted off
the walls of the drum. Some of the coke rolls off the lifter before it reaches
the horizontal plane. The coke that is not picked up slips and rolls against the
bottom of the drum. The coke that is lifted past the horizontal is dropped over a
fairly narrow angular range as the lifter approaches the vertical plane.
Coke 57
Figure 4.6 Motion of coke in a tumble test
This coke impacts with the bottom of the drum. Tests have shown that there is
a relationship between the degradation of coke in a drum test and that after a
number of drops. This makes it possible to translate the effect on coke size after
a number of drops, in meters, into a number of rotations in a drum, and vice
versa.
4.5.3 Coke hot strength and chemical reactivity
Besides a high mechanical strength, coke should have a high resistance against
chemical attack. There are two measurements for coke after the reaction with
CO: the CRI (Coke Reactivity Index) and the CSR (Coke Strength after
Reaction).
Coke Reactivity Index

Reactivity of coke can be tested in numerous ways, but by far the most common
way to determine coke reactivity is the Nippon Steel Chemical Reactivity Index
(CRI). With this test, a sample of dried coke of 19–22,4 mm is put in a 100 %
CO2 atmosphere at 1100 °C. The percentage of coke that is gasified after 120
minutes yields the CRI value. The more reactive the coke, the higher the mass
loss will be. The reactivity of the coke is mainly determined by the chemical
composition of the parent coal blend, because ash components act as catalysts
for the reaction of C with CO2.
Coke Strength after Reaction

Due to the loss of mass whilst under attack by CO2, the surface layer of the coke
particles becomes very porous and the mechanical strength against abrasion
drops rapidly. To measure this effect the reactivity test is normally followed
by a tumbler test to determine the residual coke strength. The percentage of
particles that remain larger than 10 mm after 600 rotations is called the ‘Coke
Strength after Reaction’ or CSR index. For most coke produced there is a strong
correlation between CRI and CSR. As CRI decreases, CSR increases.
58 Chapter IV
Before CRI and CSR were developed, a series of relatively expensive tests were
carried out under various research projects that involved partially gasifying
the coke in its original particle size under realistic blast furnace conditions
before subjecting it to the standard drum test. While the results of this costly
research work showed exactly how the coke in the blast furnace was subjected
to chemical attack, it provided no better information on coke quality than
the simpler method of determining CRI and CSR. These two parameters are
now generally adopted by the coke–making industry as the most important
parameters for determining coke quality, especially on large blast furnaces with
high productivity and high tuyere injectant rates.
Carburization of Hot Metal

There is no standard test for the dissolution of carbon in hot metal, the
carburization. Experiments were conducted on this by the Institute of Ferrous
Metallurgy in Germany to compare different cokes of different coal blends and
coke making technologies. The experiments showed very similar
behavior between most cokes. The only exception was the traditionally
produced beehive coke. Although it had a very good CSR and CRI it was the
only coke examined that cannot be used alone in a blast furnace because of its
poor carburization characteristics. Production trials prove that this type of coke
can only be used in a mixture with other more reactive coke.
4.6 Coke deadman
The coke “deadman” is a cone–shaped coke layer, located below the cohesive
zone and beyond the active coke zone (Figure. 4.7). The coke in this zone comes
from the coke charged into the center of the furnace. This coke is exposed to
no or only very limited attack by carbon dioxide, and consequently its strength
properties are not degraded. It maintains its size as charged more or less,
affected only by slight abrasion on its way down through the furnace. So, coke
lumps entering the deadman are bigger than those in coke layers more towards
the wall. Some companies use stronger and/or larger coke in the center to
enhance deadman permeability.
The main requirement for the deadman is to have sufficient porosity to ensure:
– From the level of the tuyeres upwards: unrestricted passage of both the gas from
the raceway and sufficient permeability for the dripping iron and slag.
– From the level of the tuyeres downwards: sufficient permeability for slag and
metal liquids on their way to the hearth.
– Within the hearth, below the taphole level: the deadman fills or at least mostly
fills the furnace hearth sump. So, its permeability for hot metal and slag flowing
towards the taphole determines the flow pattern of the liquids.
Coke 59
Figure 4.7 The coke deadman and the effect of a permeable deadman and
impermeable deadman on hot metal flow lines towards the taphole
Figure 4.7 shows two patterns of hot metal flow in the hearth: flow through the
deadman (left) and flow around the deadman (right). When the flow through
the deadman is blocked, then the flow will take place along the wall of the
furnace. This will lead to increased wear of hearth refractory lining and the
hearth refractory reaches its end of campaign faster. Another possibility is for
the deadman to float when it becomes dirtied. In this situation, a coke–free
layer of limited depth will develop across the full furnace diameter, and within
that layer flowlines for the metal towards the taphole will also be relatively
direct. The tendency for the deadman to float is determined not only by its
permeability but also by furnace geometry. Floating is more likely when the
hearth sump is deep.
60 Chapter IV
Example 4.2 Deadman cleanliness

Drainage capacity of the deadman can be quantified using a Deadman
Cleanliness Index (DCI). The DCI is based on the observation that in the
case of prolonged contact time between the metal and coke (both when
dripping and when within the hearth pool), the carbon content of the hot
metal approaches the saturation equilibrium value (Nightingale, 2000;
Sergeant et al, 2005). The dirtier the deadman, the closer the approach to
equilibrium becomes.
If, however, the deadman is clean, a high permeability coke bed allows faster
passage of the metal, and less dissolution of carbon from the coke results in
a greater (sub–saturation) departure from equilibrium carbon concentration
in the metal. The calculated equilibrium value of carbon in hot metal can be
estimated from the following formula:
%Csat = 1.3 + 2.5710 – 3HMT – 0.31%[Si] – 0.33%[P] – 0.4%[S] +

0.028%[Mn]
The sub–saturation departure from equilibrium ΔC is calculated as follows,

where the actual carbon value must be determined from metal:
ΔC = %Csat –[C]actual
If the difference between equilibrium and measured C is larger than 0.2 %,

then the deadman is likely to be clean. If lower, then the deadman is likely to
be “dirty” or less permeable. A value for real concern is below 0.1 %.
Some companies use the deadman cleanliness index (DCI). The DCI is
determined from the following expression

1
DCI = HMT + ΔC − 1430 − 190 × 1.23 − C/S
2.57 × 10−3
The last (bracketed) term recognizes the requirement for the deadman to
allow drainage of slag as well as the metal. That term was formulated for the
slag properties at Port Kembla, Australia and may not be strictly applicable to
other sites. If in doubt about this, it may be safer to make evaluation on the
basis of ΔC alone.
For furnaces having a volume from 2000 to 4300 m³, using natural gas
injection and PCI, DCI varies within the range of 150–250 with the higher
values resulting in better drainage capacity of the deadman.
Coke 61
As a result of carbon consumption and dissolution of coke ash in slag, coke

lumps gradually decrease in size and are completely consumed. There is a slow
and continuous replacement of coke in the deadman, which is supplemented
with fresh coke descending from the top. Residence time of coke in the
hearth is estimated to be about 2–3 weeks and depends on the blast furnace
volume and performance.
The inactivation of the deadman (also called “dirty deadman”) can be

identified from at least the following symptoms (Raipala, 2003):
1. Decrease of central temperature in the hearth bottom.
2. Hearth wall temperature increase.
3. Oxygen potential increase in slag (FeO and MnO in slag), resulting in poorer
desulfurization of the hot metal.
4. Hot metal carbon decrease
5. Slag tapping ratio decrease
6. Hot metal temperature is higher than usual with the same energy
consumption and hot metal composition
7. Shortening of the raceway can be observed
8. Shortening of the taphole length
Causes of inactivation of the deadman can be attributed to:

1. Coke properties: coke particles are exposed to mechanical stress in the stack
and in the lower part of the blast furnace. Large and strong coke results in
larger particle size and less fines in the hearth. Coke lumps are weakened in
the stack by solution loss reaction (C + CO2 2 CO) and the impact on
various coke qualities depends on its hot strength such as CSR/CRI.
2. Water leakage: Small leakages (tuyeres, cooling plates, stave circuits),
especially when they remain undetected, may lead to an inactive deadman
when water does not evaporate completely but seeks its way to cooler parts in
the furnace bottom.
3. Maintenance stops (shutdowns): when the hot metal below the taphole level
solidifies it has a large mass and good thermal conductivity to the cooled
bottom. Excessive hearth cooling systems increase the risk of inactivation
during blast furnace stop.
4. Low production rate: when the production rate is low, the residence time
of coke in contact with CO2 and alkalis increases and the coke properties
weaken. Heat losses through the hearth lining are relatively constant at the
beginning of the slow production period. Heat transport to the deadman
slows down and risk of solidification at the bottom increases.
5. Hearth cooling: hearth cooling itself is an essential factor in problems with
an inactive deadman. Because the clogged or dirty deadman is maintained by
solidified matter, it requires cooling to keep the temperature below melting
points.
62 Chapter IV
Measures to activate a deadman include:

1. Increase the coke quality (no small coke in the burden).
2. Charge large size coke (larger than 60 mm) in center.
3. Increase coke rate and reduced PCI rate.
4. Reduction of hearth bottom cooling.
An example of deadman inactivation is shown in Figure 4.8, where a

period with poor coke quality coincided with low values for the DCI. The
effects were also visible in poorer casting behavior as well as higher sidewall
temperatures.
1. Irregular coke quality (CSR) 2. Not permeable dead man (DMI)
3. Difficult to cast slag from one side 4. Hearth temperature differences
Picture shows the temperatures differences in a period of time (AT)
Figure 4.8 Poorer coke quality and effect on Deadman Cleanliness Index
(DCI), showing increased sidewall temperatures and poorer
casting on one taphole
Coke 63
4.7 Overview of international quality parameters

Figure 4.2 and Table 4.4 provide an overview of typical coke quality parameters
and their generally accepted levels for a ‘good’ coke quality. Although not
complete, the values given in the table represent coke qualities that have assisted
in securing excellent blast furnace results over a long period.
What is Results Accept. Best Reference

measured? Range
Mean Size Size Distribution AMS mm 40–60
HMS mm 35–50
% < 40 mm < 25
% < 10 mm <2
Cold Strength Size Distribution I40 % > 40 mm > 45 60 Irsid Test
after Tumbling I10 % < 10 mm < 20 16
M40 % > 40 mm > 80 87 Micum Test
M10 % < 10 mm <7 5.5
Stability at Wharf % > 1” > 58 ASTM Test
Stab. at Stockh. % > 1” > 60
Hardness % > ¼” > 70
Strength after CSR % > 9.52 mm > 58 70 Nippon
reaction Steel Test
Reactivity CRI % weight loss < 29 22 Nippon
Steel Test
Table 4.4 Acceptable range for coke quality parameters
Which is the most important quality parameter?
Different operators have their own preferences. A Russian ‘rule–of–thumb’ table

gives a strong relationship between M10, M25 and fuel rate and productivity
(Danshin and Chernousov, 1989), whilst a German review (Grosspietsch and
Lüngen, 2000) demonstrates the impact of CSR on fuel rate, achievable PCI
rate and productivity. Strong relationships have been observed between I10
and fuel rate. It can be argued that size, cold strength and hot strength are
all important because they all impact on the permeability of the coke that is
required for both liquid drainage and for efficient gas/solid contact, with I10
and CSR probably the two parameters most often cited as the most critical.
Hearth voidage is more likely to be affected by CSR than I10, because the fines
generated by abrasion should be preferentially consumed by direct reduction
and not reach the hearth.
With high PCI rate, the coke spends longer (more than 20 %) in the furnace
above the raceway due to its slower consumption rate. It is therefore exposed to
process conditions for longer. Once in the raceway, coke burns more slowly and
tumbles around for longer, generating more fines, which are collected in the
“bird’s nest”. Hence the need for coke of higher cold and hot strength to permit
stable operation at high PCI rates.
64 Chapter IV
Example 4.3 Consistency of coke quality, coke from stock

We have to stress that blast furnace operation is very much influenced by coke
variability: the gas flow in the furnace can only be held consistent if the layer
build–up is consistent and if day–to–day consistency of the coke is very good.
There are, however, no international standards or criteria for day–to–day
consistency.
The variability of coke quality has not been investigated in detail. Some
companies use their own “fresh” coke and purchased coke. It is inevitable
that even purchased coke with very good metallurgical properties behaves
differently to in–house coke. Figure 4.10 shows “fresh” coke (left) and
purchased coke. The purchased coke has suffered from abrasion and fine
particles have disappeared. It looks more rounded. As will be described in
the burden distribution section, the stockyard material will have different
properties with respect to the angle of repose and also different flow
characteristics from the top bin into the furnace.
Figure 4.9 Fresh coke and coke from stockyard compared

Coke 65
4.8 Coke, nut coke and high PCI

For cost–efficient operation the undersize of the coke after screening has to be
reused. Some of the undersize can be used as coke breeze for sinter production,
sometimes a small amount can be used for pellet production. The balance has
to be efficiently used in blast furnaces (Dharm Jeet Gavel, 2020). Depending
on the local situation, operators can choose to put more small coke in smaller
furnaces, or screen coke at a smaller aperture in the smaller furnaces.
If the operational principle is that all coke produced at site has to be used at
site, what is then the optimum for screening? An example is shown in Table 4.5,
where the amount of coarse metallurgical coke is shown as well as the amount
of nut coke in relation to the screening aperture.
size (mm) > 125 > 100 > 90 > 75 > 63 > 50 > 40 > 35.5 > 30 > 25 > 20 > 10 < 10
fraction 0% 0% 1% 4% 7% 17 % 24 % 17 % 13 % 8% 6% 2% 1%
kg/tHM 4 12 22 50 72 50 40 25 17 6 2
40 mm Metallurgical Coke: 70 % Nut Coke: 30 %
screen

Table 4.5 Coke fractions and screening aperture; data are from an operating
furnace and total coke rate, coarse and nut coke, is 300 kg/tHM
With increasing screening aperture,

– the coarse coke layer becomes thinner,
– the coarse coke layer becomes more permeable
– more nut coke is generated, which has to be blended into the ferrous layers.
250
200
Coke voidage
(dm³/tHM)
150
100
40 35 30 25 20
Coke Screen Size (mm)
Figure 4.10 Coke voidage per tonne of hot metal (in dm³/tHM) increases with finer
screening because of increased layer thickness
66 Chapter IV
What is the optimum? Nut coke is very useful since it makes the cohesive zone
more permeable for gas, but excess nut coke is not completely consumed in the
furnace and will reach the hearth, thus worsening hearth permeability. So, nut
coke has to be consumed at or above the tuyere level.
The melting of the ferrous burden requires that hot gas transfers its heat to the
softening and melting materials, where it is mainly used to drive the direct
reduction, and only a small part of the heat is used to heat up the materials. For
this reason, gas permeability of the coarse coke layer in the cohesive zone is very
important for melting. The thicker the coke layer per tonne of hot metal, the
more voidage and the more permeable the cohesive zone (Figure 4.10).
At high PCI rates (over 200 kg/tHM) and low coke rates, the permeability
of the coarse coke layer becomes critical, and so the nut coke should be at a
reasonable level of 20–40 kg/tHM. In other words, if more nut coke is available,
the coarser fractions should be considered coarse coke. Nut coke should also
be distributed in the ferrous burden in the wall area because this part of the
ferrous layer is more difficult to melt than –more inwardly located material. In
addition, nut coke provides greater benefit because it improves gas permeability
and melting. Moreover, if the nut coke is not fully consumed by direct
reduction, it will probably be gasified at the tuyeres.
Example 4.4 Blast Furnace operation: redistributing nut coke to wall area
In a blast furnace (1880 m³ inner volume) nut coke was previously distributed
over 30 % of the length of the ferrous burden on the conveyor belt. The nut
coke was redistributed to reach the two outer rings with the ferrous burden.
As a consequence, fluctuations of stave temperatures decreased as shown in
Figure 4.11 before and after change to more nut coke in wall area. Similar
results were observed for variation of cooling water temperatures, which
correspond with heat losses (not shown).
Figure 4.11 24 hour graphs of stave temperatures (rows 6 and 7) – nut coke being
distributed in central area (left image) and wall area (right image)
V Injection of Coal,
Natural Gas and Oil
The coke rate of a modern, big blast furnace operating with only coke as a
reductant (fuel) is typically 500–520 kg coke per tonne of hot metal. The
coke rate can be decreased by injection of coal, oil and natural gas (and other
hydrocarbons), as was suggested already back in 1831 by John Samuel Davis.
Use of natural gas injection started in 1957 in the Ukraine, with the technology
spreading quite fast, especially in the USSR. Starting in the 1960s fuel oil
injection was implemented, while industrial coal injection started at Amanda
furnace in the USA in the 1960s. The technology of coal injection spread quite
fast especially from the 1980s. The recent low cost of natural gas has led to the
application of co–injection of coal and natural gas, especially in North America.
Since coke is a major part of the hot metal cost, use of injectants is economically
attractive. Coke rates can be decreased by using injectants:
– By using coal injection, to 320 kg/tHM and even lower to 250 kg/tHM.
– By using natural gas: to 350 kg/tHM, but more typically about 400 kg/tHM.
– By using fuel oil: to about 375 kg/tHM.
The investments for using coal injection are considerable, while the investments
for gas or oil injection are low. The relative price of oil is often too high to
consistently be an attractive alternative.
Figure 5.1 Schematic overview of energy inputs and outputs

68 Chapter 5
The apparent versatility of the blast furnace for the use of various type of
injectants comes from the fact, that in the raceway the temperatures are so
high that the injectants and moisture are all converted to carbon monoxide and
hydrogen within milliseconds after injection, and that the furnace “does not
know” where the hydrogen and carbon monoxide are coming from.
5.1 Properties of coal, oil and gas
The energy inputs and outputs of the blast furnace are shown schematically
in Figure 5.1. The major sources for energy in the furnace are the coke and
injectants (coal, gas, oil), and the sensible heat of the hot blast. The main part
of the energy is used to drive the change from iron oxides to iron and the other
chemical reactions. The remaining energy leaves the furnace as top gas, as
sensible heat of iron and slag and as heat losses to the cooling system.
The differences between the various injectants are based on their chemical
composition. The carbon in the injectants becomes carbon monoxide (CO)
and the hydrogen becomes molecular H2. So carbon content, hydrogen content
and the presence of water and oxygen mainly determine the amount of heat
generated in the raceway. When comparing the injection of coal (common),
natural gas (less common) or oil (outdated), it is found that coal generates most
heat: when burning coal with pure cold oxygen to CO and H2, it generates
a flame temperature (Tflame) of around 2000 °C, while for natural gas this is
around 850 °C (Table 5.1).
Tflame (°C) H2 in gas CO in gas
coal ~2000 25 % 75 %
natural gas 850 66 % 34 %
oil 1350 45 % 55 %
Table 5.1 Tflame and gas composition for incomplete combustion of coal, gas and oil
to CO and H2 (stoichiometric)
The generated heat is used in and below the cohesive zone to:
– Heat up and melt the material from where it starts softening, about 1200 °C, to
casting temperature of 1500 °C.
– Remove the remaining oxygen from the ore burden and other hot metal
components like silicon; this chemical reaction costs a lot of energy/heat and is
called “direct reduction” (see Section 8.3.3).
The largest amount of energy (more than 80 %) is used for this direct reduction.
However, the amount of direct reduction that has to take place also depends on
the amount of hydrogen generated at the tuyeres, since hydrogen is very efficient
in removing oxygen from the iron at high temperatures (900–1100 °C). As a
consequence, the most important properties of injectants are: the amount of
heat generated when gasified to carbon monoxide and the hydrogen content of
Injection of Coal, Natural Gas and Oil 69
the gas (Table 5.2). Another effect is that the gas volume generated at the tuyeres
increases when more hydrogen is present. This is shown as the gas volume
generated by gasifying 1 kg of injectant with oxygen in Table 5.1. The amount
of injectant needed to replace 1 kg of coke is indicated in the table as well as the
effect of an additional 10 kg/tHM injectant on the flame temperature.
C H Ash Heat Gas RR Tflame
% % % MJ/kg m³ STP/kg kg/kg °C

Coke 86.0 0.2 11.0 8.5 1.6 1.00
Coal 80.0 4.4 8.8 6.7 2.0 0.88 –27
Natural Gas 72.6 23.5 0.0 2.3 4.0 0.97 –76
Oil 87.0 12.0 0.0 6.5 3.0 1.09 –46
Table 5.2 Properties of injectants: Heat of incomplete combustion to CO and H2

of various injectants, gas volume generated (complete gasification with
pure oxygen), replacement ratio (RR) relative to coke and Tflame effect if
injection is increased by 10 kg/tHM. Estimated for the reference furnace.
The consequence is that the lowest coke rates can be reached with coal injection.
Table 5.3 shows worldwide experience with different injectants, based on about
70 % of the maximum reached. Good, but not excellent, levels of injection are:
170–200 kg/tHM for coal, 80 kg/tHM for gas and 90 kg/tHM for oil. Note
that when using natural gas, higher oxygen enrichment of the blast is required
to keep the flame temperature in the required working area. The minimum
annual average coke rate, which includes coarse metallurgical coke and nut
coke, is 250–260 kg/tHM, of which at least 210–230 kg/tHM metallurgical
coke, depending on slag volume (200–300 kg/tHM). The highest PCI rate
reported is 252 kg/tHM (Lüngen and Schmöle). The maximum natural gas rate
reported has been 130 kg/tHM, enabled by 9 % oxygen enrichment.
All Typical Max. Typical Max. Oil

coke PCI PCI Gas Gas
Coke kg/tHM 517 300 256 399 354 375
Coal kg/tHM 200 252
Oil kg/tHM 90
NG kg/tHM 80 130
enrichment % 0 6.1 9.1 7.1 11.1 3.1
Gas make GJ/t 5.4 5.3 5.5 5.7 6.6 5.1
Stoves gas GJ/t 1.9 1.8 1.6 1.7 1.6 1.9
Export gas GJ/t 3.5 3.6 3.9 4.0 5.0 3.3
CV BF Gas MJ/m³ STP 3.1 3.7 4.1 4.0 4.6 3.5
Flame temp. °C 2160 2200 2155 2145 1899 2210
Blast humidity g/m³ STP 40 12 12 12 12 12
Hot blast temp. °C 1000 1200 1200 1200 1200 1200
Table 5.3 Typical operating conditions for different types of injectants

70 Chapter 5
5.2 Coal injection
5.2.1 Coal injection: equipment

The basic design for coal injection installations requires the following functions
to be carried out (Figure 5.2):
– Grinding of the coal. Coal has to be ground to very small sizes. Most
commonly used is pulverized coal: around 60 % of the coal is under 75 μm
(like powder). Granular coal is somewhat coarser with sizes up to 1 to 2 mm
(like beach sand).
– Drying of the coal. Coal contains substantial amounts of moisture, from 8 %
to more than 10 %. Since injection of moisture increases the reductant rate,
moisture should be removed as much as possible.
– Transportation of the coal through the pipelines. If the coal is too fine or
contains moisture, the pneumatic transport will be hampered. It may result in
formation of minor deposits on the transportation pipe walls and also lead to
coal leakage from the transportation pipes.
– Injection of the pulverized coal: Coal has to be injected in equal amounts
through all the tuyeres. Particularly at low coke rate and high productivity, the
circumferential symmetry of the injection should be maintained. Low injection
rates (50 kg/tHM) at all tuyeres should be facilitated by the injection system.
This is sometimes a problem with “dense–phase” coal injection systems.
Figure 5.2 Example of PCI installation

There are various suppliers available for pulverized coal injection (PCI)
installations. The reliability of the equipment is of the utmost importance, since
a blast furnace has to be stopped within one hour if the coal injection stops.
5.2.2 Coal specification for PCI
Coke replacement
Coal types are differentiated according to their volatile matter content. The
volatile matter is determined by weighing coal before and after heating for
three minutes at 900 °C. Coals that have between 6 and 12 % volatile matter
are classified as low volatile, mid volatile is 12–20 % in China and in different
areas 12–30 %, while high volatiles are over 20 % in China and over 30 %
elsewhere. All types of coal have been used successfully. The most important
property of the injection coal is the “Replacement Ratio” (RR) relative to coke.
The composition and remaining moisture content of the coals determine the
amount of coke replaced by 1 kg of that type of coal. The replacement ratio
of coal can be calculated with a mass and heat balance of the furnace. The
chemical composition of the coal (i.e. carbon percentage, hydrogen percentage,
oxygen content and ash content), the remaining moisture and the heat required
to decompose or crack the coal chemical structure (especially the C–H
bonds) have to be taken into account. Young coals contain relatively high O
percentage, which means that the heating value of the coal is lower, since the
CO bonds have already been formed in the coal structure. A simplified formula,
provided by a coal supplier, for the replacement ratio is in Formula 5.1:
RR = 0.998  C%/100 + 2.217  H%/100 – 0.077  O%/100

– 0.067  N%/100 – 0.073  (Qcrack  (100 – moist)/100) – 1.1  moist/100
where moist is the moisture content of the injected coal, and where RR is given
as the amount of coke, with a C–content of 87.5 %, replaced by 1 kg of coal.
The formula shows that coke replacement depends mainly on the carbon and
hydrogen content of the coal. Any remaining moisture in the coal consumes
energy introduced with the coal.
Cracking heat or the heat of decomposition is the difference between theoretical

and measured calorific value. Accurate measurement of the calorific value
is required to enable accurate determination. The following formula for the
cracking heat (in MJ/kg) was generated from a large dataset of coal samples
whose calorific value was measured using a high–quality bomb calorimeter.
Qcrack = 0.0007  VM2 + 0.0126  VM – 0.3687 (if Qcrack < 0 then Qcrack = 0)
Cracking heat is generally accepted to represent the heat of dissociation of C–H

bonds (MJ/kg) plus the heat of oxidation that is not realized due to the coal
carbon having already reacted with oxygen. The higher volatile matter coals
generally contain more oxygen.
72 Chapter 5
Coal quality
Besides the carbon and hydrogen content, the most important parameters for
coal quality to be injected are the following:
– Ash content and slag fusion temperature: the ash content of coal should be as
low as possible, since it consumes high temperature heat of the raceway for
heating and slag melting. Coal with high slag fusion temperature is helpful,
because it is not easy to form scabs at the lance tip or inside a tuyere.
– Volatile matter: high volatile coals are easily gasified in the raceway, but have
lower replacement ratio in the process.
– Composition: high sulfur and high phosphorus are likely to increase costs in
the steel plant. These elements should be evaluated prior to the purchase of a
certain type of coal. Young coals (high oxygen content) are known to be more
susceptible to self–heating and ignition in atmospheres containing oxygen.
This is also an important factor that must be considered with regard to the
limitations of the ground coal handling system.
– Hardness. The hardness of the coal, characterized by the Hardgrove
Grindability Index (HGI) must correspond to the specifications of the grinding
equipment. The resulting size of the ground coal is also strongly dependent
on this parameter and must correspond to the limits of the coal handling
and injection system. The HGI is a measure of the fines < 74 μm (200 mesh)
generated during a standard grinding test, with softer coals having higher HGI.
– Moisture content. The moisture content of the raw coal as well as the surface
moisture in the ground coal must be considered. Surface moisture in the
ground coal will lead to sticking and handling problems.
– Chloride content: The chloride content of coal can cause corrosion of the steel
in the blast furnace gas cleaning system. Many blast furnaces use a coating
inside the gas scrubber with a corrosion resistant paint or epoxy coating.
– Alkali (K 2O, Na 2O) content: The alkali in the coal for injection should be
considered in the total alkali loading into the blast furnace, which is known to
attack the refractory lining.
Potential injection coals can be evaluated on the basis of “value in use”, where
all effects on cost are taken into account. It is often possible to use blends of two
or three types of injection coals, so that unfavorable properties can be diluted
(Section 5.2.4).
Coal types
There are many different types of coal. Among others, coals are characterized
by the ash content and “volatile matter” content: the longer coals are left at high
temperatures in the earth’s crust, the more volatiles escape.
The special aspect in a blast furnace is that the most valuable heat is generated
by the “incomplete combustion” of coal to carbon monoxide (and hydrogen),
while in a power plant the coal is combusted completely to carbon dioxide and
water (H2O). The carbon monoxide and hydrogen are used to reduce the oxygen
content of the iron ore and are called “reductants”. Since in the lower part of
the furnace we need heat (enthalpy) at a high temperature, the most appropriate
coals are the coals that have lower ash content, as well as lower oxygen content
in the structure. This is because the fewer oxygen bonds with carbon that are
present, the more heat is generated when forming carbon monoxide.
For heat generation, the better coals are low volatile, low ash coals. Table 5.4
compares HV (high volatile) and MV (mid volatile) coals for the reference
furnace. MV have a higher replacement ratio. Productivity (at constant O2
enrichment and ΔP) is ~2.5 % lower for high volatile coals, which can be
compensated by increasing oxygen enrichment of the hot blast. The maximum
production rate determined by maximum ΔP is equal for high and low volatile
coals, although higher volatile coals require more oxygen enrichment.
High volatile coals (above 25 % VM) are easier to gasify in the raceway. For this
reason, for many years, high volatile coals were the preferred coals, especially
in Europe and the USA. Although experience in China shows that LV and MV
coals can be used very efficiently.
50/50 blend HV MV
Volatile matter % 24 31 17
C % 80.0 78.0 82.0
H % 4.4 4.5 4.3
O % 4.9 8.0 1.7
N % 1.4 1.3 1.4
S % 0.7 0.7 0.6
Ash % 8.8 7.5 10.0
Qcrack GJ/t 0.4 0.7 0.0
Moisture % 1.0 1.0 1.0
Lower heating value GJ/t 31.1 30.3 31.9
RR 0.86 0.82 0.91
Tflame (+10 kg coal/tHM) °C –26.6 –27.1 –26.1
Process Results Reference Furnace
Production tHM/d 12000 11845 12140
Coke Rate kg/tHM 299.7 299.7 299.7
Coal rate kg/tHM 200.0 210.6 190.4
O2 in blast % 27.0 27.0 27.0
Top temperature °C 150 159 143
Tflame °C 2204 2165 2240
ΔP bar 1.775 1.775 1.775
Max production at ΔP = 1.8 and coke rate = 300 kg/tHM
Production tHM/d 12865 12870 12860
O2 in blast % 30.05 30.8 29.4
Tflame °C 2303 2282 2323
Table 5.4 High and mid volatile coals compared at constant coke rate and ΔP
74 Chapter 5
Coal with a volatile matter content of 15–21 % can be considered. These coals
can be comfortably injected into a blast furnace. If the VM content is lower
than this, the coal is likely to be less reactive and so the rate of conversion may
be insufficient. Coals with a VM below 6 % should not be used for this reason.
Anthracites are very low VM coals, which fall into this category. They are very
unreactive due to their more ordered carbon structure. Trials in the 1990s using
a high proportion of anthracite experienced a significantly higher degree of coal
carryover into the dust. It should be noted that the real volatile content of coals
is 1.2 to 2 times the reported because the real heating rate and final temperature
are much higher in the raceway than in the proximate analysis test, which was
developed for coking and thermal coal evaluation.
Coal blending
Most companies use coal blends for injection. Blending allows for (financial)
optimization of coal purchases. E.g. a company with a grinding mill for hard
coals can use a considerable percentage of softer coals by blending them into
hard coals. In doing so, an optimized value can be obtained. Blending dilutes
the disadvantages of some coal types. Every material has disadvantages, like
high moisture content, sulfur or phosphorus level, a relatively poor replacement
ratio and so on. The blending can be done rather crudely. Depositing materials
in the raw coal bin by alternating truck loads can be sufficient. Proper control
of coal blend have to be put in place. In some blast furnaces, further cost
reductions have been realized by blending lignite (brown coal), charcoal,
petroleum coke or anthracite with the normal coal blends.
Research in coal gasification rigs has shown that when a high volatile coal is
mixed with a low volatile coal, they combust more effectively than individually.
A suggested mechanism is that the volatiles from the high VM coal are released
first, combust around the low VM coal particles, and the thermal shock then
helps the low VM coal to release its volatiles and combust. Different coals
combust at different temperatures, times and rates. For this to be effective the
coals must be well blended.
80
75
Combustion Rate (%)
70
65
60
55
50
Coal A Coal B Coal C 50 % A 50 % B 20 % A
(8.7 % VM) (17.1 % VM) (36.7 % VM) 50 % B 50 % C 50 % B
30 % C
Figure 5.3 Effect of coal blending on gasification in a test rig

(Santos Assis et al, 2004)
Figure 5.3 shows the rate of combustion of individual coals and blended coals
(Santos Assis et al, 2004). These results were measured on a pilot scale test rig
simulating a single tuyere. Others have noted that taking this to extremes can
be detrimental because the high volatile coal can consume most of the oxygen
meaning that burnout of the remaining low volatile coals will be low, which
reduces their char reactivity.
5.2.3 Coal injection in the tuyeres

Coals are injected via lances into the tuyeres, then gasified and ignited in the
raceway. The coal is in the raceway area only for a very short time (20–50
milliseconds – Steer et al, 2018), so the characteristics of the gasification
reaction are very important for the effectiveness of a PCI system. Coal
gasification consists of several steps as outlined in Figure 5.4. First, the coal is
heated and the moisture evaporates. Gasification of the volatile components
then occurs after further heating. The volatile components are gasified and
ignited; this causes an increase in the temperature. All of these steps occur
sequentially with some overlap.
Ignition/Oxidation of Char
Ignition/Oxidation of Volatiles
Temperature
Gasification of Volatiles
Heating of Coal
Evaporation of Moisture
t=0 t = ~ 10 ms
Time (ms)
Figure 5.4 Coal gasification after injection
The effects of lance design, extra oxygen and coal type on the coal combustion
have been analyzed. Originally, the coal lances were straight, stainless steel
lances that were positioned at or close to the tuyere/blowpipe interface as
indicated in Figure 5.5. Occasionally, very fine carbon (soot), formed from
volatile cracking, is detected as it leaves the furnace through the top. To avoid
this problem, especially at high injection rates, companies have installed
different types of injection systems at the tuyeres, such as:
– Use of two lances per tuyere at different entry position or angles to cause more
turbulence or better mixing with the blast. This approach is becoming more
common with those operators aiming for very high PCI rates (> 220 kg/tHM).
– Co–axial lances with oxygen flow in an outer concentric pipe and coal flow in
an inner concentric pipe.
– Lances with a tip specifically designed to cause more turbulence at the lance tip.
– Bent lance tips, positioned more inwards in the tuyere, closer to the tuyere nose.
76 Chapter 5
The lance tip must be positioned at the center of the tuyere. It the coal impacts
on the tuyere inside it can be damaged by abrasion. The lance tip must be in
front of the blowpipe/tuyere interface to ensure that the coal does not burn in
the blowpipe.
When using PCI, deposits of coal ash are occasionally found at the lance tip or
within the tuyere. The deposits can be removed by periodic purging of the lance
by switching off the coal while maintaining air (or nitrogen) flow.
The rate of gasification increases as:

– The volatility of the coals increases.
– The size of the coal particles decreases.
– The blast and coal are mixed better.
The gasification of coal also depends on the percentage of volatile matter

(VM). If low volatile coals are used, a relatively high percentage of the coal is
not gasified in the raceway and is transported with the gas to the active coke
zone. This “char” will normally be used in the process, but might affect the gas
distribution. The high volatile (HV, over 30 % VM) and ultra high volatile
coal (over 40 % VM) produces a large quantity of gas in the raceway and a
small quantity of char. So, higher VM coals cause a greater increase in blast
momentum meaning that a stronger coke may be required.
If the gas combustion is not complete, soot can be formed. Blending a variety of
coals, especially high and low volatile coals, yields the advantage of being able
to control these effects. It has been found that the coke at the border between
raceway and deadman contains more fines when working at (high) injection
rates. This packed coke region has been termed the “bird’s nest” (Figure 5.5).
Figure 5.5 Coal injection in the tuyeres, single lance and double lances
5.2.4 Process control with coal injection: Oxygen and PCI

At high PCI operation, about 40 % of the reductant is injected via the tuyeres.
For this reason, it is important to control the amount of coal per tonne of hot
metal as accurately as the coke rate is controlled. The feed tanks of the coal
injection system are weighed continuously, and the flow rate of the coal is
controlled. It can be done with nitrogen pressure in the feed tanks, or a screw or
rotating valve dosing system. In order to calculate a proper flow rate of coal (in
kg/minute) the hot metal production rate must be known. There are several
ways to calculate this. The production rate can be derived from the amount
of material charged into the furnace. Short–term corrections can be made by
calculating the oxygen consumption per tonne of hot metal from the blast
parameters in a stable period and then calculating the actual production from
blast data. Systematic errors and/or the requirement for extra coal can be
included in a process control model.
The heat requirement of the lower furnace is a special topic when using PCI.
Coal is not only used for producing the reduction gases, but use of coal has
an effect on the heat balance in the lower furnace. The heat of the bosh gas
has to be sufficient to melt the burden. The heat requirement of the burden is
determined by the “pre–reduction degree”, or how much oxygen still has to be
removed from the burden when melting. The removal of this oxygen requires a
lot of energy. The “melting capacity” of the gas is defined as the heat available
with the bosh gas at a temperature over 1500 °C and depends on:
– The quantity of tuyere gas available per tonne of hot metal. Especially when
using high volatile coal there is a high amount of H2 in the bosh gas.
– The flame temperature in the raceway.
The flame temperature itself is determined by coal rate, coal type, blast
temperature, blast moisture and oxygen enrichment. From the above, the
oxygen percentage in the blast can be used to balance the heat requirements of
the upper and lower furnace. The balance is dependent on the local situation. It
depends on burden and coke quality and coal type used. For the balance there
are some technical and technological limitations, which are presented as an
example in Figure 5.6.
Figure 5.6 Working area (white) showing limiting factors with coal injection as
calculated for a ΔP of 1.8 bar
78 Chapter 5
For higher injection rates more oxygen is required, but this has to remain
within the limits of the working area, which is the area between a minimum
top temperature and a minimum Tflame.
– Minimum top gas temperature: If the top gas temperature falls too low, it takes
too long for the burden to dry and the effective height of the blast furnace
shortens. Deposits of wet dust particles in the first stage of gas cleaning (dust
catcher or cyclone) can result in plugging of the dump valves.
– Minimum flame temperature: Low flame temperature will hamper coal
gasification and melting of the ore burden. In Figure 5.5 (next page), 2050 °C is
used as the minimum flame temperature.
Further restraints have to do with maximum flame temperature and technical
limitations. If the flame temperature rises too high, burden descent can become
erratic. Technical limitations are for instance the allowed or available oxygen
enrichment in the plant’s oxygen production facility or pipeline size or pressure.
The higher the oxygen injection, the higher the productivity of the furnace can
be, as shown in Figure 5.6, which is based on the mass and heat balance of the
reference furnace. The highest productivity is reached, with an oxygen level set
so that the top gas temperature is at the minimum. The minimum is the level
where all water of coke, burden and process is eliminated from the furnace, i.e.
slightly above 100 °C. From a technological perspective it can be said, that the
heat balances over the lower part of the furnace (i.e. from 900 °C to tuyere level)
and over the upper part of the furnace (i.e. from top to the 900 °C isotherm)
are in balance (Section 8.5). In operational practice the availability of oxygen
in a plant is often the limiting factor. Sometimes “unburnt coal” is observed in
top gas. In Example 5.1, coal gasification is discussed. Besides incomplete coal
gasification, char can be observed in top gas, leading to black foam on the water
treatment plant for the following mechanisms:
– Unburnt material can be found in periods with poor burden descent. Our
explanation is that in these situations channeling of the gas flow from tuyere to
top takes place.
– High coal injection requires a central coke chimney and results in high top gas
temperatures in the center of the furnace. A possible explanation for some of
the “char” found at the top is Boudouard carbon (solution loss reaction), formed
when the central high temperature (over 800 °C) of the top gas with a high
CO percentage cools down over the temperature range where the Boudouard
reaction takes place.
5.2.5 Effect of additional PCI

The effect of the use of extra coal injection for the recovery of a cooling furnace
is three–fold. By putting extra coal on the furnace:
1. the production rate decreases, because the production rate is determined by the
amount of coke gasified by the hot blast;
2. the flame temperature drops;
3. the gas reduction increases by increased H2 content in bosh gas, and thus direct
reduction decreases.
Responses 1 and 2 are fast and are only delayed by the response time of the
hearth heating up – so two to three hours. Response 3 is slower, because the
cohesive zone has to be processed before it becomes effective, and response
will take an additional 2–3 hours. A furnace recovers from a cold condition
by increasing PCI, because it slows down the production rate, because the
coke burning rate decreases. If simultaneously Tflame is kept constant, there is
even more heat per tonne available to melt the ferrous burden. If the effect of
hydrogen is taken into account, which reduces direct reduction and therefore
the heat required for melting, the effect becomes even more pronounced. For
daily operations: remedial action for a cooling trend is:
– to use additional PCI
– at constant Tflame and
– constant (or decreasing) hot blast oxygen input per hour.
Example 5.1 Gasification of injected coal

Coal is not completely gasified in the raceway. Some material leaves the
raceway as char. Observation of “un–burned char” at the top has led to
numerous investigations with respect to coal gasification. At high coal
rates, an estimated 30 % of injected coal leaves the raceway as char. In most
situations only very small amounts of char leave the furnace with top gas.
So, most char is used in the process, most likely by its use in direct reduction
reactions or by adhering to coke. Char is much more reactive than coke; so
will preferentially be consumed at high temperature by Boudouard reaction
with CO2 to form CO. There is no standard test for char reactivity. One test
reported (Steer et al, 2018) uses 100 % CO2 at 900 °C and measures the time
taken for 50 % of the reactive portion of the char to be consumed. Chars
were produced in a drop tube furnace (DTF) and tested by this method. The
results show that char from LV coal is less reactive than char from HV coal.
In general, coals with a higher burnout (proportion of carbon burnt in the
drop tube furnace), generate a more reactive char, and HV coals yield better
burnout. For an individual coal, the longer the residence time in the DTF,
the higher the burnout, but the poorer the char reactivity.
There is debate over the factors affecting char reactivity. Char morphology
influences the reactivity due to its effect on available surface area for reaction.
The internal surface area available for reaction may be increased by the
presence of large ‘feeder pores’ at the char surface (Hippo & Walker, 1975).
The char pore structure is not determined by the original coal porosity (Yu
et al, 2007). Swelling, fragmentation and agglomeration all impact available
surface area. It is also reported that ‘reactive sites’ within the pores affect the
intrinsic reactivity (Bar–Ziv and Kantorovich, 2007; Gupta et al, 2006).
Higher ash coals, particularly those predominantly formed from

aluminosilicates, can inhibit burnout in the raceway by the molten ash
encapsulating organic material, thereby inhibiting its availability for reaction.
80 Chapter 5
5.2.6 Effect of a tuyere without PCI injection

At many operating furnaces one or more tuyeres are not injecting PCI. What
occurs in the blast furnace?
Suppose that all tuyeres accept the same hot blast volume per hour. In one
tuyere no coal is injected, but the same amount of hot blast as in its neighboring
tuyeres. At this tuyere, only coke will burn, leading to an increase in coke
burning rate (= production rate) and a strong increase in Tflame. The coal will
be divided over the other tuyeres, which decreases the local coke burning rate
(= production rate) and Tflame. This is shown in Figure 5.8 and results are given
Table 5.5. The cohesive zone becomes asymmetric. Missed coal injection coal
into a tuyere becomes more serious at higher injection rate. If more than one
tuyere is not on coal injection, the situation is more severe, when they are
located close to each other.
Figure 5.8 Tuyere without PCI and its effect on cohesive zone
base 39 tuyeres 1 tuyere

on coal without coal
Production tHM/d 12000
Tuyeres 40 39 1
Production tHM/h.tuyere 12.5 12.1 29.7
Tflame °C 2200 2187 2732
Production rate tHM/h.tuyere 12.5 12.1 29.7
Production rate % of basis 100 96 238
Table 5.5 Process effects when there is no coal injection on a tuyere

Example 5.2 Coal backing up

With coal injection, it is very important that the tuyeres are clear and open,
allowing the coal plume to flow into the raceway to optimize combustion.
If the tuyere should become blocked, or a blockage in front of the tuyere
appears, injection must be stopped immediately. If it is not, the coal will flow
backwards into the tuyere stock and can ignite further up in the connection
with the bustle pipe (Figure 5.9). This can cause serious damage or even
explosions. It is especially prone to happen within an hour after restart
of a furnace, since unmolten material can block the tuyere gas flow. The
phenomenon has also been observed with natural gas injection. In many cases
this has been a cause of a chilled hearth since the furnace is shut down in an
unprepared state for extended periods to replace tuyere stock or repair the
bustle main.
Figure 5.9 Coal (or natural gas) backing up into the bustle pipe, caused by scab
in front of tuyere, leading to the possibility of explosion. The upper
picture shows the damage at the hot blast main.
To prevent an incident, a light sensor may be fitted in front of the peepsight

to detect a blockage at the end of the tuyere, or the ΔP can be measured over
the tuyere stock to detect gas flow into the furnace. If there is no flow, this
indicates that a blockage is present. The coal to that tuyere is automatically
switched off and restarted only once an operator has checked the tuyere.
Many furnaces using natural gas injection rates over 90 kg/tHM are also
installing the ΔP blocked tuyere detection to shut off the natural gas to
prevent back–ups. Tuyere cameras allow the operator to monitor blockage
from the control room.
82 Chapter 5
5.3 Natural gas injection

Natural gas typically consists of methane with some minor quantities of other
hydrocarbons and nitrogen. Most frequently, natural gas injection is applied in
the range of 60–90 kg/tHM, although higher levels have been reached up to
150 kg per tonne of hot metal (Agarwall et al, 1999) or on other furnaces more
than 100 kg natural gas per tonne of hot metal (see e.g. Lingiardi et al, 2001).
The replacement ratio of natural gas is typically 1–1.1 kg coke/kg natural gas
and the replacement ratio is non–linear, so decreases as the natural gas rate
increases This happens especially above 80 kg/tHM because the water content
of top gas increases, and the gas volume decreases because of the high level of
oxygen enrichment required.
5.3.1 Injection systems for natural gas

Natural gas injection is mostly into the tuyeres with one lance per tuyere, but
it can also be injected through a port in the tuyere copper casting sidewall.
Natural gas is injected into the tuyeres at ambient temperatures and is gasified
in the tuyeres. The residence time of the gas in an oxidizing atmosphere of the
raceway is very short: 4–7 milliseconds. Some companies apply premixing of
gas and oxygen and injecting it with gas–oxygen mix lances aimed at complete
combustion in the raceway.
In order to keep the natural gas rate constant on the basis of kg natural gas per
tonne of hot metal, the production level of the furnace has to be constantly
monitored and calculated, so that the total flow control can be adjusted
in real time. The control system of natural gas normally is based on a flow
measurement and control valve for injection in every tuyere.
In contrast to PCI, an injection system for natural gas is relatively simple and
requires limited investment.
5.3.2 Process control with natural gas

As in Figure 5.6 for coal injection, the working area for natural gas and oxygen
enrichment is shown in Figure 5.10. When using natural gas, the minimum
Tflame (for example 2050 °C) and minimum top gas temperature cross each
other; this will be the point with the highest possible productivity and lowest
coke rate. The exact position of the crossing point depends on local conditions
(like moisture input with burden and coke) and on the minimum possible
flame temperature. Fortunately, because of the hydrogen effect, the minimum
Tflame with gas injection is lower than with coal injection; this is shown with the
flame temperature of 2000 °C in Figure 5.10, resulting in a potential higher gas
injection rate of about 20 kg/tHM.
Figure 5.10 Working area (white) for a furnace on natural gas injection
5.3.3 Natural gas and thermal control

The usability of coal–injection for thermal control is based on the fact that, as
soon as the coal injection rate is increased, the production slows down, which
compensates the effect of the lower flame temperature. Natural gas works much
slower in the furnace: the immediate effect of increase in natural gas injection
is that the flame temperature drops three times more than with coal, while
the production rate decreases slower than with PCI. Natural gas works more
indirectly by changing the reduction reactions. Additional coal injection works
immediately and is visible in the casthouse after 3–6 hours. For additional gas
injection the effect of, for example 10 kg, natural gas is smaller, and the reaction
is further delayed by 2–3 hours.
5.3.4 Flame temperature, hydrogen and total energy approach

The adiabatic flame temperature reflects the heat which is used to melt and
heat up the hot metal and slag, as well as the heat used in the final reduction
of oxides. As shown above, the flame temperature effect of natural gas is much
larger than that of coal, oil or tar. However, the decrease in direct reduction
reaction means that less heat is required for the final melting and heating.
Furnaces operating with natural gas can operate on lower flame temperatures.
Furnaces operating with coal injection typically operate in a range of flame
temperatures of 2050–2200 °C, while furnaces with natural gas can operate at
flame temperatures well below 1900 °C. The effect is caused by the relatively
large percentage of hydrogen in natural gas. It is discussed here, but equally well
applicable for hydrogen coming from other sources (coal, oil and moisture in
hot blast).
The hydrogen is generated by the incomplete combustion of natural gas and
partly replaces the carbon monoxide. Hydrogen is very efficient for reduction
in the lower part of the furnace. The presence of hydrogen in the bosh gas
lowers the direct reduction, and gas reduction in the shaft increases. So, higher
hydrogen in the bosh gas lowers the heat requirement of the lower part of
the furnace. This approach has been quantified by Charles River associates
(Agarwall et al, 1991). The basis of their approach is that the hydrogen effect is
recalculated to a lower heat requirement for the lower furnace.
The total energy is the thermal energy plus the “chemical” energy. The thermal
energy is the amount of heat generated in the raceway above a reference
temperature of 1482 °C (= 2700 °F). So, it is the gas quantity (m³ STP/
tHM) times (Tflame –1482) times the heat capacity of the gas (kJ/m³ STP).
The chemical energy is the equivalent energy of the bosh gas hydrogen and
the hydrogen utilization, taking into account that additional gas reduction by
hydrogen gives an energy advantage of 132 MJ/kmole FeO. This is based on the
following reaction equations:
FeO + C Fe + CO ΔH = 159 MJ/kmol FeO

FeO + H2 Fe + H2O ΔH = 27 MJ/kmol FeO
The difference is 132 MJ/kmol FeO, the total energy is the sum of thermal and
chemical energy. Analysis of an operating blast furnace at varying gas rates has
shown that the total energy required in the lower furnace is constant (Lingiardi
et al, 2001). This means, that at higher gas injection rates and thus higher
hydrogen input per tonne, the chemical energy increases allowing the thermal
energy (Tflame) to decrease.
But hydrogen has other effects on the process. Firstly, since hydrogen is a very
light molecule, the density of the blast furnace gas decreases with increasing
hydrogen content and more wind can be blown for the same ΔP. This can be
quantified as follows. The Ergun equation is
ΔP/L= Kρv2
where ΔP/L is the pressure difference per unit height – and is related to burden
properties, permeability index (K), the gas density (ρ) and the gas speed (v). For
5 % more hydrogen in the bosh gas, the density decreases by 4.3 % and the
blast volume can increase by 2.3 % for the same ΔP.
Secondly, since part of the reduction process is done with hydrogen, the
requirement for carbon monoxide decreases. The amount of carbon – the
carbon footprint – required per tonne of hot metal decreases. The difference is
that the carbon requirement for a furnace on natural gas is 10–15 kg lower than
for a furnace with coal injection, when we neglect any coke effects.
Third, standard deviation of hot metal silicon and sulfur decrease with
increasing injection levels. This is ascribed to the observation that the variability
in direct reduction is largely responsible for the variability of the thermal state
of the blast furnace.
It should be noted that even though the additional hydrogen produced by

high natural gas injection (above 100 kg/tHM) improves the process, as
described above, the top gas hydrogen can be between 10–14 %, which is highly
flammable. For this reason, the furnace top and charging equipment must
be maintained free from leakage, or a top fire could result when mixed with
oxygen in the outside air.
5.4 Coal–gas co–injection

Depending on the local situation, operators choose to co–inject coal and gas
simultaneously, especially in North America. Co–injection can be done by
using two lances per tuyere as shown in Figure 5.11.
Figure 5.11 Through the peepsight: co–injection of coal and natural gas
Figure 5.12 shows the annual average operating results of North American
blast furnaces over 2016, with 14 furnaces co–injecting coal and natural gas.
Typically, tonne of hot metal about 20–40 kg natural gas and 125–160 kg coal
are used per tonne of hot metal.
86 Chapter 5
Since in the raceway, the gas and coal are (nearly) completely converted to CO
and hydrogen, the effects of both injectants are additive. Coal still generates
the most heat, while gas generates hydrogen, which is efficient in the lower
part of the furnace. In comparing the operating furnaces, it was found that
in some furnaces the amount of coal replaced by natural gas was significantly
higher than that calculated from models. This is probably due to the fact that
direct reduction is lower when using natural gas. As a consequence, the furnace
becomes thermally more stable, which allows the operator to work on a lower
thermal level (Geerdes, 2016).
200
Coal Injection (kg/tHM)
150
300 kg/tHM
100 350 kg/tHM
50
0
0 20 40 60 80 100
Natural Gas Injection (kg/tHM)
Figure 5.12 Co–injection of natural gas and coal in North American blast furnaces.
Dotted lines indicate coke rates.
VI Blast Furnace Gas and
Processing the Ferrous Burden
6.1 Introduction
The blast furnace is charged with pellets, sinter, lump ore and coke, while
additional reductant (pulverized coal, natural gas) can be injected through
the tuyeres. The steel plant requires a defined quality of hot metal and the slag
chemistry has to be designed for optimum properties with respect to liquidus
temperature, fluidity, desulfurising capacity, alkali removal and slag markets.
For this reason, the blast furnace operator has to make calculations to select
the blast furnace burden. The present chapter first details the conditions for a
burden calculation, which is then illustrated with a practical example. Then
it is shown how and where the gas processes the ferrous burden. In order to
aim for maximum productivity, the permeability for gas and efficiency of the
process have to be understood. Finally, a furnace is efficient if every tuyere
contributes the same amount to the process. So, circumferential symmetry
is of major importance, while the blast furnace process is very asymmetric in
the radial direction. An important lesson from this chapter is that in a blast
furnace the local temperature determines the progress of the chemical reactions.
Understanding blast furnace operation is about understanding heat transfer and
to a much lesser extent understanding chemistry.
6.2 Hot metal and slag composition
6.2.1 Hot metal and slag quality objectives

Starting points for burden calculations are the hot metal and slag quality.
– Hot metal quality: hot metal contains around 94.5 % iron (Fe), some 4.5 %
carbon (C), 0.4–0.6 % silicon (Si). Low sulfur (S, under 0.03 %) and defined
phosphorus (P) levels, typically ranging from 0.05 to 0.13 %, which depends
on the phosphorus content of the burden materials. The balance consists of
manganese (Mn) and a number of different elements including titanium.
– Slag quality: generally, the lower the slag volume the better. Typically, the four
major constituents of slag represent some 96 % of its volume: Al 2O3 (8–20 %),
MgO (6–12 %), SiO2 (28–38 %) and CaO (34–42 %). For slag design, see
Chapter VIII.
88 Chapter VI
Example 6.1 Burden Calculation

The composition of burden components and their distribution over hot
metal and slag can be calculated with a one–stage mass balance. The
present example is restricted to the components required to calculate the
slag composition. The four main components (SiO2, CaO, MgO and Al2O3)
represent 96 % of the total slag volume. The remaining 4 % consists of MnO,
S, K 2O, Na 2O, TiO2 and many more. Yield losses when charging the furnace,
have to be taken into account for operation, but are left out here. Yield losses
collect in the dust catcher and gas scrubber systems.
The calculation is done as follows. The targets for the calculations are taken;
in most models the basicity, either the 2 or 3 components basicity are chosen.
In this example, a B2 basicity (CaO/SiO2) of 1.15 is used. The chemical
analysis of the input materials is shown in the upper part of the table. The
quantities of the various materials are used to make 1 tonne of hot metal,
which contains 945 kg of Fe. In the example 70 % sinter, 20 % acid pellets
and 10 % lump ore are used. The final burden is calculated in an iterative
process, where more elements are considered and more restrictions can apply:
for instance with respect to the slag Al2O3 or MgO content as well as with
respect to the input of different materials like alkali and zinc.
Chemical Analysis (weight %) ash Fe SiO2 Al2O3 CaO MgO

coke 11.0 0.5 5.5 3.3
coal blend 8.8 0.5 4.4 2.6
sinter 58.3 4.6 1.0 8.9 1.8
pellets 65.0 4.0 1.4 0.5 0.3
lump 62.0 5.0 1.8
Slag Analysis (per tHM) kg/tHM Fe SiO2 Al2O3 CaO MgO
coke 300 2 17 10 0 0
coal blend 200 1 9 5 0 0
sinter 1099 641 51 11 98 20
pellets 314 204 13 4 2 1
lump 157 97 8 3 0 0
Total 1570 945 96 33 99 21
Hot metal Silicon (0.46%) –10
In slag kg/tHM 86 33 99 21
Results SiO2 Al2O3 CaO MgO
slag composition % 34.5 13.3 39.8 8.3
slag rate* kg/tHM 250
B2, CaO/SiO2 1.15
B3, (CaO+MgO)/SiO2 1.39
Table 6.1 Calculation of slag composition (starting points in Annex II)

Blast Furnace Gas and Processing the Ferrous Burden 89
The burden has to fulfil requirements with respect to the highest allowable
input of:
– Phosphorus – since phosphorus leaves the furnace with the iron.
– Alkali input – sodium (Na 2O) and especially potassium (K 2O), which can
attack the refractory and affect the process. Typically, a limit of 1 to 3 kg/tHM
is used.
– Zinc: zinc can condense in the furnace and can, similar to alkali, lead to a zinc
cycle. Limits for zinc input are typically 100–150 g/tHM. With high central gas
temperatures, zinc and alkali are partly removed with the top gas.
6.2.2 Burden calculation for slag composition

The burden calculation uses the chemical composition (on a dry basis) and
the weights of the various materials in a charge as input parameters, usually
expressed per tonne of hot metal. A charge consists of a layer of burden material
and a layer of coke with its auxiliary reductants as injected through the
tuyeres. The mass balance consists of the input and output of the iron and slag
components (SiO2, CaO, Al2O3, MgO), as described in Example 6.1.
6.3 Top gas composition from a one–stage

mass balance
The blast furnace process is driven by blowing hot blast into the furnace that
gasifies coke and injectants. So, shortly after the raceway, the gas temperature
(Tflame) is high (2200 °C); bosh gas consists of nitrogen (N2), carbon monoxide
(CO) and hydrogen (H2). The gas heats and reduces the ferrous burden. The
overall balance of the gas is presented in Example 6.2 on the next page. Details
of what happens inside the blast furnace are discussed in Section 6.6.
The top gas in the blast furnace is formed as follows.

– Nitrogen (N2): the major source of nitrogen is the hot blast. Nitrogen is inert in
the furnace and escapes from the charged burden 6–10 seconds after it has been
blown in.
– Carbon (C), contained in coal and coke, that is used in the process for chemical
reactions and for dissolution of carbon (around 4.5 %) in the hot metal. The
balance of the carbon leaves the furnace with the top gas.
– Oxygen (O2): oxygen is blown into the furnace with the hot blast and is charged
with ferrous burden. Some minor quantities of oxygen are contained in the coal
and in moisture in wind and injected coal. Oxygen leaves the furnace with the
top gas, again as CO and CO2 and as water (H2O) formed in the process of
reduction of the burden with H2.
– Hydrogen (H2): the major source of hydrogen is the fuel injection, with minor
amounts coming from moisture in the hot blast and coal. The water charged
with the burden and coke is not considered, because it evaporates before the
blast furnace process starts.
90 Chapter VI
Example 6.2 One–stage gas balance of a blast furnace

The balance can be used to estimate the bosh gas composition as well as the
top gas utilization. It is made up as follows based on the input data on the
next page:
– Count all inputs of oxygen, carbon, hydrogen and nitrogen together in kmol
per tonne of hot metal.
– Subtract the amount of carbon in the hot metal. The balance is eliminated
with the top gas.
– Part of the hydrogen is used for reduction of the ferrous burden generating
H2O. Typically, hydrogen utilization is 40 %.
– The result is an amount of nitrogen and hydrogen in the top gas (in kmol), as
well as an amount of kmol C and O. Since every CO contains 1 atom O per
C and CO2 contains 2 atoms O per C, the amount of CO2 is kmol O minus
kmol C, as illustrated in the table below.
– With CO and CO2 available, the gas utilization (CO2/(CO+CO2) can be
calculated.
Input tuyeres (per tHM) kmol N2 kmol H2 kmol C kat O kmol

Fe
wind m³ STP/tHM 900 29.9 21.7
moisture in wind g/m³ STP 12 0.6 0.6
coal kg/tHM 200 0.1 4.4 13.3 0.6
coal moisture kg/tHM 2 0.1 0.1
Input tuyeres per ton HM 29.4 5.1 13.3 23.0
Input top
coke kg/tHM 300 0.1 0.3 21.5
burden kg/tHM 1570 24.6 16.9
Total 29.6 5.4 34.8 47.7 16.9
Output taphole
hot metal kg/tHM 1000 –3.8 –16.9
slag kg/tHM 250
Output top gas total kmol N2 kmol H2 kmol C kat O
H2 eliminated as H2O –2.2 2.2
top gas (minus H2O) 29.6 3.2 31.1 45.5
top gas (composition) N2 H2 CO CO2 H2O
kmol/tHM 29.6 3.2 16.7 14.4 2.2
% 46.3 5.1 26.1 22.6
m³ STP/tHM 1431 662 73 373 323 48
kg/tHM 1935 828 6 467 634 39
ηCO 46.4 %
Composition Fe C H O N moist ash

coke kg/tHM 300 86.0 0.2 1.3 11
coal kg/tHM 200 80.0 4.4 4.9 1.4 1 8.75
hot metal kg/tHM 1000 94.5 4.5
slag kg/tHM 250
Operating conditions
wind volume m³ STP/tHM 900
O2 % in wind 27
blast moisture g/m³ STP 12
coal moisture % 1
utilization H2 % 40
O/Fe burden atom/atom 1.46
1 kmol gas m³ STP 22.4
Table 6.2 Gas balance in the blast furnace
6.4 The temperature profile and chemical reactions
The temperature profile and the chemical reactions in a blast furnace are closely
related. They are summarized in Figure 6.1 on the next page. The ferrous
burden is charged mainly as hematite (Fe2O3) with some magnetite (Fe3O4).
The reduction of the oxides to wustite (FeO1.05, we use FeO) takes place at
temperatures between 600 and 900 °C. Thereafter, in the temperature range
of 900 to 1100 °C, the wustite can be further reduced by gas. The chemical
preparation zone can take up to 50 to 60 % of the height of the furnace and has
a relatively constant temperature. This region is called the thermal reserve zone.
Figure 6.1 Reduction of iron oxides and the locations where the reactions have taken
place in a quenched blast furnace from Omori et al, 1985
92 Chapter VI
At temperatures higher than 1100 °C, carbon dioxide can react with coke
carbon to produce carbon monoxide The total reaction is known as direct
reduction, because carbon is directly consumed. The reactions can be indicated
as below:
2 FeO0.5 + CO 2 Fe + CO2
+ CO2 + C 2 CO
Total 2 FeO0.5 + C 2 Fe + CO ΔH = +155 kJ/kmol FeO
The direct reduction reaction requires an enormous amount of heat, which is

provided by the heat contained in the hot raceway gas. The direct reduction
reaction is very important for understanding the process. In a modern blast
furnace, direct reduction removes about a third of the oxygen from the burden,
leaving the remaining two thirds to be removed by the gas reduction reaction.
The amount of oxygen to be removed at high temperatures, as soon as the
burden starts to melt, is very much dependent on the efficiency of the reduction
processes in the shaft. For an explanation of the process, it is may be assumed
that the gas reduction reduces the iron oxides to FeO0.5. More efficient furnaces
do exist and reach O/Fe ratio of 0.4. Note the following important observations:
– Direct reduction uses carbon (coke) and generates extra CO gas.
– Direct reduction costs a lot of energy.
– A significant part of the direct reduction is believed to occur in the solid phase,
through reduction of CO2 with C before the burden starts to soften. Note that,
where direct reduction occurs, it does not affect the heat balance.
In operational practice the direct reduction can be monitored. In many blast

furnaces the direct reduction rate (the percentage of the oxygen removed from
the burden by direct reduction) or the solution loss (the amount of coke used for
the reaction) are calculated on–line. Experienced operators are well aware that
as soon as the direct reduction rate increases, the blast furnace burden starts to
descend faster, the cohesive zone comes down as the coke is consumed faster,
and the furnace chills. When properly monitored, chilling can be prevented, for
example by using additional coal injection.
6.5 Carbon balance: estimation of direct reduction
Considering the carbon balance in more detail shows that not all the coke
and coal is gasified at the tuyeres. Some carbon leaves the furnace with the
hot metal, around 45 kg. But even then, there is a large difference: the process
consumes an additional amount of carbon of approximately 100 kg/tHM.
This is the carbon used for direct reduction, mostly of iron oxide. The direct
reduction reaction 6.1 is:
FeO + C Fe + CO ΔH = +155 kJ/kmol FeO (6.1)
The direct reduction reaction of iron oxides costs heat and consumes coke.
About 80–90 % of the heat in the bosh gas above a temperature of 1400 °C is
needed for the direct reduction reactions. So, an increase in direct reduction
will start a cooling trend in the furnace as is shown by a decrease of hot metal
temperature and silicon. The cooling trend has a self–accelerating effect
because: the increase in direct reduction costs heat – this consumes coke – this
increases the production rate – causing an even larger heat shortage – resulting
in a further increase in direct reduction. For this reason, the furnace can cool
down from normal operation condition to very cold in 2–8 hours.
The variability of direct reduction is often the most important factor affecting
process stability. The trend in direct reduction can be monitored in real time
in the control room. It gives the operator an early warning signal for a cooling
trend in the production process, because the direct reduction reaction consumes
extra coke. In the event of a cooling trend, the operator has the burden descent
rate as a check. Increased direct reduction and increased burden descent rate are
a far faster indication than hot metal temperature or hot metal silicon measured
in the casthouse. A cooling trend of this kind can be stopped by additional
reductant injection. Some companies have an automatic thermal control based
on the heat balance of the lower furnace.
The direct reduction can be estimated on–line from the difference between the
carbon that leaves the top of the furnace and the carbon that is gasified at the
raceway, as in Example 6.3.
The operator has to react to severe cooling in good time. There are basically two
control methods:
– React with extra fuel (PCI) to a decrease of the thermal condition as indicated
by hot metal temperature and silicon.
– React with extra fuel (PCI) to a simultaneous increase of direct reduction and
burden descent speed. For direct reduction there are several calculation methods
available. For descent speed several indicators can be used: descent speed as
measured from stockrods, number of charges per hour, of the time needed for 5
charges.
The first method is delayed by 4–6 hours compared to the second method, since
casthouse data are collected less frequently and the hearth has to be cooled
down as well.
94 Chapter VI
Example 6.3 Direct reduction estimation and operational example

The direct reduction consumes coke, which can be estimated from the
charged amount of coke (method 1) or from the amount of carbon in top gas
estimated from the top gas chemical composition (method 2). In the example,
direct reduction consumes about 96 kg carbon/tHM. This means that the
iron oxides are reduced by gas to a ratio of 0.47 O–atom per Fe atom.
Estimated from charges per tonne ton ton kg/tHM C%

hour and hot blast volume burden HM coke
One charge 133.5 85 25.5
Coke 300 86
Coal 200 80
Production tHM/hr 500
charges/hr 5.9
Carbon input
C from coke t/hr 129
C from coal t/hr 80
Total input t/hr 209
Carbon output
in hot metal t/hr 22.5
in top gas t/hr 186.5
gasified in raceway t/hr 138.1
Carbon used for DR t/hr 48.4
for DR of Si, Mn t/hr 2.9
for DR of iron oxides t/hr 45.5
kg/tHM 91.1
O/Fe burden for DR 0.45
% DR (oxygen removed from iron oxides) 30.8
Estimated from top N2 H2 CO CO2 m³ STP kg C/tHM

gas composition
Bosh gas m³ STP/tHM 659 114 516 1289 276.2
% 51.1 8.9 40
Top gas % 46.3 5.1 26.1 22.6
m³ STP/tHM 659 72 372 321 1425 371.2
Carbon used for DR kg C/tHM 95.1
Hydrogen utilization 37 %
Table 6.3 Calculation of amount of carbon used for direct reduction
In addition to the direct reduction of iron (typically from FeO0.5) some

other materials are also directly reduced in the high temperature area of the
furnace.
The amount of carbon used for these direct reduction reactions is indicated
in Table 6.4. This can be calculated from the chemical composition and the
atomic weights, considering that the amount of oxygen removed reacts with
the carbon in the coke. Assuming that the iron oxides are directly reduced
from an O/Fe = 0.47 the carbon used for direct reduction is 101 kg C/tHM
corresponding to 117 kg coke/tHM.
Material reduced to hot metal kg C/tHM

FeO 0.5 Fe 94.50 % 101.5
SiO2 Si 0.40 % 3.4
MnO Mn 0.30 % 0.7
TiO2 Ti 0.05 % 0.3
P 2O 5 P 0.07 % 0.7
S in hot metal S in slag * 0.90 % 0.8
Total accompanying elements 5.8
Table 6.4 Typical coke consumption for direct reduction, including oxides
* – Desulphurization reaction (CaO) + [C] + [S] (CaS) + CO
In operating furnaces the variability of direct reduction is the major cause

of variation of the thermal level. Since direct reduction consumes coke as
soon as it increases, the burden descent starts to increase. Simultaneous
increases of the burden descent and the direct reduction as calculated from
the gas composition are a strong indication that preventive measures have
to be taken, such as increasing PCI. Figure 6.2 shows an example, where
direct reduction (white) and burden descent (see charges per hour in yellow)
increase, while hot metal silicon (green) decreases from 0.43 to 0.2 %. This
phenomenon can be observed in every furnace every few days, because
variability of direct reduction is due to variability of heat losses and/or raw
material quality.
Figure 6.2 Example of increased direct reduction and related effects

96 Chapter VI
6.6 Gas and burden composition during processing

In the preceding section the temperature profile in the blast furnace has been
detailed. In this section the gas in the furnace will be dealt with in more detail.
Wind is blown through the tuyeres, along with coal or natural gas and
moisture, generating gas at high temperature. The gas temperature is called
Raceway Adiabatic Flame temperature (RAFT, Tflame).
6.6.1 Raceway adiabatic flame temperature

The flame temperature in the raceway is the temperature that the raceway gas
reaches as soon as all carbon, oxygen and water have been converted to CO
and H2. The flame temperature is a theoretical concept, since not all reactions
are completed in the raceway. From a theoretical point of view, it should be
calculated from a heat balance calculation over the raceway. For practical
purposes linear formulae have been derived (see Example 6.4). But every
company has its own formula, which differs in the details considered and the
starting situation for which the formulae have originally been derived.
Flame temperature is normally in the range of 1900 to 2300 °C and is

influenced by the raceway conditions. The flame temperature increases if:
– Hot blast temperature increases.
– Oxygen percentage in the blast increases.
The flame temperature decreases, if:

– Moisture increases in the blast.
– Reductant volatiles increase, due to energy required for cracking the chemical
bonds.
– Reductant injection rate increases, since cold reductants are gasified instead of
hot coke. The precise effect depends on auxiliary reductant composition as well.
Table 6.5 provides some basic rules with respect to flame temperature effects.
Tflame Ttop ηCO

Blast temperature °C +100 52.8 –15.2 0.6
Coal kg/tHM +10 –26.6 9.4 –0.1
Natural gas kg/tHM +10 –75.8 20.1 –0.1
Fuel oil kgtHM +10 –45.7 13.5 –0.1
Oxygen % +1 33.9 –14.0 0.0
Moisture g/m³ STP +10 –42.5 8.6 –0.4
Table 6.5 Flame temperature effects, rules–of–thumb (calculated with mass and
heat balance for the reference furnace).
Example 6.4 Estimation of Flame Temperature (Tflame)

Flame temperature (Tflame) can be estimated from linearized formulae. The
formula below is in close agreement with more rigorous heat and mass
balance model calculated values. A notable improvement over the original
simple formulae is that it takes coal chemistry and cracking heat into account.
With the variety in coal chemistry and VM now in use, it is important that
the coal type is taken into account. For example, a high VM coal at a high
injection rate can have a real flame temperature 200 °C lower than a low
volatile coal. Note that for injectant, it is the concentration of injectant in the
blast (kg/m³), rather than kg/tHM, that affects flame temperature. Higher
oxygen enrichment and lower fuel rate operation both give a lower blast
volume/tHM leading to a greater impact on flame temperature of a given
injectant rate.
Tflame = (BT + 1961 – Coal/WC + 112.5OE – 1.992BM – 5203.9Gas/WC)

/
(1.287 + 0.02133OE + Coal/WC + 0.00257BM + 2.77687Gas/WC) (6.1)
= 1.1622.9439(5.467C(1–moist/100) + Qcrack /4.1868 + 11.945moist)

= 2.9439(0.03726H(1–moist/100) + 0.0086475moist) + 0.006938Ash(1–moist/100)
C = C in coal (dry %)
H = H in coal (dry %)
moist = coal moisture, as injected (%)
Ash = Ash in coal (dry %)
BT = hot blast temperature (°C)
Coal = coal rate (kg/tHM)
Gas = Natural gas rate (kg/tHM)
WC = specific blast volume, including oxygen enrichment (m³ STP/tHM)
OE = oxygen enrichment (O2 % – 20.9)
BM = blast moisture (g/m³ STP)
VM = coal volatile matter (dry %)
Qcrack = cracking heat of coal (MJ/kg) – can be measured or estimated from Equation 5.2:
Qcrack = 0.0007VM² + 0.0126VM – 0.3687 (If Qcrack < 0 then Qcrack = 0)
There is a connection between Tflame and top gas temperature: the higher Tflame,
the less gas is required for the heat required in the lower furnace (lower furnace
is the area with a temperature above 900 °C). As a consequence, there is less gas
in the upper furnace, and so top gas temperature decreases. However, there are
considerable response times for the increase and decrease of top temperature
after the change of blast conditions, up to 10 hours or more. Note that the
oxygen percentage in the hot blast determines the bosh gas volume and so has
an effect on the changes in Tflame. An example as estimated for the reference
furnace is given in Figure 6.3 on the next page.
98 Chapter VI
55
45
O2 +1 %
ΔTflame (°C) 35
25 PCI –10 kg/tHM
15
20 22 24 26 28 30
Oxygen in hot blast (%)
Figure 6.3 Effects on Tflame vary with blast conditions and depend especially on blast
oxygen contect. Effect of changes in Tflame are shown for 1 % additional
O2 enrichment and for a decrease of PCI rate by 10 kg/tHM.
6.6.2 Gas composition and ferrous burden processing

The gas in the blast furnace is processed in five steps. In the real blast furnace
these steps are overlapping.
Step 1: Wind is blown through the tuyeres along with coal or natural gas and moisture.
All these components react to form carbon monoxide (CO), hydrogen and
nitrogen. So, the conditions at the end of the raceway are a high temperature of
2000 to 2200 °C and CO, H2 and N2 in gaseous form.
Step 2: The gas ascends in the furnace and cools down to 1200 °C. The direct reduction
reactions take place, generating additional CO gas. On reaching 1200 °C, the
gas leaves the cohesive zone and enters the furnace stack filled with granular
materials. At temperatures over 1000 °C gas reduction is very limited, as any
CO2 formed reacts instantaneously with coke to CO, a reaction which is
thermodynamically equivalent to direct reduction.
Step 3: The gas continues to ascend, its temperature decreases from 1200 to 900 °C.
In this temperature range, the hydrogen is very effective and about 35 % of the
hydrogen picks up oxygen from the ore burden (against about 24 % of the CO).
At 900 °C the burden O/Fe ratio is 1, ηCO is around 24 % and ηH around 35 %.
2
Step 4: The gas continues to ascend cooling to 600 °C. At this temperature, the ore
burden has the composition of magnetite, Fe3O4.
Step 5: The gas cools down further to the temperature at which it will leave the furnace
at the top (110 to 150 °C). In this area, the carbon monoxide is utilized further
and removes more oxygen from the ore burden, although at temperatures below
500 °C, the reduction no longer proceeds.
In terms of gas volume, once the temperature of the gas has dropped below
1000 °C, the total gas volume in m³ STP per tonne of hot metal is constant,
only the composition of the gas changes, as shown in Figure 6.4. In this book
the basis of mass and heat balances are treated; an excellent overview has
recently been published (Cameron et al, 2020).
N2 O2 CO CO2 H2 H2O
Top gas, O/Fe=1.46, 150°C
Magnetite, O/Fe=1.33, 600°C
Wustite O/Fe=1, 900°C
DR O/Fe=0.5, 1000°C
Coke/coal gasification, 2200°C
Input tuyeres
0 500 1000 1500

Gas composition (m³ STP/tHM)
Figure 6.4 How top gas is formed from hot blast
It is clear from Figure 6.4, that the major part of the gas through the furnace
consists of nitrogen. Nitrogen is chemically inert and delivers only its heat from
the hot blast to the burden. During its five to eight seconds journey through the
furnace it cools down from the blast temperature to the top gas temperature.
For comparison in Figure 6.5, processing the burden is shown by demonstrating
the weight and composition of gas and burden in kg/tHM. The weight at the
top of approximately 1600 kg ferrous burden including gangue and 300 kg coke
decreases gradually to 1000 kg of hot metal and 250 kg of slag. The balance
is eliminated from the furnace with the top gas. How hot metal and slag are
formed is discussed in the next chapter.
How top gas is formed from hot blast How Burden is processed
N2 O2 CO CO2 H2 H2O Coal Iron (oxides) Si Carbon slag/gangue
Hematite Charged at top

O/Fe=1.46, 150 °C O/Fe 1.46
Magnetite Magnetite
O/Fe=1.33, 600 °C 600 °C, O/Fe=1.33
Wustite Wustite
O/Fe=1, 900 °C 900 °C, O/Fe=1
DR 1000 °C, O/Fe=0.5

O/Fe=0,5, 1000 °C
Coke/coal gasification Reduced iron

2200 °C
Input tuyeres HM and slag

1500 °C
0 250 500 750 1000 1250 1500 1750 2000 2000 1750 1500 1250 1000 750 500 250 0
Gas weight (kg/tHM) Burden weight (kg/tHM)
Figure 6.5 Blast furnace process showing counter–current gas and material flow and
the weights at varying temperatures in kg/tHM
6.6.3 The heat or energy balance

The energy or enthalpy input into the blast furnace consists of the energy
represented in the coke and coal and the sensible energy in the hot blast. Coke
and coal have an energy content of about 30 GJ/tHM. Per tonne, approximately
500 kg is used. Some 45 kg carbon is eliminated with hot metal, which contains
4.5 % carbon.
100 Chapter VI
Typical energy input into the blast furnace:

– 500 kg coke and coal, 15 GJ/tHM.
– Hot blast, typically 1.5 MJ/tHM.
– In total, some 16.5 MJ/tHM.
Typical outputs are:

– Chemical and thermal energy top gas: 5.5 MJ/tHM, of which around 35 %
is being reused for fueling the hot blast stoves. The balance is mostly used for
power generation.
– Chemical energy hot metal: Fe 7 GJ/tHM, C 1.5 GJ/tHM and Si 0.2 GJ/tHM
– Sensible heat around 2 GJ/tHM: hot metal (64 %), slag (14 %), top gas (12 %)
and heat losses (10 %).
– In total it is 16.2 GJ/tHM. The small differences are for instance due to use of
rounded numbers, neglect of chemical reactions of e.g. Mn and tuyere heat loss.
Basically, producing hot metal costs 16 GJ/t, of which 5 GJ/tHM is lost/

expelled with top gas. The heat balance as derived for the standard furnace is
presented in Figure 6.6. To put this into context, the net energy requirement for
the reference furnace equates to the output of one of the two units of the most
recent nuclear plant commissioned in China (Taishan), at 1.75 GW.
18
15
Top Top Gas CV

Energy (GJ/tHM)
12 Coke
Heat Loss
9
6 Hot Metal
Taphole
Coal
3
Tuyeres
Heat Heat Liquids
0
Input Output
Figure 6.6 Energy input and output in a blast furnace; heat is the sensible heat of the
hot blast (left) and cooling water, top gas, hot metal and slag (right).
Another method to describe the counter–current mass and heat exchange in

the blast furnace is with a graphical tool in the Rist–diagram. This is more
advanced and shown in Annex VII.
6.6.4 Where are coke and coal consumed?

In the typical blast furnace, the oxygen blown into the furnace via the tuyeres
is converted to CO in the raceway. The oxygen input per tonne of HM coming
from the hot blast is typically 240–260 Nm³ O2/tHM. The blast oxygen is used
to gasify the coal and some additional coke (see Example 5.1 for gasification of
coal).
The direct reduction consumes coke. It takes place in and probably above the
area of the cohesive zone at burden temperatures between 1000 and 1400 °C.
The carbon dissolution takes place at slightly higher temperature as soon as the
hot metal is liquid. The results are presented graphically in Figure 6.7.
Figure 6.7 Coke and coal use in a blast furnace (left), local basicity and liquidus
temperatures
As a result of coke being consumed, the coke ash is blended into the slag.
Since coke and coal ash consist mainly of acid components (SiO2 and Al2O3),
the basicity of the slag decreases and the slag liquidus temperature is reduced
as well. The example above is derived from the reference blast furnace used
throughout this book.
It has to be mentioned, that slag formation at the tuyeres is a special case.

For every tonne of hot metal, 200 kg coal and 135 kg coke are gasified in
the raceway generating some 32 kg slag per tonne of hot metal. Some slag
components are gasified, especially SiO2, of which around 10 kg is reduced to
atomic silicon dissolving in the hot metal. This makes the slag components in
front of the raceway a slag high in Al2O3 with liquidus temperatures well above
1800 °C.
Another important aspect is, that injection coals with a high ash percentage
force the operator to make the “primary slag” in the cohesive zone more basic
in order to reach the same final slag basicity. This can jeopardize melting of the
ferrous burden.
102 Chapter VI
6.7 Permeability
There is a large difference between the permeability of coke and burden. Coke is
larger than ferrous materials and has more voidage. Ferrous burden average size
is much smaller and contains more fines (material smaller than 5 mm); hence
its gas permeability is poorer. Figure 6.8 shows the pressure drop required for
a standard flow in mm water gauge per meter bed height. From the data of the
figure: resistance to gas flow is 4–5 times as high for a sinter burden as for coke.
1000
(mm WG/m of bed depth)
800
Sinter
Pressure Drop
600
Pellets
400
200
Stockline
Coke
0
0 10 20 30 40 50 60
Harmonic Mean Size (mm)
Figure 6.8 Permeability difference: coke (HMS 40 mm) and sinter (HMS 14 mm)
In a typical furnace with a sinter burden, over 80 % of the resistance to gas flow
comes from the ferrous burden. The surface mean (harmonic mean) diameter
for coke is about 40 mm whereas for sinter it is typically 14 mm. The pressure
drop chart shows that coke is permeable for gas, and ferrous burden has a
relatively high resistance for gas flow. Gas flowing through a blast furnace takes
the path of least resistance, making the burden distribution the major tool for
gas flow control as discussed in the next chapter.
6.7.1 Permeability: the K–factor

The hot blast is pressurized by a blower and is transported through the furnace,
because the pressure at the top is lower than the pressure in the hot blast main.
Blast furnace operators are interested in the permeability of the material column
inside the furnace, because the more blast can be transported through the
furnace, the higher the production, since more coke is gasified. The furnace can
be driven to higher production to a maximum ΔP (blast pressure minus top
pressure), but when exceeding the ΔP above this maximum, the burden descent
deteriorates, the furnace becomes less efficient and productivity will decrease.
How are blast pressure and blast volume related? In its simplified form the
equation of Bernouilli states that
P = ½  ρ  v² (6.7.1)
Where P is the (dynamic) pressure, ρ the density (kg/m³), v the velocity (m/s).
Or in other words, to transport twice the amount of gas through a blast

furnace, doubling v, pressure has to increase by a factor of four.
Blast furnace operators are interested in the permeability. This is characterized

by monitoring how much pressure is required to transport the gas through the
furnace, as indicated by the permeability factor K:
K = [(BP + 1.033)² – (TP + 1.033)²] / (BGV)1.7 (6.7.2)
Where BP is blast pressure and TP is top pressure in g/cm², BGV is bosh gas
volume in m³ STP/min. The factor 1.033 has to do with adjustment from
atmospheric to absolute pressure (g/cm² = bar  1019.72). Note that the K–factor
is called the permeability factor, but the higher the K–factor the poorer the
permeability.
Some companies use partial–K–factors for upper and lower part of the furnace.
This provides good information over the permeability of the cohesive zone
(lower part) and upper part of the furnace. The total K–factor remains the sum
of the partial–K factors. Different companies use different characterizations, for
instance a resistance index or use ΔP/blast volume.
K–factors are thought to reflect the properties of the burden. However, there are
many more influences that have impact on the ΔP and hence on K–factors.
1. Variability: the pressures in a blast furnace are rather variable with typical
minute–to–minute variations of ±5 %. Since in the short term the material
column is constant, this is caused by the constantly changing permeability in
the cohesive zone, as iron and slag drip through it.
2. Channeling. Channeling is the process whereby the gas follows a preferential
path along the wall in the furnace. The hot gas will also lead to high heat loads
on the cooling system and to short–circuiting in the pressure taps; that means
the same pressure at different heights.
3. Top gas temperature and water vapor are not taken into account in the K–
factor, while top gas temperature has an impact (section 6.7.2).
6.7.2 The pressure differences over the furnace height

The pressure difference between hot blast and top pressure depends on the
blast volume and the furnace itself. The ‘allowable’ maximum ΔP depends
on the furnace and the burden, and is a specific value for each furnace. Large
14 meter hearth diameter furnaces can work up to a maximum ΔP of 2 bar
or even higher. Note that this is an instantaneous maximum. Some furnaces
have systems which automatically lower the blast volume when reaching the
maximum ΔP in order to prevent hanging, slips and process upsets. Smaller
furnaces have maximum ΔP of 1.5–1.8 bar.
104 Chapter VI
The pressure difference can be subdivided into three parts:

– over the tuyeres
– over the lower part of the furnace (some 8 meters above the tuyeres)
– over the upper part
Figure 6.9 Typical pressure difference over a blast furnace
The pressure difference over the tuyeres (ΔPtuy) is typically 0.35–0.65 bar,
depending on wind velocity in the raceway. The wind velocity in the tuyeres can
be estimated from Formula 6.7.3.
Vtuy = ((Tblast+273.15)/273.15)((1.01325/(Pblast+1.01325))V blast/(Atuy60) (6.7.3)
where Vtuy is the tuyere velocity in m/s

Tblast the hot blast temperature, °C
Pblast the blast pressure, bar
V blast the blast volume, m³ STP/min
Atuy the total tuyere surface diameter, m²
The pressure difference over the tuyeres is dependent on wind velocity. A higher
wind velocity yields a higher ΔPtuy as shown in Figure 6.10. The ΔPtuy (in bar)
can be estimated from a linearized formula 6.7.4 below.
ΔPtuy = Vtuy  0.0035 – 0.353 (6.7.4)
0.6 300
0.5
Tuyere velocity (m/s)
ΔPtuyeres (bar)
0.4 250
0.3
0.2 200
0.1
Europe SE Asia
0 150
150 175 200 225 250 275 10 11 12 13 14 15 16
Wind velocity (m/s) Hearth Diameter (m)
Figure 6.10 Wind velocity and pressure difference over the tuyeres (left) and
benchmark of tuyere velocity in selected operating furnaces (right)
Internationally, there are different tuyere velocities being used by South–East

Asian operators and European and North American operators (Figure 6.10).
South East Asian operators prefer to work at tuyere velocities of 240–260 m/s
while in Europe and North America, 200–220 m/s is the preferred range. The
former prefer a large penetration of the hot blast in order to minimize the size of
the deadman. To this end, long tuyeres (longer than 60 cm) are frequently used
as well. The latter take into account a lower minimum tuyere velocity, because
wind distribution over the tuyeres is sufficiently symmetric. Moreover, they
prefer to control the gas flow with burden distribution and use shorter tuyeres
(40–50 cm). This makes the tuyeres less vulnerable.
Some companies place a greater importance on the kinetic energy, also called
blast momentum, per tuyere (kg.m/s), rather than the tuyere velocity. Raceway
probing with an accelerometer has demonstrated that this is the more important
parameter to maintain the raceway stability. If the kinetic energy is too low
then the raceway becomes unstable and the active raceway shrinks periodically.
This will increase the heat load on the tuyeres, and cause more wall gas flow.
KEblast = 0.5  (Bflow/Ntuy)  Vtuy  Vtuy/9.81 (6.7.5)
where KEblast is the blast kinetic energy (momentum) in kg.m/s

Bflow is the blast flow in kg/s
Ntuy the number of tuyeres
Vtuy as in 6.7.4
Biswas quotes a range of 3000–9000 kg.m/s. European plants tend to operate

between 3000 and 9000 kg.m/s depending on hearth diameter. On large Asian
furnaces values up to 12500 kg.m/s are more typical.
The pressure difference over the upper part of the furnace is determined by the
size distribution of the burden and coke rate. But this is difficult to use as a tool
for the quality of the burden, since the pressure difference varies not only with
gas flows through channels, but also with the top temperature and moisture
content of the top gas. This is because at higher temperature gas expands and
more real m³ of gas have to be transported through the upper furnace.
An example of the relation between ΔP of the upper furnace and the top
temperature is shown in Example 6.5 on the next page. High moisture to be
eliminated from the furnace has a similar effect on ΔP of the upper furnace.
Some operators have reported an increase in pressure difference over the upper
zone when they have charged excessive fines; this has led a few hours later to
problems in the lower zone.
106 Chapter VI
Example 6.5 Pressure drop and top temperature

The pressure needed to drive gas through a blast furnace increases with
increasing top gas temperature. Why? At higher temperature gas expands,
so more real m³ of gas have to be transported, especially through the upper
furnace. As a consequence of the increase in ΔP the K–value deteriorates.
This can be observed in operational practice, and an example is shown below
in Figure 6.11. In this case the relatively high top temperature was caused by
asymmetric process as a consequence of worn throat armor.
0.51
0.49
ΔP in upper zone (bar)
0.47
0.45
0.43
0.41
0.39
75 100 125 150 175 200
Top Gas Temperature (°C)
Figure 6.11 A typical example of the relation between top gas temperature and ΔP
upper furnace (14 m furnace, inner volume 4050 m³)
6.7.3 PCI and permeability

In current operation scenarios, a relatively large part of the reductant
requirement of a blast furnace comes from auxiliary injectants like pulverized
coal (PCI) and/or natural gas. When increasing the injection level, what
happens with the permeability of the blast furnace? When increasing reductant
injection, part of the coke is replaced by burden, hence the permeability of the
blast furnace becomes poorer. This is illustrated in Figure 6.12.
What happens if the PCI rate is increased from 60 kg/tHM to 200 kg/tHM?
At 60 kg PCI/tHM the coke layer and ferrous layer are the same thickness. If
the PCI is increased by 140 kg/tHM to 200 kg/tHM, some 120 kg/tHM coke
has to be taken out of the charge. So, the coke layer decreases and the resistance
will increase (Figure 6.12).
Operational practice is somewhat more complicated.

1. When increasing PCI, ΔP increases and this is compensated by increasing the
oxygen enrichment of the hot blast. Since the furnace then operates with less N2
per tonne, the gas volume per tonne decreases and the same production can be
made with less gas, so with a lower ΔP.
2. If PCI is increased without increasing oxygen enrichment, then top temperature

will increase. And this will lead to a higher ΔP over the upper furnace. An
operational example is elaborated in Example 6.6.
3. However, if the furnace is operated at higher PCI, the weight of the material
column above the cohesive zone increases, and hence the maximum acceptable
ΔP increases as well. This is illustrated with an operational example in Figure
6.12, where ΔPmax is the momentary value of ΔP, which increases at decreasing
coke rate and the weight of the burden column above the cohesive zone is
estimated. When ΔPmax is exceeded, the blast volume is reduced.
Figure 6.12 Effect of Increasing PCI on burden layers
When coke rate decreases, the weight of the burden increases. As will be
explained in Chapter VII, the higher weight will allow for a higher maximum
value of ΔP, because the weight of the burden and coke above the cohesive zone
increases. It can be estimated and Figure 6.13 shows the weight (in kgf) and the
maximum allowable ΔP as a momentary value for an operating furnace.
2.50
Weight (kgf/cm²) ΔPmax (bar)
Weight (kgf/cm²)
DeltaPmax (bar)
2.25
2.00
1.75
250 300 350 400 450
Coke rate (kg/tHM)
Figure 6.13 Increasing burden weight (calculated) allows a higher maximum

allowable ΔP for a blast furnace. The dotted line is from operating
instructions of a 4500 m³ inner volume blast furnace.
108 Chapter VI
Example 6.6 PCI rate, coke rate and permeability.

Increasing PCI rate at constant oxygen enrichment will lead to a higher top
temperature (Table 6.5). Hence, increasing PCI with its consequent reduction
of coke rate (at constant blast volume and oxygen enrichment) will result in:
– An increase of ΔP because permeable coke is replaced by less permeable
ferrous burden.
– An increase of ΔP because top temperature increased.
An example of blast furnace operation is shown in Figure 6.11 from the

early days of coal injection in Europe. At the time, the plant was short on
oxygen capacity and thus operators tried to reach as high PCI rates as possible
without additional oxygen enrichment. It is clear from the figure that the
increase in ΔP is caused by the increased upper ΔP.
More than 60 % of the ΔP increase was due to the increased top temperature
and less than 40 % to the replacement of coke by ferrous burden. The
dotted line in Figure. 6.14 is the calculated pressure difference taking only
replacement of coke by ferrous burden into account. The difference between
the calculated (red–dotted) line and the measured (solid) line is due to
increased top temperature.
1.2
1.0
0.8
ΔPprocess (bar)
0.6
0.4
Total ΔP (measured)
0.2
Total ΔP (estimated)
Lower furnace
0
275 300 325 350 375 400
Coke rate (kg/tHM)
Figure 6.14 ΔPprocess as measured inside a blast furnace at decreasing coke rate. The
dotted line is calculated taking only the higher burden fraction at lower
coke rate into account
6.8 Blast furnace efficiency

The efficiency of the furnace is the amount of reductant used per tonne of
hot metal. It is the sum of coke, nut coke and auxiliary injectants. The total
reductant rate (in kg/tHM) can be standardized calculated as
Reductant rate = coke rate + nut coke rate + RRcoal rate + RRgas rate (6.8.1)
Some companies use a standardized coke, for instance with 87.5 % C, to

make a consistent total reductant rate over the years. Actual coke and nut coke
chemistry are recalculated to the standard coke and replacement ratios (RR) are
determined with respect to the standard coke. Some companies use as the total
reductant the Formula 6.8.2:
Reductant rate = coke rate + nut coke rate + coal rate + gas rate (6.8.2)
If management focuses too much on the results of 6.8.2, then optimization by

higher coal injection is frustrated, since injection coals have a replacement ratio
of typically 0.9, so the reductant rate increases at higher injection rates. In fact,
higher PCI is attractive:
– costs are lower since PCI coals are much cheaper than coking coals, especially if
the PCI equipment has spare capacity.
– environmentally more friendly, since the coking process is partially avoided.
The total amount of reductant per tonne decreases when the heat input into
the furnace is larger (for example higher hot blast temperature) or the heat
requirement is smaller. The major parts of the heat requirement are
– the heat used for the direct reduction reactions,
– the heat required for bringing hot metal and slag to casting temperature and
– the heat losses through the furnace wall.
The gas composition of the top gas is often indicated as “Top Gas Efficiency”,
“Top gas Utilization” or ηCO (as % in dry top gas):
CO
CO = CO + CO
2
# 100 %
2 (6.8.3)
ηCO is typically in the range of 45–50 %. Increase of ηCO means, that the oxygen
from the burden is removed with less gas thus with less coke and the furnace
becomes more efficient. A similar factor is the hydrogen utilization,
HO
H = H O 2+ H # 100 %
2
2 2 (6.8.4)
The ηH can be estimated from the input of H2 through the tuyeres and the
2
output with the top gas, the balance being the H2 utilized in the process. ηH is
2
especially important for furnaces using natural gas.
110 Chapter VI
The main factors determining the efficiency are:

– The better the contact is between gas and ferrous burden, the better the gas
reduction and the lower the direct reduction reactions and the reductant rate.
Another aspect is the amount of fines smaller than 5 mm that are charged or
generated in the furnace. This is important because more fines block the gas
flow and decrease the permeability of an ore layer, especially when the fines are
concentrated in certain radial areas, as for example at the wall. This is discussed
in Chapter VIII.
– How easily is oxygen removed from the ferrous burden? This is largely
determined by the size distribution and softening/melting properties of burden
materials. Lump ore is less accessible to gas than sinter or pellets, so lump ore
has a penalty, which varies from type to type but is typically an additional 50
kg coke for a tonne of hot metal made from lump ore.
– How much slag has to be processed per tHM.
It is often difficult to grasp the actual picture of the efficiency of the furnace in
its control room, since different parameters are used for different reductants,
for example tonne of coal or m³ of natural gas per hour and coke rate per tonne
of hot metal. Operators have to know how to “translate” these parameters to
reductant rate (kg/tHM). Rules of thumb are in Annex III.
On–line monitoring of the efficiency is possible by close attention to the

chemical composition of the top gas. However, this is not easy either, since there
are many effects on the gas composition.
The complications with the interpretation of the gas utilization come from:
– Interference with driving rate of the furnace: the faster the burden descends, the
more oxygen is removed from the burden. If the driving rate increases, then the
top temperature will go down.
– Interference with burden level: if burden level is low due to charging problems,
gas utilization will decrease, since the first reduction step from hematite to
magnetite does not take place in the top of the stack. When filling the furnace,
the gas utilization improves above the earlier level.
– Competition between hydrogen and carbon monoxide. Part of the oxygen
of the burden is removed with hydrogen. So, as soon as an operator changes
the setting for moisture, the CO gas utilization changes. The same applies for
changes in set points of injectants.
Another aspect that follows from the gas analysis is the amount of coke that is
used in direct reduction reactions. This becomes manifest from the percentage
of CO + CO2 in the top gas. However, this percentage also varies with the
moisture and oxygen percentage in the hot blast, so it is recommended to have
an on–line calculation available that expresses the direct reduction in carbon
rate or coke rate per tonne of hot metal as shown in Section 6.5.
6.9 Productivity
The productivity of a blast furnace is the amount of hot metal produced per
unit time. It is often expressed as tonnes of hot metal produced per 24 hours
(without delays) either per m³ working volume or per m³ inner volume or per
m² hearth surface. High productivity for large furnaces is 2.4 tonnes of hot
metal per m³ inner volume per 24 hours. Smaller furnaces tend to have higher
productivity, because there is a smaller deadman beyond the raceway.
The limiting factors for the production rate can be

– the process: how much wind the furnace accepts?
– the casthouse: how much liquid can be drained?
What determines the productivity of a blast furnace? It is a simple question with

a complicated answer. The areas to consider are:
– Burden quality and slag volume;
– Furnace efficiency and the minimum coke rate that can be attained;
– Process conditions (hot blast pressure, temperature, moisture, O2 enrichment).
– Equipment: charging capacity, capacity of turbo blowers, availability of oxygen.
6.9.1 Burden quality and slag volume

The blast furnace burden consists of sinter, pellets and lump ore. Lump ore
is very variable in quality and is in general not helpful for high productivity.
Properties of burden materials for high productivity have been discussed in
Chapter III.
An important factor is the slag volume per tonne of hot metal. It has been
observed that over a wide range of slag volumes, productivity deteriorates with
increasing slag volume. This was found for similar size furnaces at different
plants, as well as for different size furnaces at one site. Typically, productivity
decreases by 0.05 t/m³ IV.24hrs for an increase in slag rate of 10 kg/tHM. The
slag effect is the consequence of improved permeability of the cohesive zone
resulting from less slag and of the reduced heat requirement.
The charge of metallic iron units (scrap, DRI–Direct Reduced Iron, HBI–Hot
Briquetted Iron) lowers the reductant rate and increases the furnace efficiency,
since metallic iron is already reduced. Details are in Chapter III.
6.9.2 Furnace efficiency.

The productivity of a furnace increases as the reductant rate decreases, since
in a more efficient furnace more hot metal is produced with the same amount
of reductant. In front of the tuyeres, about 350 kg coke and coal is gasified per
tonne of hot metal, since about 100 kg coke is used for direct reduction and
some 50 kg for dissolving carbon in hot metal. This means that for every 3.5 kg
lower reductant rate, production increases by 1 %.
112 Chapter VI
6.9.3 Process conditions for high productivity

The more coke that is gasified at the tuyeres, the more iron is produced. More
coke is gasified by blowing more hot blast or using more oxygen at the same
blast volume. The following limitations are met:
– ΔPmax: a furnace cannot exceed the maximum ΔP: burden descent starts to
deteriorate above ΔPmax or, with constant top pressure, maximum blast pressure.
– O2 enrichment: the more oxygen that is present in the hot blast, the more coke
that is gasified and the more productivity increases. However, there is a limit
to the amount of oxygen enrichment that can be used: increase of oxygen
percentage in the hot blast leads to a lower top gas temperature. As soon as the
top temperature fallsbecomes too low (below 100 °C), the furnace becomes less
efficient, since it takes more time to start the reduction reactions.
– Metallurgical coke rate: the metallurgical coke layers (so the coke without the
nut coke) form the grid in the cohesive zone through which gas ascends and
the melting hot metal and slag descend. If the coke layers become too thin,
there will be contact between sequential ferrous layers and the gas flow through
the furnace becomes unstable. The minimum metallurgical coke rate also
depends on the slag volume that has to percolate through the grid. A minimum
metallurgical coke rate means a maximum PCI with the highest oxygen
enrichment, and this drives productivity. Minimum metallurgical coke rates are
reported at 230–240 kg/tHM, plus nut coke rates around 30 kg/tHM.
Maximum productivity (with a given burden and injectants) is reached when:

– Top pressure is at the maximum value.
– Blast volume is set so that the furnace is operated to the maximum ΔP.
– Hot blast temperature is set at the highest achievable value, since it lowers the
fuel rate and thus the gas flow.
– Moisture is at ambient level.
– Fuel injection (coal, gas, oil) is at the maximum the furnace accepts, which
means the minimum coke rate.
– The top gas temperature is controlled to above 100 °C with oxygen injection.
– The furnace operates symmetrically circumferentially, meaning that all sections
representing one tuyere contribute equally to the process (see Section 11.4)
– Casting regime makes process independent from casthouse operation.
6.10 Process symmetry
The process in a blast furnace is very asymmetric: in the radial direction gas
flow and composition in the center are very different compared to the wall,
which means that different hot metal and slag compositions are simultaneously
produced in the center and wall area. Blending of the liquids in the taphole
means that these differences cannot be observed. Gas flow in the center has
higher temperatures and lower ηCO, which can be observed, but is also mixed
in the gas system. Around the circumference there can also be deviations in
symmetry, resulting from equipment issues, damaged throat armor, different
diameter tuyeres with varying amounts of injection and so on.
6.10.1 Radial asymmetry

Gas composition and temperature vary strongly along the radius. In Figure 6.15
the ηCO and temperature are shown, as measured with a retractable in–burden
probe. In this example, the ηCO varies from 10 % in the center to close to 60 %
at the wall, and the temperature from 100 °C at the wall to 600 °C in the center.
60
50
40
ηCO (%)
30
20
10
0
0 20 40 60 80 100
wall Radial distance from wall (%) center
Figure 6.15 Radial asymmetry of ηCO and temperature, 6 m below stockline
The burden descend velocity varies over the radius as well. This is to be
expected, since the burden descends where the ferrous burden is placed
(Chapter VII). An example of radial burden descent is shown in Figure 6.16, as
measured in an operating furnace.
60
Descending velocity (cm/min)
30
0
0 20 40 60 80 100
wall Radial distance from wall (%) center
Figure 6.16 Radial descending velocity in an operating blast furnace (2800 m³ IV)
6.10.2 Circumferential symmetry

For high performance all tuyeres have to contribute equally to the process.
Consider a tuyere as representing a section of the furnace as shown in Figure
6.17, Example 6.7. If one part of the furnace does not produce the same amount
as other parts, the total will suffer. Operators can observe asymmetry in the
process from various parameters: for instance, from the burden level: one stock
rod at the top gives a different burden level from another; from differences in
gas temperatures in the uptakes; from asymmetry in the heat losses; and from
differences in hot metal temperature between two tapholes.
114 Chapter VI
6.10.3 Circumferential symmetry of injection, wind distribution

If every tuyere in a blast furnace is considered as part of the blast furnace
“pie” and is responsible for the process to the stockline, it is self–evident that
the circumferential symmetry of the process has to be assured to reach high
performance. This means: symmetrical wind distribution and symmetrical
injection of coal and gas.
The various systems in use for PCI have different methods to ensure a good
distribution. However, the largest deviation from circumferential symmetry
occurs when no coal is injected in a particular tuyere. If no injection is applied,
the production rate at that particular tuyere increases substantially (Figure 6.17
and Table 5.5). Consequently, the blast furnace operator has to take care that
all tuyeres are injecting coal. In particular, where two tuyeres next to each other
are not injecting coal the equalizing effects between the tuyeres are challenged.
Especially if the furnace is operating at high PCI rates, the situation is rather
serious and short–term actions have to be taken to correct the situation.
The wind distribution is equally as important as the distribution of injectants.

Operators tend to work with smaller tuyeres above the tapholes, since a lower
production rate above the tapholes stabilizes the taphole mushroom and taphole
length. When applying smaller tuyeres, the injectants should be maintained
on the tuyeres, otherwise the production increases, even with smaller tuyeres.
When replacing a 145 mm tuyere with a 125 mm tuyere the wind volume
decreases by 74 %. A similar way of reasoning applies when a tuyere is partially
blocked, by slag or ash deposits.
6.10.4 Factors affecting circumferential symmetry

Asymmetry in the process can arise from various sources:
– Asymmetry of the charging. With a bell–less top this can be prevented by
changing the rotational direction of the chute and by alternating the coke and
ore top bins. Note that the changes have to be made on a time scale smaller
than the blast furnace process i.e. every 1–2 hours or 5–11 charges.
– Blast distribution. Blast distribution is affected by plugged or small tuyeres
(above a taphole or refractory hot spots) and slag deposits in the tuyere.
Asymmetry is also the result of too low blast speeds (under 100 m/s): tuyeres
will not function efficiently as blast distributors.
– Worn refractory or throat armor plates at the top of the furnace.
– From uneven coal injection. Especially tuyeres without PCI.
– Deviation of furnace center line from vertical line. This is especially a concern
in older furnaces.
Example 6.7 Circumferential symmetry in operation

In Figure 6.17 one tuyere is not injecting coal. What is the effect? As
described in Chapter V, one section of the furnace receives more coal and thus
has a lower production rate and another section the reverse. As a consequence,
the cohesive zone will become asymmetric. This can be observed in heat losses
as shown in Figure 6.18, leading to a net shortage on one side of the furnace.
Normal Operation One Tuyere Off
Figure 6.17 Effect of a tuyere without coal injection
Figure 6.18 24 hrs heat loss distribution (blue). Note a slight process asymmetry.
One day graph of eight sections, four levels.
116 Chapter VI
A relatively unknown cause of asymmetry is that the rotation of the chute

does not reach perfect symmetry of distribution, for example there is a
difference related to the flow from top bins, especially with double hopper
systems. Remedial action is to change the rotational direction of the chute
frequently, for example every 1–2 hours. Results are shown in an example in
Figure 6.19.
Figure 6.19 Effect of frequency of changing the rotational direction of the chute as
shown by the variation of the temperature of the upper row of staves.
Change of rotation: left hand every 50 charges (every 10 hours), right
hand every 10 charges (every 2 hours).
VII Burden Distribution
A blast furnace is a huge gas reactor and the method of choice to distribute the
gas over the radius is by making the proper burden distribution. Gas always has
the highest flow at the path of least resistance, just as water always flows to the
deepest point. For this reason, a separate chapter about burden distribution is
required in a book about blast furnace operation.
7.1 The movement of gas and burden
The gas generated at the tuyeres and at the melting zone has a short residence
time of 6 to 12 seconds in the blast furnace (Section 2.3). During this period,
the gas cools down from the flame temperature to the top gas temperature,
from 2000 to 2200 °C down to 100 to 150 °C, while simultaneously removing
oxygen from the burden. The vertical distance between tuyeres and stockline is
approximately 22 meters. So, the gas velocity in the furnace is rather limited,
in a vertical direction about 2 to 5 m/s during the 6 to 12 seconds the chemical
reactions take place. The burden descends in the blast furnace from top to
bottom. Figure 7.1 shows a representation of the burden descent. It is indicated
with mechanical stock rods, which are resting on the burden surface and
descending with the burden between charging. The burden surface descends
with a speed of 8–15 cm/min.
Zero level burden
1.30 m (Stockline)
1. Stock rod descending 1

2. Stock rod on burden 2 4
3. Stock rod descending
with burden 3
4. Stock rod ascending
for charging
6m
Figure 7.1 Stable burden descent measured with mechanical stockrods

118 Chapter VII
In order for the burden to descend, voidage has to be created somewhere in the
furnace. Where is this voidage created? (See also Figure 7.2)
– Firstly, coke is gasified in front of the tuyeres, thus creating voidage at the
tuyeres.
– Secondly, the hot gas ascends the furnace and melts the burden material. So, the
burden volume is disappearing into the melting zone.
– Thirdly, the dripping hot metal consumes carbon. It is used for carburization of
the iron, as well as for the direct reduction reactions; so, below the melting zone
coke is consumed.
It is possible to indicate how much each of the three mechanisms contributes

to the amount of voidage created. A large part of the voidage is generated at the
melting zone. In a typical blast furnace with a high PCI rate, only about 20 %
of the voidage is created at the tuyeres and less than half of the 300 kg/tHM
coke is gasified at the tuyeres.
Figure 7.2 Creation of voidage in the blast furnace
The creation of voidage is quantified for the furnace example and shown in
Figure 7.3. Example 7.1 shows that the mass flow of material is strengthened
towards the ring where the highest amount of ore is charged into the
furnace. Therefore, at low coke rates high ore concentration at any ring in the
circumference, especially in the wall area, has to be avoided.
Burden Distribution 119
Example 7.1 Where and when is voidage generated?

Voidage in a blast furnace is mainly generated where iron ore melts. By
melting, the solid ferrous burden becomes liquid and drips down to the
hearth. Direct reduction takes place where the burden is melting, as well as
carbon dissolution. Most voidage is created where material is melting – only
20 % of the voidage is created by gasifying coke (Figure 7.3).
Cohesive Zone
Dripping Zone
Raceway
input coke ferrous total

charged kg/tHM 300 1570
specific weight kg/m³ 550 1800
+10 % compression kg/m³ 605 1980
volume m³/tHM 0.50 0.79 1.29
eliminated coke ferrous m³/tHM %

melting kg/tHM –1570 –0.79 62
direct reduction kg/tHM –110 –0.18 14
C dissolution kg/tHM –52 –0.09 7
Coke gasified kg/tHM –137 –0.23 18
Total m³/tHM –1.29
Figure 7.3 Creation of voidage as quantified for the reference blast furnace
Another aspect of the voidage in the blast furnace is, that the actual voidage
depends on local fluidization of material by the gas flow. If gas flow is locally
high, lumps of material are a little farther away from each other than if the
gas flow is low. So, compaction of materials is continuously changing. And, of
course, the fluidization has an effect on voidage and thus blast pressure.
120 Chapter VII
An example is in Figure 7.4, where the effect of stove changes is shown on

blast pressure. Blast furnaces that do not compensate for the additional
blast requirement during a stove change, are faced with a slightly lower blast
volume into the blast furnace and thus a lower blast pressure. In Figure 7.4b
the blast pressure during the minutes after a stove change is shown. Blast
pressure for the same blast volume is higher after a stove change and recovers
in 5–10 minutes. A similar observation can be made when the furnace is
stopped; the burden descends during a stop and voidage decreases. The
increase/decrease of voidage can also be seen in the effect of casting.
3.4 6700 3.30
Pressure Volume
Blast Volume (m³ STP/min)

3.3 6500
Blast pressure (bar)
Blast pressure (bar)

3.2 6300
3.25
3.1 6100
3.0 5900
2.9 5700 3.20

6 hours 0 5 10 15 20
Minutes after stove change
Figure 7.4 Effect of stove changes on blast pressure

Left: blast volume and pressure for 6 hours showing the stove changes
Right: the average effect of 7 changes in the 6 hours of left graph
Note that the blast pressure in any blast furnace shows a very high variability.
In Figure 7.4 and Figure 7.5 the variability is 0.12 on a total ΔPprocess of 1.2
bar within an hour – and half of it occurs within minutes. 0.12 bar means
that voidage varies by 5 %. Compared to different industrial processes, like
direct reduction plants or air separation units, the variability is extreme. The
blast furnace is extreme, because voidage can vary because of blockage or the
opening of coke slits with liquids.
Zero level burden
Hanging
Slow Descent
Slipping
6m
Figure 7.5 Irregular burden descent

Sometimes the burden descent of a blast furnace is erratic. What is the

mechanism? Cold burden materials and coke flow rather easily through bins,
as can be observed in the stock house of a blast furnace. Hence in the area in
the blast furnace where the material is solid, the ore burden and coke flow with
similar ease to the void areas. Nevertheless, blast furnace operators are familiar
with poorly descending burden (Figure 7.5).
Also, the phenomenon of “hanging” (no burden descent) and “slips” (fast
uncontrolled burden descent) are familiar. From the analysis in this section
it follows that, in general, the cause of poor burden descent must be found in
the configuration of the cohesive zone. The materials “glue” together and can
form internal bridges within the furnace. Poor burden descent arises at the
cohesive zone. The effect of a slip is that the layer structure within the furnace is
disrupted and the permeability for gas flow deteriorates (Chapter XII).
7.2 Burden descent: system of vertical forces
The burden descends because the downward forces of the burden exceed
counteracting upward forces and push material into the voidage that is created.
This idea has in particular been developed by Cor van der Vliet, co–author
of the 1st and 2nd editions of this book. The most important downward force
is the weight of the burden; the most important upward force is the pressure
difference between the blast and top pressure or, to be more precise: the pressure
difference between the lower side of the cohesive zone and the top ΔPprocess.
(Figure 7.6). Most operators, however, are used to the total ΔP between hot
blast pressure and top pressure.
Figure 7.6 Pressure difference in a blast furnace, the upward pressure is the ΔPprocess
The cohesive zone is an area with high resistance to gas flow, since cohesive
material is impermeable to gas and the coke slits fill up with melting slag and
hot metal. It leads to a large pressure drop over the cohesive zone and to a large
upward force. The upward and downward forces are estimated for the example
furnace and shown in Figure 7.7 on the following page.
122 Chapter VII
The coke submerged in hot metal also exerts a high upward force on the burden
due to buoyancy forces (Figure 7.8), as long as the coke is free to move upwards
and does not adhere to the bottom. In the figure it was calculated as coke
submerged in 1 meter of hot metal. In addition to the upward force arising from
the blast pressure, friction forces from the descending burden influence burden
descent: the coke and burden are pushed outward over a cone of stationary or
slowly descending central coke. The wall exerts frictional forces on the burden.
Figure 7.7 System of vertical forces in the blast furnace
Figure 7.8 Upward force from hearth liquids
In operational practice poor burden descent is often an indicator of a poor blast

furnace process, especially difficulty in consistent melting of ferrous materials.
The reasons for poor burden descent can be:

– The upward force is too high. Experienced operators are well aware of the
maximum pressure difference over the burden that allows smooth operation.
If the maximum allowable pressure difference is exceeded (generally 1.6 to 2
bar), the process is pushed beyond its capabilities: burden descent will become
erratic, resulting in frequent hanging, slipping and chills.
– A hot furnace is also known to have poorer burden descent. This is because
the downward force decreases due to the smaller weight of burden above the
melting zone. In addition, there is more slag hold–up above the tuyeres, because
of the longer dripping distance to the hearth and the (primary) slag properties.
– Burden descent can be very sensitive to casthouse operation because of the

abovementioned upward force on the submerged coke. A high liquid level in
the hearth creates greater buoyancy forces.
7.3 Burden distribution and gas flow control
7.3.1 How does gas pass through ferrous layers?

The total ΔP in the furnace in Figure 7.4 is typically around 1.6 to 2.0 bar, of
which 0.3 to 0.5 bar is caused by the tuyeres (Figure 6.8), which function as a
restriction. So, the ΔP of the blast furnace process is 1.2 bar. Since there are
30 to 40 charges in the blast furnace, the ΔP over 1 charge is 0.03–0.04 bar
or 30–40 cm water gauge. There is a large difference in permeability between
the coke layers and the ore burden layers. The ore burden has a much higher
resistance to gas flow than coke. As a consequence, the horizontal coke layers
have the same pressure over the diameter, and it is the pressure difference of
0.03–0.04 bar between sequential coke layers that drives gas through the ferrous
layers (Figure 7.9).
Figure 7.9 The pressure difference between sequential coke layers drives gas through
the ferrous layer
The required pressure to drive the gas through the furnace is determined by the
permeability of the ore burden, which is determined by the amount of fines and
its melting properties. The amount of fines is determined by:
– The screening efficiency in the stock house.
– The physical degradation during transport and charging.
– The low temperature degradation properties of the burden. These effects
cause a ring of burden material with poor permeability in many operating
blast furnaces. This ring of material, in particular, is often difficult to reduce
and to melt down. Sometimes, unmolten ore burden materials are visible
as scabs through the peepsights of the tuyeres. The unmolten material can
cause operational upsets like chilling the furnace or tuyere failures. It is a
misunderstanding to think that these scabs consist of accretions that have fallen
from the wall.
124 Chapter VII
7.3.2 Objectives of burden distribution

In a blast furnace oxygen is removed from the ferrous burden by two different
mechanisms: reduction by gas (CO and H2) and “direct reduction”, which
results in carbon used for direct reduction. Both mechanisms have to take
place, as explained in Annex IV, but in operating blast furnaces, improving
gas reduction always leads to lower fuel rate, which is manifest from improved
gas efficiency or ηCO. Technically: in order to reduce wustite, FeO, by CO, the
ηCO on the wustite level has to be below 30 % in order for sufficient CO to be
present (Annex IV). In operational practice the value at very good operating
furnaces does not exceed 25 %. This means gas reduction does not reach its
thermodynamic limits.
So, the objective of burden distribution is to maximize gas reduction. To this

end, the reduction gas has to be distributed evenly through the ferrous layers.
The unfortunate aspect of blast furnace operations is that ferrous burden
impedes gas flow because of its higher resistance compared to coke. In order to
make the gas flow as even as possible, the following conditions have to be met:
1. The ferrous layer needs to have a uniform thickness over the radius, because a
thicker layer on a certain radial position will impede gas flow.
2. The ferrous fines have to be blended well through the ferrous burden, because
fines impede gas flow.
3. Various materials have to be blended well in order to avoid radial differences in
permeability. It may be clear that at a radial position with good permeability, for
instance with a large pellet percentage, gas will flow through this area, resulting
in very well reduced pellets, but poor reduction will then occur at the radial
positions that have less gas throughput.
The situation is made more complicated in a blast furnace because

1. Gas can easily flow along the wall.
2. Softening and melting materials are impermeable for gas, which means that in
the cohesive zone gas has to flow through the coke slits (or along the wall).
Gas flow along the wall is shown by heat losses through the cooling system of
stave temperatures in Figure 7.10. These are caused by hot gas in direct contact
with the staves. The general pattern is, that stave temperatures increase fast in
5–20 minutes and then decrease more slowly over several hours.
600
Temperature (°C)
400
200
Figure 7.10 Example of stave temperatures

Gas has to flow through the cohesive zone via the coke slits. The pressure
needed to drive gas through the coke slits depends on:
1. The amount of gas per unit time, minute for instance: at high productivity more
gas has to pass through the coke slits than at low productivity.
2. The number of coke slits and width of the coke slits: at high coal injection rates
the number and/or width of the coke slits becomes smaller.
7.3.3 Two basic types of cohesive zone

The efficiency of the furnace is determined by the amount of energy used in
the process. Heat losses to the wall and excess top gas temperature are examples
of energy losses. The top gas contains CO and H2, which have a high calorific
value, therefore, the efficiency of a blast furnace is determined by the progress of
the chemical reactions and thus by the gas flow through the furnace.
Two basic types of gas distribution can be distinguished: the “central working”
furnace and the “wall–working” furnace. The typology has been developed to
explain differences in operation. Intermediate patterns can also be observed.
In the “central working” furnace the gas flow is directed towards the center.
In this case the center of the furnace contains only coke and coarse burden
materials and is the most permeable area in the furnace. The cohesive zone takes
on an “inverted V shape”. In a “wall–working” furnace, the gas flow through the
center is impeded, e.g. by softening and melting burden material. The gas flows
preferentially through the zone with highest permeability, i.e. the wall zone.
In this case the cohesive zone takes the form of a “W shape”. Figure 7.11 shows
both types.
Both types of gas flow can be used to operate a blast furnace, but each has its
own drawbacks. Gas flow control is achieved by burden distribution.
Figure 7.11 Two types of melting zone: “inverted V shape” for a central working
furnace (left) and “W shape” for a wall–working furnace (right)
126 Chapter VII
7.3.4 Central working furnace

The two types of gas flow through a furnace can be achieved with the help of
the burden distribution. In Figure 7.12 the ore–to–coke ratio over the radius
is shown for a central working furnace. In the figure, the center of the furnace
contains only coke. So, in the center of the furnace no melting zone can be
formed, and the gas is distributed via the coke slits from the center towards the
outside radius of the furnace. The melting zone gets an inverted “V” or even “U”
shape. The central coke column not only serves as a gas distributor, but also as a
type of pressure relief valve: it functions to stabilize the blast pressure.
Figure 7.12 Central working furnace
The type of burden distribution equipment determines how the coke can be
brought to the center. With a bell–less top the most inward positions or lowest
angles of the rotating chute can be used. With a double bell system, the coke
has to be brought to the center by “coke push” (see below) and by choosing
the right ore layer thickness in order to prevent flooding of the center with ore
burden materials. In the central working furnace, there is a relatively small
amount of hot gas at the furnace wall: hence low heat losses. As a result, the
melting of the burden in the wall area takes place close to the tuyeres, so the
root of the melting zone is low in the furnace. The risk with this type of process
is that ore burden may not be melted completely before it passes the tuyeres.
This could lead to the observation of lumps of softened ore burden through the
tuyere peepsights. This can lead to anything from slight chilling of the furnace
(by increased direct reduction) and irregular hot metal quality up to severe
chills and damage to the tuyeres.
Limiting the risk of a low cohesive zone root can be done with gas and burden
distribution. Operational measures include the following.
– Maintain a sufficiently high coke percentage at the wall. Using nut coke in the
wall area can also do this. Note that an ore layer of 78 cm at the throat needs
about 28 cm of coke for carburization and direct reduction. So, if the coke
proportion at the wall is under 27 %, a continuous ore burden column can be
made at the wall.
– Ensure a minimum gas flow along the wall in the bosh and belly, which can be
monitored from heat loss measurements and/or temperature readings. If the gas
flow along the wall becomes too small, it can be increased by means of burden
distribution (more coke to the wall or less central gas flow) or by increasing the
gas volume per tonne of hot metal (by decreasing oxygen).
– Control the central gas flow. Note that the gas flow through the center leaves
the furnace at a high percentage of CO and H2 and a high temperature. The
energy content of the central gas is not efficiently used in the process, and so the
central gas flow should be kept within reasonable limits.
The central working furnace can give very good, stable process results with
respect to productivity, hot metal quality and reductant rate. It also leads to
long campaign length for the furnace above the tuyeres. However, the process
is very sensitive to deviations in burden materials, especially size distribution.
A big increase in coke rate due, for example, to problems with PCI preparation,
may need a change in coke distribution to avoid excessive central coke, which
can then starve the wall region of gas.
7.3.5 Wall–working furnace

Figure 7.13 shows the wall working furnace. If ore is charged into the center,
melting ore burden blocks the center of the furnace and the gas flow is directed
towards the wall area.
Figure 7.13 Wall–working furnace
The gas flow causes high heat losses in the area of the furnace where a gap
can be formed between burden and wall i.e. in lower and middle shaft. The
melting zone gets a “W” shape or even the shape of a horizontal disk. In this
situation the root of the melting zone is higher above the tuyeres, which makes
the process less sensitive to inconsistencies. The process can be rather efficient.
However, due to the high heat losses, the wear of the refractory or damage to
cooling staves or plates in the shaft, is much more pronounced than with the
central working furnace. The gas passing along the wall can also cool down
128 Chapter VII
rapidly and in doing so loses its reduction capabilities. As a consequence, the

fuel rate is high. Moreover, the fluctuations in the pressure difference over the
burden are more pronounced, and this leads to limitations in productivity.
7.3.6 “Ideal” burden distribution

The ideal burden distribution for high productivity and high PCI rates is,
according to the authors, as follows (Figure 7.14):
– An ore–free center,
– Nearly horizontal layers of coke and ore burden,
– Some nut coke in the ore burden in the wall area and
– Fines distributed over the radius.
Stockline
–1 m
–2 m
–3 m
Wall Centre
Figure 7.14 “Ideal” burden distribution (left) and operational practice as

measured with profile measurements of close to ideal burden
distribution (right)
Ore–free center
The ore free center allows the gas to distribute itself through the coke layers
from the center to the wall. We can consider the coke layers as layers with
equal pressure. If the total internal pressure difference is 1.2 bar, the pressure
difference over each of the 30–40 ore layers is about 0.03–0.04 bar. The ore–free
center typically has a diameter of 15–20 % of the diameter. The ore–free center
can be made in a furnace with a bell–less top by discharging 15–20 % of the
coke on a very inward chute position. The center coke can also be discharged as
a separate dump.
The coke charged in the center is attacked least by the solution–loss reaction
and has the smallest chance of being burnt in front of the tuyeres. For this
reason, it is thought that the coke charged in the center finally constitutes the
coke in the hearth. Good permeability of the hearth helps to improve casting
and prevents preferential flow of iron along the wall, thus increasing hearth
campaign life.
Nearly horizontal layers

Using nearly horizontal layers of coke and ore minimizes the effect of natural
deviations of parameters. For instance,
– sinter and pellets have different angles of repose.
– wet pellets have a larger angle of repose as compared with dry pellets.
– stockyard material has a smaller angle of repose as well as more fines than fresh
materials.
Figure 7.15 Segregation and angles of repose.
Figure 7.15 shows the angles of repose of the various materials used in a blast
furnace. Coke has the steepest angle of repose; pellets have the lowest angle of
repose and sinter is in between. Hence, in a pellet charged furnace the pellets
have a tendency to roll to the center. Fines concentrate at the point of impact
and the coarse particles flow “downhill”, while the fine particles remain below
the point of impact. The result is fine material at the center and coarse material
at the foot of the conical pile
Nut coke
The gas in the wall area is cooled by the heat losses to the wall. Moreover, a
relatively large percentage of fine ore burden material is located at the wall, and
reduction–disintegration is highest at the wall (because of slower heating and
reduction). For these reactions, reduction and melting of the ore burden in the
wall area is most difficult. Nut coke in the wall area helps to lower reduction gas
and heat requirements in the wall area. The nut coke has a lower heat capacity
than the ore burden. Moreover, when the ore burden in the wall area starts
melting, the nut coke is immediately available for direct reduction. In this way,
it prevents direct reduction attack on the metallurgical coke.
Fines distributed over the radius

During discharge of the ferrous materials, fine material is discharged at the
lower part of the falling curve (Figure 7.16 on the next page). As a consequence,
if several rounds of the chute are maintained at the same chute angle, the fines
are collected above each other in a ring, and this makes that specific ring less
permeable. Segregation is minimized by using a large number of chute angles
(more than 8 positions) and distributing the ferrous burden in horizontal layers
with a minimal discharge time (less than 12 rings).
130 Chapter VII
Coarse
Coarse
Fine
Fine
Figure 7.16 Segregation of fine material in a chute (left) and as shown in a laboratory
test with coke
Coarse and fine sinter

Some companies separate coarse and fine sinter and charge the materials
separately, for example as in Figure 7.16. The good aspect of the method is
that the furnace suffers less from mixed layers (see Section 7.4), since coarse
sinter does not penetrate well into the coke. The disadvantage is that there are
relatively large ferrous layers.
Fine sinter
Coke
Sinter
Coke
0 20 40 60 80 100
wall Distance from wall (%) center
Figure 7.16 Burden profile with coarse and fine sinter

Example 7.2 Example: from W shape to inverse V shape cohesive zone

An example of a furnace changing over from “W shape” cohesive zone to an
“inverse V shape” (central working) is shown in Figure 7.17. In an 1800 m³ IV
furnace, 19 % of the coke was charged into the center, with the chute starting
in the center for coke (moving from center to wall) and the ferrous burden
charged from outside inwards. Heat losses fell from 50 GJ/hr to 30 GJ/hr,
and the furnace became more stable and productive within 24 hours. The top
camera showed a relatively cold spot in the center before the change, which
disappeared after the change.
80 2800
Blast volume (m³ STP/hr)

Heat loss (GJ/hr)
60 2100
40 1400
20 700
0 0
–10 –5 0 5 10 15 20
Hours
before after
Temperature at burden surface
Figure 7.17 Effect of charging coke into the center. At 0 hrs, 19 % of coke was
charged into the center. Dotted line represents heat loss, drawn line
blast pressure. Details in text.
132 Chapter VII
7.4 Coke and ore layer thickness

The ratio between coke and ore layer thickness is dependent on the coke
amount charged per tonne in the top. For permeability: the thicker the coke
layer and the more permeable the coke (large coke), the better the permeability.
For reduction of the ferrous layer: the thinner the ferrous layer, the better the
contact between gas and the ferrous burden and the better the reduction.
Figure 7.18 Mixed layer formation
What determines the optimum ferrous layer thickness? First consider formation
of a “mixed layer”. The average diameter of coke (40 to 50 mm) is much larger
than that of pellets and sinter (typically under 15 mm and 25 mm respectively).
Burden components dumped on a coke layer will tend to form a “mixed layer”
(Figure 7.18). This mixed layer will have permeability comparable with the ore
layer. The larger the coke and ferrous layer thicknesses, the fewer mixed layers
are present in the furnace. However, very thick ferrous layers are difficult to
reduce (Figure 7.19). The thicker ore layers heat up more slowly and have poorer
gas reduction. When the layers are softening, a very thick cohesive zone is
formed, which melts from the surface. Thicker layers are also more difficult to
melt, so for permeability, thick coke layers are required, but for good reduction
and melting, thin ferrous layers are to be preferred.
100% Thickness 150% Thickness
Figure 7.19 Reduction of thin and thick ore layers

7.4.1 Ore layer thickness

Figure 7.20 shows a benchmark of the ferrous layer thickness used in operating
blast furnaces. The results indicate that the optimum layer thickness also
depends on quality and type of raw materials, screening in the stockhouse,
charging capacity, productivity and injection rate.
200
Ore base (ton/charge)
160
120
80
40 y = 18.58x – 128.33
R² = 0.95
0
7.5 10.0 12.5 15.0 17.5
Hearth diameter (m)
Figure 7.20 Ferrous layer thickness expressed as tonne ferrous/charge versus hearth
diameter. Red dots represent blast furnaces that perform excellently.
7.4.2 Coke layer thickness

Coke layers separate the ore layers. While the ore is melting, the ferrous
layers have to be prevented from making contact with each other, causing
the formation of scabs. Example 7.3 presents operational examples, where the
coke layers between the ferrous layers become too small. It can cause poor
permeability, erratic burden descent and/or tipping tuyeres.
In order to prevent contact between the ferrous layers, coke required for direct
reduction and carburization has to be present. This can be estimated as in
Figure 7.21. The amount of coke required to prevent ferrous layer contact
is about 25 % of the volume. The coke requirement at the wall can also be
met using nut coke blended into the ore layer. In this case, the nut coke
is preferentially available for direct reduction and will preserve the larger,
metallurgical coke in the layer structure.
Figure 7.21 Coke required for direct reduction including carburization

134 Chapter VII
Contact between ferrous layers will lead to disturbed gas flow. There is also a
risk that unmolten material will rest on the tuyere nose and cause the tuyere to
tip. This can be observed through the peepsight where an oval opening of the
tuyere is seen rather than a round one.
7.4.3 Minimum coke rate

When reaching higher and higher coal injection levels the question arises
whether a minimum coke layer thickness exists, and what would it be?
The gas ascending the furnace from the tuyeres to the top is distributed through
the coke layers, so the coke layers must be present at all elevations of the furnace
for this to continue. As the layers are made up of discrete coke particles, the
theoretical minimum coke layer thickness translates into a number of coke
particles. To produce a path for the gas it is believed that the minimum number
of coke particles that should be present in the height of one layer is three. The
minimum thickness is therefore three times the mean size of coke in the belly
of the blast furnace. Taking an average coke size of 50 mm, it would therefore
be reasonable to expect that the minimum coke layer thickness in the belly is
15 cm. As the effective ratio of the surfaces of belly to throat is generally around
2.25, the coke layer thickness at the throat should be at least about 34 cm.
In operational practice at furnaces operating at high coal injection levels, the

coke layer tickness at the throat has reached values as low as 32 cm and in the
belly 14 cm.
The minimum “metallurgical” coke rate (this is the coke rate excluding nut
coke), as a monthly average for operating blast furnaces, has reached levels of
210–220 kg/tHM in the best months and 230–240 kg/tHM as a longer term
average for well–operated blast furnaces.
7.4.4 Optimizing ore and coke layer thickness

The blast furnace operator wants good permeable coke layers (i.e. thick layers)
and good melting ore layers i.e. thin layers. As is often the case in blast furnace
operation the best operational results can only be reached with a compromise
between these two factors. Generally speaking, from operational observation,
the ore layers should not exceed 70–80 cm in the throat of a blast furnace, and
coke layers should not be smaller than 34 cm. The operational optimization
depends on local situations.
Experience has shown that:

– Permeable ore layers can be maintained even when the layers have become quite
thick, provided a permeable ore burden is used. For pellet burdens this would
require screening of the pellets, and for sinter it would have to be sized to a
relatively large diameter (more than 5 mm).
– The minimum coke layer thickness experienced was 14 cm metallurgical coke in
the belly.
Conveyor belt fed furnaces tend to work with thicker ore layers. This is caused
by the fact that in a conveyor–fed furnace the charging capacity increases with
increasing layer thickness. In skip–fed furnaces the optimum charging capacity
is reached with full skips of coke. In the past the volume of coke was normally
the determining factor, so furnaces tended to work with full skips of coke. At
high coal injection rates, the skip weight is normally the determining factor and
thus furnaces now work with full skips of ore.
Another aspect of the optimization of the coke layer thickness has to do with
the gas permeability of the coke layer. The coarser the coke is screened in the
blast furnace stockhouse, the more permeable the layer is. There are, however,
two drawbacks to the coarse (35 mm or more) screening of coke.
Consequence 1: The coarser the coke is screened, the more nut coke or small
coke is produced. The nut coke is added to the ore burden layer, increasing the
its thickness and decreasing that of the coke layer.
Consequence 2: The coarser the coke is screened at the stockhouse, the thicker
the formation of a mixed layer at the coke–burden interface.
7.5 Finding an optimal burden distribution
It can take many years to find good burden distribution for a furnace. But if
found, it can be quite stable for many years with slight and controlled steps in
burden distribution.
Companies with proper instrumentation for analysis of burden distribution,

like a profilometer and burden distribution model, have generally succeeded in
achieving efficient operation.
Blast furnaces operated without these tools have greater difficulty in optimizing
burden distribution. In operational practice the following misunderstandings
have been observed by the authors.
1. In the event of high heat loads through the wall, operators sometimes tend
to charge more ferrous at the wall. This can help but more often the furnace
receives a coke percentage that is too low at the wall, thus making the situation
worse. Note that gas in a blast furnace always takes the path of least resistance,
so the solution for high heat loads is adjusting the resistance pattern in the
furnace, for instance by charging central coke.
2. Sometimes the operator is forced to aim for high productivity. Excessive
amounts of coke in the center (> 25 %) and high oxygen enrichment with a low
top temperature (below 80–90 °C) are used. Note that this is an average – it
will be even lower over the outer radius, indicating insufficient gas flow. Furnace
efficiency declines (ηCO < 44%). More details in Section 11.1.2.
3. Operators are trying to improve permeability and lower the heat loads to the
wall by increasing the central gas flow. The method of choice is to move the
136 Chapter VII
distribution of coke and ferrous to the furnace center. This can lead to an
“inverse profile” (i.e. a profile where the central area is higher than the wall).
To our knowledge, the inverse profile has not been applied successfully and
should therefore be avoided at all times. This method is an extrapolation of the
burden distribution control used in the past with double bell equipped furnaces.
Finding that central gas flow could be increased by extending a platform at
the wall and making the V–shaped center smaller was an extremely fruitful
research outcome. The method is sometimes referred to as “platform–V” burden
distribution. It was extremely useful for double bell furnaces, but not applicable
or obsolete for furnaces with a bell–less top.
7.6 Double bell furnaces
Furnaces operating with a double bell top have less flexibility in burden
distribution. The following points are of importance for these furnaces.
– As shown in Figure 7.22, material charged from a double bell segregates
into fine and coarse material, forming rings of fine materials that are poorly
permeable for gas. These materials melt relatively poorly and can become visible
as scabs through the peepsights. In the event that movable armor is available,
the radial location of the ring of fines can be varied, which improves the gas
reduction and furnace efficiency and decreases the frequency of scabs at the
tuyeres.
– The central coke in the furnace comes from the “coke push” effect (Figure 7.22).
Coarse coke on the surface of the coke layer is pushed towards the center upon
charging of the next ferrous layer. The central gas flow can be controlled by
controlling the amount of ferrous burden that reaches the center. This can be
done using three different methods.
a. Creating and extending a platform at the wall with movable armor positions.
On an extended platform more ferrous is held at the wall and less ferrous
reaches the center.
b. By varying the layer thickness (weight) of the ferrous burden. If the batch
weight of the ferrous is decreased, for every charge the same coke amount is
still pushed towards the center, but less burden reaches the center.
c. Coke charged first on the double bell before ferrous will be pushed towards
the center.
Operators have developed additional methods to influence gas flow for a

double–bell blast furnace. These methods have to do with the charging
sequence of coke and ferrous, the stockline level and the opening speed of the
double bell. There is often considerable experience with these types of methods
on site, but application is very furnace–dependent.
Figure 7.22 Burden charging with double bell top
7.7 Permeability of cohesive zone
Materials soften and melt in the temperature range of 1200–1400 °C as shown

in Figure 7.23a. Softening materials are impermeable to gas: in Figure 7.23b
shown by a sample of burden consisting of 100 % pellets taken from a blast
furnace after blowdown, and in Figure 7.23c by a sinter sample during a melting
test. If nut coke is added to the sinter, the nut coke remains intact and can
create paths for gas to pass (Figure 7.23d). For this reason, the cohesive zone
becomes more permeable with nut coke, which makes the cohesive zone melt
ore easily. The cohesive material breaks into smaller pieces when descending
into voidages, and helps to promote smoother burden descent, especially if
charged in the wall area.
Since scrap has a high melting temperature, it also remains solid in the
temperature range of the cohesive zone, so scrap has a similar effect to nut coke
on the permeability and melting of the cohesive zone.
Figure 7.23 A: cohesive zone; B: softened pellets sampled from a furnace; C: sinter
during melting test; D: same as C with nut coke added
(C and D from Quinshi Song, 2013)
138 Chapter VII
Example 7.3 Operational challenges and burden distribution solutions

Radial nut coke position, 4600 m³ blast furnace
In an operating blast furnace burden descent was poor (Figure 7.24, left).
This improved greatly after bringing nut coke from the intermediate to the
wall area (Figure 7.24, right). Explanation: operator moved ferrous outwards
to control high heat loads to the wall, but exceeded the limit where ferrous
burden at the wall started to become vertically connected.
17 January nut coke in wall area 24 January

6 hrs stockrods
Figure 7.24 Stockrods, 6 hour graphs, before and after moving nut coke to wall
Inverse profile, 2800 m³ furnace

Because of tuyere tipping, the operator wanted to charge more coke at the
wall, which was executed by moving pellets (100 % pellets furnace) inwards.
This caused an inverse profile with a very large pellet percentage at the wall
because of the shallow angle of repose of pellets (Figure 7.25). The result was
more severe tipping. Solved by moving coke and pellets towards the wall.
Figure 7.25 Inverse profile at 100 % pellet furnace

Inverse profile, 5000 m³ furnace

In order to improve central gas flow and decrease heat loads to the wall, coke
and ferrous layer were shifted inwards. The burden profile as calculated by the
plant research showed an inverse profile, see Figure 7.26.
At about 2.2 m from the center, the coke percentage was very low. On
melting, a vertical column of cohesive material was formed without coke slits.
This caused poor permeability. After a horizontal profile was implemented,
the furnace had lower heat loads and was accepting higher pellet percentage.
5.6 2.8 0 2.8 5.6
Figure 7.26 Inverse profile with very low coke percentage 2 meters from the center,
resulting in cylinders of melting ferrous material in and below the
cohesive zone
VIII Hot Metal and Slag
How is the ferrous burden transformed to hot metal and slag? In the preceding
chapter gas flow through the furnace was discussed. The present chapter shows
how the gas affects burden, first in the “dry” area above the cohesive zone,
then softening, melting and finally how hot metal and slag reach the final
composition and properties.
8.1 Processing the ferrous burden above

the cohesive zone
Ferrous burden is the collective term used to describe the iron–containing

materials that are charged to the furnace, namely, sinter, pellets and lump ore.
The melting properties of these materials depend on the local chemical slag
composition. Lump ore has its natural slag composition as it is found in the
earth, gangue consists mainly of acid components like SiO2 and Al2O3. Pellets
and sinter have an artificial composition with components added to the natural
iron ores, such as limestone (CaCO3), dolomite (MgCO3.CaCO3), olivine
(2MgO.SiO2) and others. Sinter basicity CaO/SiO2 is typically above 1.6 and
can even reach levels of 2.8 or higher. Pellets have a wide variety of chemical
compositions, most notably acid pellets (CaO/SiO2 < 0.2) or fluxed pellets
(CaO/SiO2 > 0.8).
The chemical composition of the materials is not only based on the design of
the optimum properties of the final slag, with respect to melting temperature,
fluidity and desulfurizing properties, but also on the design of the metallurgical
properties of the sinter and the pellets. Optimal metallurgical properties mean
that the materials should have good reduction–disintegration properties and
melting temperatures as high as possible. The reason for these requirements is
defined by the nature of the blast furnace process, this being a gas–reduction
process. If the material breaks down into small particles, the gas flow through
the ore layer is impeded and the normal reduction process is limited. In
addition, materials which start to melt form an impermeable layer and will also
affect the reduction progress.
Note that the efficiency of a blast furnace is largely determined by the gas
reduction process, along with the amount of oxygen bound on the iron, which
is removed by gas (CO and H2).
142 Chapter VIII
8.1.1 Reduction from hematite to magnetite and

reduction–disintegration
The reduction process starts at temperatures of about 500 °C in the atmosphere
of a reducing gas, that is, the blast furnace top gas. The reduction of hematite
(Fe2O3, O/Fe = 1.5) to magnetite (Fe3O4, O/Fe = 1.33) takes place rather easily
and generates a small amount of heat. In hematite, 6 atoms of iron are bound
to 9 atoms of oxygen, which changes to 8 atoms of oxygen upon transition to
magnetite. The extra oxygen is bound to the CO gas, forming CO2.
The first step in the reduction process has a profound effect on the properties
of the ferrous burden. The crystal structure where 6 iron atoms and 9 oxygen
atoms were happily conjoined is forced to change to 6 iron atoms on 8 oxygen
atoms. The crystal structure changes with an associated volume increase of
about 10 % and this leads to stress within the particles, which can fall apart.
This is called reduction disintegration, and is represented by the Reduction
Disintegration Index (RDI) or, sometimes by Low Temperature Breakdown
(LTB, Chapter III). Note that the reduction disintegration generates a huge
amount of fines. After screening, sinter has 2–4 % of fines (smaller than 5 mm);
reduction disintegration generates 20–30 % fine sinter and lump (< 3.15 mm)
(Chapter III, Table 7.1).
Pellets are not very prone to reduction disintegration, as pellets have about 30 %
voidage in the structure, which can take care of local expansion. As they retain
their round shape, they do not impede the local permeability for gas.
Some lump ores have a very tight structure and are difficult to reduce, with
the reduction starting on the outside of the particle. These lump ores will have
reasonable RDI values. However, if a lump ore has a relatively open structure,
which is easily permeable for gas, then the RDI will be poor. Lump ores with
this characteristic are less suitable for direct use in the blast furnace for more
than a few percent.
Sinter, on a micro–scale has a relatively tight structure with limited possibilities

for local expansion. This means that sinter has inherently poor RDI unless
measures are taken to improve it. The RDI can be improved by impeding the
formation of the secondary hematite on the sinter strand. Secondary hematite
is the material that is re–oxidized on the sinter strand, from magnetite back
to hematite. This takes place when sinter is cooled with air. It is the secondary
hematite that is very prone to reduction disintegration in the blast furnace. The
reduction disintegration stops when all the hematite is reduced to magnetite.
8.1.2 Gas reduction of magnetite

The magnetite (Fe3O4) is further reduced by gas (CO and H2) to wustite
(FeO1.05). At around 900 °C ,equilibrium is reached between the reducing power
of the gas and the composition of the iron oxides, i.e. the FeO level of one atom
of oxygen per atom of iron. In this area the temperature is relatively constant
Hot Metal and Slag 143
(thermal reserve zone), as is the chemical composition of the gas (chemical

reserve zone). When blast furnaces are operated at very high productivities the
reserve zone becomes smaller and is ultimately eliminated.
At temperatures around 900 °C the temperature of the coke is still too low to
react with the CO2 gas. The coke reactivity reaction (CO2 + C 2 CO) starts
at around 1000 °C. This means that all the reduction is taking place by means
of gas reduction (Fe2O3 + CO 2 FeO + CO2), and in this temperature range
also for a small part by hydrogen reduction (Fe2O3 +H2 2 FeO + H2O). The
gas reduction continues to a gas temperature above 1000 °C and a reduction
of iron oxide to a level of FeO0.5. The higher the temperature, the more H2
contributes to the gas reduction. The gas reduction continues, but coke starts to
react with CO2 and H2O – which is chemically identical to the direct reduction
reaction. If material starts to soften and melt (around 1200 °C) the direct
reduction reaction (FeO + C Fe + CO) will take place. The direct reduction
starts more or less at a ratio of O/Fe slightly below 0.5 atom O per atom of Fe.
In the case of natural gas injection, this O/Fe for direct reduction can decrease
to around 0.35 atom O per atom Fe.
8.2 Softening
Softening starts at local chemical compositions with the lowest softening

temperatures. This is where there are high local concentrations of SiO2 and FeO.
Internal migration of atoms will cause larger and larger parts of the particles to
soften. Softening starts as soon as internal ‘melts’ of material are formed. The
softening temperature depends on burden type and composition. The details are
in Section 10.9. In the book a softening temperature of 1200 °C is used. At this
temperature, the O/Fe ratio in the ferrous burden is 0.35–0.5, which means that
the FeO content in the “primary slag” is 60–70 %. The softened materials are
called “primary slag”. Its composition is the gangue of the ferrous burden with
large quantities of wustite (FeO); slag component from coke, coal and nut coke
are not taken into account.
In the case of fluxed pellets, softening starts at approximately 1200 °C. Acid
pellets soften at a temperature about 100 °C. Fluxed sinter starts to soften at
temperatures above 1200 °C, which increases to 1300 °C for super–fluxed
sinter. Lump ore softening depends on quality. It may start below 1000 °C and
can reach considerably higher softening temperatures. It is very acceptable for
high performance, especially dictated by the amount of acid slag components
present.
As soon as the gangue starts to melt, it will come into contact with the slag
components of other parts of the ore burden and the slag composition will be
averaged. This happens at high FeO concentrations. Figure 8.1 on the next page
shows the liquidus temperatures of “primary slag” as dependent on B2 and FeO
content. Slag melting starts at low temperatures and high FeO levels, but the
144 Chapter VIII
basicity of the burden affects melting. As the FeO is reduced out of the primary
melt, the slag liquidus temperature increases. The primary slag of the reference
furnace has a basicity when melting of 1.29, which is the gangue of all ferrous
components plus the coke ash of the coke consumed by direct reduction.
The figure shows that the liquidus temperature depends very much on slag FeO
percentage as well as basicity. The primary slag liquidus temperature increases
by around 50 °C for every 0.1 basicity increase.
In this respect it is important to understand, that the ash percentage of injection

coals plays an important role, since coal ash contains approximately 50 % SiO2.
With high ash coals there is a need for a higher primary basicity. One of the
advantages of using natural gas as an injectant is the lower primary slag basicity.
1700
B2 = 1.6
B2 = 1.5
B2 = 1.4
1600 B2 = 1.3
Liquidus Temperature (°C)
B2 = 1.2
B2 = 1.1
1500
1400
1300
1200
40 30 20 10 0
FeO content (%)
Figure 8.1 Slag liquidus temperatures of bosh slag and the effects of FeO content and
basicity. The dark red line (B2 = 1.3) is the reference furnace. “MT data”
software was used for calculations.
Softening is the process of collapse of the burden and depends on local

chemical composition and the reduction progress. So, in order to keep the
variation within the layers as small as possible, good blending of components is
recommended. It is shown elsewhere (see Section 3.8) that blending improves
metallurgical and softening and melting properties. Also, the use of nut coke in
a burden layer results in better permeability, since nut coke does not soften with
the burden and maintains a path for gas through the burden layer. Charging
scrap or DRI in the ferrous burden also maintains the permeability of softened
material up to higher temperatures.
In summary, softening starts in material with acid slag components mixed with
iron oxides, which start to form melts, which in turn cause the ore bed collapse.
The order of events is: First. the lump ore structure collapses due to the acidic
gangue. Next the acid pellet structure collapses. This is followed by the collapse
of the fluxed pellet and sinter structure. As soon as the layers have collapsed,
the permeability for the gas decreases. It is estimated that permeability for gas
disappears more or less completely between 1200 and 1350 °C. In this situation,
the layers of cohesive material are heated only by gas flowing along its surface.
Reduction by hydrogen plays a special role in this situation. Since hydrogen can
easily diffuse into a more compact structure, the hydrogen reduction continues
after CO reduction has stopped.
8.3 Melting
Material is completely molten if the temperature is higher than the melting or

liquidus temperature. Below that temperature solids are present, and it has a
high viscosity and does not flow properly.
In a blast furnace we have the hot metal and slag components. The liquidus
temperature depends on chemical composition especially on the FeO content
and basicity.
In Figure 8.2, the melting temperatures of hot metal is given. Comparison with
slag liquidus temperatures of Figure 8.1 shows that primary slag with a high
FeO content melts at a lower temperature than hot metal with low carbon. That
is the reason that primary slag melts in the first instance at temperatures of
1300 °C. Reduced iron does not contain carbon, which is dissolved as soon as
FeO content in the direct environment is low (below 10 %).
1600
Melting temperature (°C)
1500
1400
1300
1200
1100
1000
0 1 2 3 4 5 6
Carbon content (%)
Figure 8.2 Melting temperatures of iron
For this reason, reduced iron, for example a sponge iron skull of a pellet, has
a much higher melting temperature than hot metal. The sponge iron does not
yet contain carbon, and its melting temperature comes closer to the 1535 °C of
the elemental iron rather than the 1147 °C of hot metal with a 4.2 % carbon
content. The reduced iron does not melt, but remains as solid particles in the
melting slag–FeO mix.
When softened materials are heated further and start to drip, the “primary
melt” consists of a blend of the gangue, FeO and finely dispersed iron, which
has not been separated from the melt. The first process in the ‘primary’ melt
146 Chapter VIII
is that the gangue loses its FeO. The reduction of FeO is by direct reduction,
which requires heat that has to be supplied by the bosh gas. So, the contact
between bosh gas and primary melt determines the rate of melting. 80–90% of
the heat present in the bosh gas at a temperature above 1400 °C is required to
drive the direct reduction reaction. The direct reduction consumes coke. This
makes the acid coke ash available, which lowers gradually the basicity of the
primary slag.
As soon as the FeO is removed and the primary melt flows over the coke, the
iron starts to dissolve carbon from the coke, which rapidly lowers the melting
temperature of the iron. This has the effect of making the iron much more
liquid when flowing over the coke. The carbon of the coke diffuses into or is
taken up by the metallic Fe, allowing the iron droplets to separate from the
primary melt. After this process has taken place, the iron starts to increase in
silicon content, which comes from the SiO gas that was created in the raceway
flame.
As long as the slag contains FeO, the silicon in the hot metal will be oxidized
back to SiO2 and the FeO in the slag reduced to Fe. At the wall of the furnace,
the root of the cohesive zone is located a small distance above the tuyeres,
while in the center the cohesive some is located higher in the furnace. As a
consequence, the slag formed at the wall will have relatively high FeO and the
hot metal formed at the wall will have low silicon. While the hot metal formed
and dripping down in the center of the furnace will have high silicon. The
final silicon level observed during a cast is a blend of these two ‘hot’ and ‘cold’
components.
The formation of the final composition of hot metal and slag is a stepwise
process, which is summarized in Figure 8.3.
Figure 8.3 Melting of Iron Ore
Liquid hot metal and slag drain down through the coke. The available voidage
of the cokes has to hold the slag and may be the rate limiting factor for the
production level. Figure 8.4 shows a comparison between several furnaces
regarding the amount of liquids that are processed per m² of hearth surface.
Very few furnaces exceed a liquid hot metal and slag production level of about
22 m³ liquids/m² hearth.24 hrs. The limitations are influenced not only by the
amount of liquids, but also by the area for the liquids to flow through (furnace
dimensions as well as coke voidage) and the amount of gas that ascends through
the dripping liquids and coke voidage. The available coke voidage comes from
the coarse coke only and is determined by coke quality as well as the amount
of coke present per tonne of hot metal. The latter decreases with increasing PCI
rate. If the slag volume per tonne of hot metal increases, the required voidage is
larger and as a consequence the PCI rate or productivity has to decrease.
500
400
Slag rate (kg/tHM)
300
200
22 m³/m² hearth.24 hrs
20 m³/m² hearth.24 hrs
100
50 55 60 65 70 75 80 85 90
Productivity (tHM/m² hearth.24 hrs)
Figure 8.4 Liquid (sum of hot metal and slag volume) and production levels of
operating furnaces with a hearth diameter over 10 m are indicative of
limiting flooding conditions
8.4 Formation of hot metal and slag as cast
8.4.1 Radial asymmetry of hot metal and slag

The hot metal and slag as dripping down from the cohesive zone have been
formed from local conditions. In the center of the furnace hot metal with high
silicon is formed, at the wall low silicon. The elimination of FeO from the
primary slag is faster in the center than at wall. In the raceway coke and coal
are gasified generating acid coal and coke ash, which condenses on coke close
to the tuyeres. Part of the SiO2 in the coal and coke ash is reduced to generate
silicon monoxide gas, that is then further reduced to form atomic silicon which
dissolves in the hot metal. Basicity is high in the center and lower in the wall
area, because there is more SiO2 available in the wall area. The final hot metal
and slag compositions are formed in the taphole by blending liquids from the
different regions of the hearth.
148 Chapter VIII
8.4.2 Hot metal–slag interactions

The final hot metal and slag compositions are the result of a complex process of
iron–slag interactions as the various elements are divided over the slag and iron
phases. The dispersion of an element over the two phases depends on the slag
and hot metal composition as well as temperature, as discussed below. As an
illustration the typical percentages of elements entering the slag and iron phases
are indicated in Table 8.1.
The following points should be noted:

– Silicon, titanium and sulfur are concentrated in the slag.
– Manganese is concentrated in the hot metal.
– Most alkali (potassium and sodium) are eliminated with the slag, some of the
alkali is discharged as vapor with the top gas.
– Nearly all the phosphorus goes to the hot metal.
Input Output to Iron Output to Slag

Element kg/tHM kg/tHM % kg/tHM %
Silicon 46 5 11 41 89
Manganese 6 3.6 80 2.4 20
Titanium 3 0.7 23 2.3 77
Sulfur 3 0.3 10 2.7 90
Phosphorus 0.7 0.7 98 0 0
Potassium 0.15 0 0 0.11 73
Table 8.1 Typical distributions of selected elements over iron and slag
Hot metal temperature and hot metal silicon content are both used as an
indicator of the thermal state of the furnace. The hot metal temperature is
the result of the heat input in the furnace minus the heat used in the lower
part of the furnace. The major heat consumption in this area comes from the
(endothermic) chemical reactions taking place below the cohesive zone. These
are the direct reduction reactions of iron, silicon, manganese, titanium and
phosphorus oxides. While the hot metal silicon shows the result of the actual
chemical balance in the furnace, the hot metal temperature is somewhat delayed
compared to the hot metal silicon, since there is a large heat buffer in the
hearth.
Silicon, manganese, titanium and phosphorus oxides are reduced via the direct
reduction reaction. These reactions consume much heat. Most of the heat
is consumed for the SiO2 reduction and consequently, the hot metal silicon
content reflects the thermal state of the furnace. The manganese distribution
over hot metal and slag, [Mn]/(MnO), is an even faster indicator of the thermal
state because of the smaller MnO content of slag. For operational purposes, the
silicon reactions are of particular interest. The hot metal silicon is a sensitive
indicator of the thermal state of the furnace, and the silicon variation can be
used to analyze the consistency of the process.
8.4.3 Hot metal silicon

The reaction of silicon reduction in a blast furnace are now discussed in more
detail. Reduction of silicon from silica by carbon is possible at a temperature
above 1400 °C:
SiO2 + 2 C Si + 2 CO ΔH= + 637 MJ/kmol
Silicon transition into hot metal is promoted by reaction of iron silicide

formation:
SiO2 + 2 C + Fe FeSi + 2 CO ΔH= + 542 MJ/kmol
Silica reaches the high temperatures required for these reactions when coke and
coal are burned. Coal and coke ash contain typically 45–50 % SiO2. SiO2 is also
part of the melting burden materials which contain gangue. Silicon reduction in
blast furnace takes place via two stages (Figure 8.5).
Figure 8.5 Silicon reduction
At stage 1 in the combustion zone, volatile silicon monoxide gas is formed from
the silica of the coke ash and injected pulverized coal ash.
SiO2 + C SiO + CO
SiO formation in raceway zone increases:

– when ash content in coke and PCI increases, and/or SiO2 content in the ash
increases,
– when the flame temperature (RAFT) increases,
– when the residence time of coke in the high–temperature zone increases, which
means that smelting rate and the furnace production output are decreasing.
– silica reduction is accompanied by generation of CO gas. So, an increase in the
pressure in the furnace impedes silica reduction and its formation to SiO gas.
150 Chapter VIII
At stage 2, the silicon is reduced from SiO by coke in the area of the raceway
and cohesive zone. It is dissolved in hot metal droplets. The following reactions
take place:
SiO + [C] [Si] + CO ΔH= – 10 MJ/kmol
SiO + [C] + [Fe] [FeSi] + CO ΔH= – 105 MJ/kmol
Research has shown that reduction of silicon takes place in the high–
temperature area above the raceway. When hot metal is sampled through the
tuyeres the hot metal has a higher silicon content than when sampled from
the cast. The higher the level of the cohesive zone, the larger the volume of the
high–temperature zone and the more time there is for contact of volatile SiO
with carbon of coke and hot metal. This leads to an increase in silicon reduction
and consequently higher hot metal silicon content and higher hot metal
temperature.
The contact time of hot metal with gas and coke in the coke layer between
tuyere level and cohesive zone is dependent on the structure of the coke bed and
the size of coke as well. The more coke fines, the smaller the fraction of middle–
size lumps of coke and the better the adhesive behavior for molten slag, and the
more the porosity of the coke layer decreases. This increases the time for the hot
metal and slag to descend through it, and there is thus more intensive contact
between the (SiO–containing) gas, which results in additional Si reduction. As
discussed above, silicon is partially reduced by coke carbon from slag. Silicon
reduction from slag is impeded by an increase in basicity, which decreases the
rate of Si reduction from calcium silicates.
In summary: the Si content in hot metal can be decreased by the following

measures:
– lowering the cohesive zone inside the furnace, which means running the
furnace at a lower thermal level.
– using a lower flame temperature.
– increase of coke CSR and average size of coke lumps charged into furnace.
– reduction of ash content in coke and injection coal.
– increase of top gas pressure.
– increase of the percentage of CaO and MgO in slag.
– increase of oxidation potential in the hearth (see below).
– increase of the production levels (melting rate) with increasing blast volume.
One aspect not yet mentioned is that in the hearth hot metal comes into contact
with slag and remains in equilibrium with it. When hot metal droplets pass
through the slag layer in the hearth, Si may be re–oxidized by ferrous oxide, if it
is present in slag:
[Si] + 2 (FeO) + 2 [C] (SiO2) + 2 [Fe] + 2 CO

Example 8.1 Properties of high silicon hot metal

During blow–in of a blast furnace or during a chill recovery a high coke
rate and low basicity burden are often used. This can lead to high hot metal
silicon levels, for instance 4 % or more. This type of hot metal can be very
viscous and can generate scabs in the runner system. When the hot metal
cools down graphite can be generated in the hot metal flow. The reason for
the high viscosity is that solid particles can form in the hot metal.
A relatively new insight is that this phenomenon of poorly flowing hot metal
takes place not only in the runner system, but can also be observed when
a blast furnace is operated for a prolonged period on very high coke rates
and high silicon. The phenomenon is illustrated in Figure 8.6, where the
measured hot metal silicon and carbon during a furnace recovery are shown
in a silicon–carbon phase diagram, where it is assumed that hot metal can be
oversaturated with carbon. SiC can form at a silicon content above 4 % and
Fe8Si2C at temperatures below 1300 °C. During a recovery it is important to
go towards silicon levels below 2 % fast. If such segregation takes place below
the tuyeres, unexplained tuyere failures (on bottom) can be observed, because
hot metal does not drain to the hearth.
Figure 8.6 Fe–Si–C phase diagram of hot metal oversaturated with C (Mn = 1 %).
Hot metal composition during a chill recovery indicated with red dots.
FeO is observed in slag when the furnace is very cold. Slag is very heavy and
granulated slag is black. In these situations, the furnace has a very low silicon
level. From a chemical point of view this is called increase of oxidation potential
in the hearth. This mechanism also explains why the hot metal sampled at the
tuyeres has a higher silicon% than hot metal coming from the taphole.
Finally, although we observe a uniform hot metal silicon level coming from a
taphole, in reality the hot metal is a blend of hot metal generated in the central
part of the furnace with a higher Si content and hot metal generated in the
peripheral zone with a low hot metal Si content.
152 Chapter VIII
Hot metal silicon varies from cast to cast and during a cast as well. Changes
in energy input and output become manifest in the hot metal silicon. Very
stable hot metal silicon is the result of stable operating conditions, the most
important being stable burden descent. If the burden descent is not smooth, the
gas reduction in the furnace varies, resulting in variation in direct reduction
in the high temperature area below the cohesive zone. In this situation the re–
oxidation of silicon will vary more.
For operators it is well known that the hot metal silicon increases when the
blast volume is decreased. From the previous section it is clear that when wind
is reduced the following effects will take place: lower driving rate of the furnace
allowing the hot metal more time for contact with (SiO–containing) gas,
improved gas reduction by better contact between burden and gas, sometimes
improved burden descent and finally a lower top pressure. All these effects result
in higher hot metal silicon at lower blast volume.
8.4.4 Hot metal Sulfur

The major sources of sulfur input are the reductants, i.e. coke and PCI, which
introduce 80–90 % of sulfur to the furnace, while only 10–20 % comes with
the burden materials. Most sulfur in the fuel (60–80 %) is organic, whereas the
sulfur in burden materials is present in the form of sulfides and sulfates. Sulfur
input into the furnace for modern blast furnaces is 2.5–3.5 kg/tHM.
As soon as the coke temperature exceeds 1000 °C, part of the sulfur volatilizes.
During combustion of coke and injected fuel, all sulfur is oxidized in the
form of gaseous SO2 and SO. In the more reducing atmosphere, sulfur oxides
react to free sulfur (S, S2) and its compounds CS, CS2, COS, H2S, HS. Sulfur
reacts when ascending through the furnace with burden components, forming
calcium, iron and manganese sulfides. These sulfides descend to the hearth
again. Calcium, magnesium and manganese sulfides dissolve in slag, while iron
sulfide dissolves in hot metal.
More than half of the input sulfur is transferred to slag by the following
reaction:
[FeS] + (CaO) + C (CaS) + [Fe] + CO ΔH= + 143 MJ/kmol (8.1)
Where ( ) means concentration in slag and [ ] in hot metal
The distribution of sulfur between slag and hot metal (LS) is estimated as
relation LS = (S)/[S] and depends on the slag basicity and on the temperature of
hot metal and slag. Sulfur quantity transiting to slag also depends on the slag
volume per tonne. The sulfur content in gas, solid and liquid phases over the
height of the furnace are shown in Figure 8.7.
The sulfur brought into the furnace with fuel, is first brought to the gas
phase before being eliminated from the furnace. As a result at each moment a
significant quantity of sulfur is retained in the blast furnace, estimated as 4–8
tonnes sulfur for 3000–5000 m³ furnaces (at a coke sulfur content of 0.6 %).
Figure 8.7 Sulfur content of gas, solid and liquid phases over the blast furnace height
Desulfurization of hot metal takes also place in the taphole when hot metal
and slag are cast simultaneously. In the taphole, pressure decreases from a high
value inside (around 5 bar) to atmospheric pressure outside the blast furnace,
which makes the equilibrium (8.1) shift to the right, to lower hot metal sulfur.
This is illustrated by an example of the hot metal sulfur in consecutive ladles,
showing about 20 % lower levels as a consequence of slag and hot metal being
cast simultaneously (Figure 8.8).
0.06
0.05
Hot metal sulfur (%)
0.04
0.03
0.02
0.01
0
1st 2nd 3rd 4th
Ladle in cast
Figure 8.8 Sulfur content in consecutive ladles of a cast from a 2600 m³ IV furnace
154 Chapter VIII
Example 8.2 Silicon, sulfur and basicity correlations in operation

Hot metal silicon and temperature are correlated as indicators of the thermal
state of the hearth. In the graphs below, these correlations are illustrated for
an operating blast furnace with 10 months of cast by cast data. The data are
“cleaned”: gap times shorter than 30 minutes, HMT 1450–1530 °C.
Trend comparisons can be made by making correlations over long periods or

by considering relative changes, for example the change in hot metal silicon
corresponding with a change in hot metal temperature of the preceding casts.
Correlation of hot metal temperature and silicon is shown in Figure 8.9. Hot
metal sulfur decreases with increasing silicon as shown in Figure 8.10.
1.0
y = 0.0037x – 4.9914
0.4
0.8 0.3
Hot metal Silicon (%)
Change in HM Si (%)
0.2
0.6 0.1
–60 –40 –20 20 40 60

0.4 –0.1
–0.2
0.2 –0.3
–0.4
0
1440 1460 1480 1500 1520 1540
Hot metal temperature (°C) Change in HMT compared to preceding cast (°C)
Figure 8.9 Correlation of hot metal silicon and temperature. The right hand graph
shows the difference in hot metal silicon and temperature compared to
the preceding cast.
Change in hot metal Sulphur (%)
0.015
0.010
0.005
–0.4 –0.2 0.2 0.4

–0.005
–0.010
–0.015
Change in hot metal silicon compared to preceding cast (%)
Figure 8.10 Change in hot metal S and Si
For sulfur the operator can consider the ratio (S)/[S], the sulfur in slag
divided by the sulfur in hot metal. This ratio is independent of slag volume.
Correlations with 2– and 3–component basicities shown in Figure 8.11.
100 90
y = 80.006x – 44.171 y = 74.989x – 57.38
80
80 70
60
60
(S)/[S]
(S)/[S]
50
40
40
30
20 20
10
0 0
0.8 0.9 1 1.1 1.2 1.3 1.0 1.1 1.2 1.3 1.4 1.5 1.6
Basicity CaO/SiO2 Basicity (CaO+MgO)/SiO2
Figure 8.11 (S)/[S] correlation with basicity B2 (left) and B3 (right).

8.5 Slag properties
8.5.1 Slag composition and basicity

Slag is formed from the gangue material of the burden and the ash of the coke
and auxiliary reductants. During the process primary slag develops to a final
slag. Four major components make up about 96 % of the slag, these being
SiO2, MgO, CaO and Al2O3. The balance is made up of components such
as manganese (MnO), sulfur (S), titanium (TiO2), potassium (K 2O), sodium
(Na 2O) and phosphorus (P). These components have a tendency to lower the
liquidus temperature of the slag. The definitions of basicity are given in Table
8.2 and typical compositions of hot metal and slag in Table 8.3.
B2 CaO/SiO2
B3 (CaO+MgO)/SiO2
B4 (CaO+MgO)/(SiO2+Al2O3)
Table 8.2 Definitions of basicity (weight percentage)
Hot metal Typical Slag Typical Range

Iron Fe 94.5 % CaO 40 % 34–42 %
Carbon C 4.5 % MgO 10 % 6–12 %
Silicon Si 0.40 % SiO2 36 % 28–38 %
Al2O3 10 % 8–20 %
Manganese Mn 0.30 %
Sulfur S 0.03 % Sum 96 %
Phosphorus P 0.07 % Sulfur 1%
Table 8.3 Typical hot metal and slag compositions
8.5.2 Slag properties

The final slag has a higher melting temperature than hot metal. In practice
it is more correct to think in temperature ranges than in melting points, as
composite slags have a melting trajectory rather than a melting point. The
liquidus temperature is the temperature at which the slag is completely molten.
At temperatures below the liquidus temperature solid crystals are present. These
solid crystals increase the viscosity of the slag. In our experience the behavior
of slag can be well understood on the basis of its liquidus temperature, except
for high Al2O3 percentages. Figure 8.12 on the next page shows the viscosity of
slags. Acceptable flow rates are at a viscosity below 5 poise. It is clear from the
figure that these values are exceeded at 20 % Al2O3.
156 Chapter VIII
10
Viscosity (Poise)
6
0
5 10 15 20 25
Slag Al2O3 content (%)
Figure 8.12 Effect of Al2O3 on slag viscosity at 1500 °C, in poise (10–1Pa.s).
Acceptable flow is effected below 5 Poise. Dots are calculated for
operating blast furnaces, the line is based on Slag Atlas (1995) with a B2
basicity of 1.05 and an MgO level of 10 %.
Liquidus temperatures are presented in ternary diagrams such as Figure 8.13.

These diagrams have been developed for pure components and in practice the
liquidus temperatures are somewhat lower.
Figure 8.13 Phase diagram of liquidus temperatures of blast furnace slag for 10 %
Al2O3. The red dot indicates a slag composition of 40 % CaO, 10 %
MgO and 36 % SiO2. To this end, these main components have to be
recalculated from 96 to 100 % of the slag. The area where the liquidus
temperature of the slag is lower than 1350 °C is indicated in yellow.
(After Slag Atlas, 1981)
Since in the ternary diagrams only three components can be indicated, one
of the major slag components is taken as fixed. i.e. Al2O3 content is 10 %.
Diagrams at different Al2O3 percentages are presented in Figure 8.14. The
typical slag composition for a blast furnace slag is also indicated (Table 8.3).
Note that the liquidus temperature (red dot) is about 1400 °C and that the
liquidus temperature increases when CaO increases (i.e. when the basicity
increases). However, the viscosity decreases as CaO and MgO increase because
these basic oxides break up the more strongly bonded network formed by the
acidic oxides SiO2 and Al2O3.
SiO2 SiO2
5% Al2O3 10% Al2O3
We
We
O
O
a
a
ig
ig
%) e C
%) e C
ht (0 –9
ht (0 –9
95 t ag
9 0 t ag
pe 5%
pe 0%
(0 – rcen
(0 – rcen
rce )
rce )
nt
nt
pe
pe
ag
ag
ht
ht
e
e
ig
ig
SiO
SiO
We
We
2
2
CaO MgO CaO MgO
Weight percentage MgO Weight percentage MgO

(0–95%) (0–90%)
SiO2 SiO2
15% Al2O3 20% Al2O3
We
We
aO
aO
ig
ig
%) e C
%) e C
ht (0 – 8
ht (0 – 8
85 t ag
8 0 t ag
pe 5%
pe 0%
(0 – rcen
(0 – rcen
rce )
rce )
nt
nt
pe
pe
ag
ag
ht
ht
eS
e
ig
ig
SiO
iO 2
We
We
2
CaO MgO CaO MgO
Weight percentage MgO Weight percentage MgO

(0–85%) (0–80%)
SiO2
25% Al2O3
< 1350 °C
We
1350–1400 °C
O
Ca
ig
ht (0 –7
1400–1500 °C
%) e
75 tag
pe 5%
(0 – rcen
> 1500 °C
rce )
nt
pe
ag
ht
eS
ig
iO 2
We
CaO MgO
Weight percentage MgO

(0–75%)
Figure 8.14 Phase diagrams of slag liquidus temperatures at various Al2O3 levels.
Areas where liquidus temperature is lower than 1350 °C are yellow.
(After Slag Atlas, 1981)
High Al2O3 slags

The Al2O3 content of the final slag varies according to the raw materials in use.
Especially in India, the raw materials are rich in Al2O3. Operational practice
shows that is possible to run a blast furnace with around 18 % Al 2O3. However
further increase of the Al2O3 content to 20–22 % is done by some companies,
but even higher content is (almost) impossible.
In Figure 8.15, the composition of the slag resulting from a burden of self–
fluxed sinter and pellets is indicated with a red dot. The liquidus temperatures
of the “pure” components show high liquidus temperatures for the slag, well
above 1500 °C. How is it possible that the material melts in the cohesive zone?
The secret behind the melting of sinter and pellets is that the ore burden
contains a high FeO percentage, which lowers the melting temperature or, as
mentioned earlier, lowers the liquidus temperature and solidus temperature.
158 Chapter VIII
This is shown in Figures 8.15 and 8.16.

SiO2 BF Slag
1 Olivine fluxed pellets
5
2 Typical basic pellets
We
O
3 Superfluxed sinter
ig
%) e C
h t (0 – 9
4 Fluxed sinter
9 0 t ag
1
pe 0%
Acid pellets
(0 – r c e n
5
r ce )
nt
pe
ag
ht
eS
ig
iO 2
We
CaO MgO

(0–90%)
10% Al2O3
Figure 8.15 The slag composition of typical pellets and sinter qualities
Figure 8.16 Slag formation
8.5.3 Hot metal and slag design for stops and starts
During special blast furnace situations, such as a blow–in or a very hot furnace,
the hot metal silicon can rise to very high values. Since the silicon in the hot
metal is taken from the SiO2 in the slag, the consequence is that the basicity
increases. This leads to high slag liquidus temperatures (Figure 8.17).
In a situation with very high basicity, the final slag does not remain liquid in the
furnace and cannot be cast. It will accumulate in the furnace, where it can form
a ring of solid slag, particularly in the bosh region. Burden descent and casting
will be disrupted. For this reason, for special situations where hot metal silicon
is expected to be high, the slag should be designed to handle the high hot metal
silicon. A reduced slag basicity is required. To this end, extra SiO2 has to be
brought into the furnace, and the recommended method is the use of siliceous
lump ore.
Figure 8.17 Slag properties at increasing hot metal silicon, a typical example
Some companies use quartzite, which is suitable for correcting the basicity in
normal operation. However, it is less suitable for chilled situations, since the
liquidus temperature of quartzite itself is very high (1700 °C). The effect of the
use of a siliceous ore is shown in the ternary diagram in Figure 8.18: by working
at a lower basicity, the liquidus temperature decreases along the line indicated.
SiO2
BF Slag
Low Basicity
Burden
We
O a
ig
%) e C
h t (0 – 9
9 0 t ag
pe 0%
(0 – r c e n
r ce )
nt
pe
ag
ht
e
ig
Si O
We
CaO MgO

(0–90%)
10% Al2O3
Figure 8.18 Effect of low basicity burden on slag liquidus temperatures. The red dot
respresents a normal slag composition, the brown dot a 20 % lower B2
basicity.
IX Casting Hot Metal and Slag
9.1 Introduction
The blast furnace process produces liquid hot metal and slag. The two liquids
drip down into the coke–filled hearth of the blast furnace, where they wait to
be tapped, or cast, from the furnace. Liquid hot metal and slag do not mix well
and separate. The densities of the two liquids are quite different; with hot metal
(7.2 t/m³) being three times that of slag (2.3 t/m³). The hot metal and the slag
separate inside the hearth before the liquids are tapped, as well as outside in the
runner system, with slag floating on hot metal. The present chapter is about the
question of how hot metal and slag are cast from the furnace.
A cast has a typical duration of 60–240 minutes. It starts with drilling a hole
(diameter 40–70 mm, length 2,5–4,5 meter). During a cast, hot metal and slag
flow within the hearth towards the taphole. Hot metal has a very low viscosity
and flows easily, but slag has higher viscosity and flows with greater difficulty.
At the end of a cast, the “dry” condition is indicated by a blowing taphole: gas
escaping with liquids causing spraying at the taphole face. At that point, the cast
has to be plugged. Note that at the end of a cast, the liquid hot metal surface in
the hearth is quite horizontal, but the slag level slopes downwards to the casting
taphole. As a consequence, a large amount of slag remains in the furnace at the
end of a cast, shown in Figure 9.1.
Figure 9.1 “Dry” taphole showing ingress of gas and the slope of the slag surface
162 Chapter IX
After that a cast is plugged, and another cast has to be opened. Most large
furnaces have continuous casting, which means that another taphole is opened
just before or as soon the cast has been plugged. The time between the end of a
cast on one taphole and opening the next is called gap time. To a large extent,
the gap time determines the highest liquid level in the furnace during next cast!
The majority of modern high productivity blast furnaces have been between
2 and 5 tapholes. During In normal operation of a furnace with two or more
tapholes, the tapholes will be used alternately, with one cast being on one
taphole, and the next cast being on the other. This also applies to furnaces with
up to five tapholes. The reason for having at least three tapholes is to ensure
that there are always two or more in operation, even during times of casthouse
repair, or emergency breakdown. This allows the production level to be
maintained, since two or more tapholes can drain the hearth.
Single taphole furnaces do exist, most of them being smaller size furnaces that
show different behavior and requirements for casting. Single taphole operation
is described in Section 9.6.
The objective of casthouse operation is to prevent interference in the liquid level

inside the furnace. To this end the liquid level has to be kept as low as possible.
Continuous or alternate casting is the basis for the description in this chapter.
– How liquid level interferes with the process
– Casting regimes
– Flow of hot metal and slag through the taphole
– Taphole operation: drilling, plugging, clay quality
– Casthouse operational challenges.
9.2 Liquid level and the blast furnace process
Why is the blast furnace operator concerned about the liquid level in the
hearth? How much time does he have to take decisions? About the latter
question: how long does it take after a dry cast before the liquid level reaches
the level of the tuyeres? Because, as soon as slag is at tuyere level, the furnace
can no longer be stopped without filling all tuyeres with slag. In Example 9.1, it
is calculated to be 79 minutes, so timely actions are required in case of delays.
How does liquid level affect the blast furnace process?

Experienced operators will tell you that with poor casting, so increasing liquid
level:
– The blast pressure may increase,
– The burden descent may slow down,
– Blast pressure increases and heat losses (or stave temperatures) increase.
The operator will also tell of burning tuyeres, tuyere stocks that fill with slag on
occasion, and of blowpipes that “burn out of the furnace”. The latter two are
incidents with major damage.
Casting Hot Metal and Slag 163
Example 9.1 How long does it take to fill hearth with liquids
from taphole to tuyeres?
The time available to fill the area between the taphole and tuyeres can be
estimated by taking the volume of the produced liquids into account and
the voidage of coke, here taken as 20 %. According to Example 9.1: for the
reference furnace it takes 79 minutes to fill the voidage between taphole and
tuyeres with fresh liquids from the production process. The voidage is not
uniform through the hearth, there is low voidage in the deadman and close to
the bottom, while there is high voidage expected in a ring around the hearth.
Distance taphole–tuyeres m 4.6

Diameter m 15
Volume tuyeres–taphole m³ 812
Voidage % 20
Volume available m³ 162
Hot metal production tHM/d 12000
m³/hr 69
Slag rate kg/tHM 250
m³/hr 54
Liquids production m³/hr 124
Time to fill min 79
Table 9.1 Time to fill the reference furnace at full production from “dry” to
liquids at the tuyeres
Burden descent
The liquid level in a blast furnace affects burden descent. Figure 9.2 shows an
operational example of burden descent and casting for single taphole operation.
Descending so fast that
the charging system can’t
keep up—stockline lost
Charging speed slows as Increased speed of burden

furnace hearth fills descent as liquids are tapped
Figure 9.2 High residual liquid levels and burden descent in a singe taphole furnace
(1400 m³ IV)
164 Chapter IX
There is a large gap time, and so in between casts, the liquid level rises and
burden descent slows. When the taphole is casting, burden descent speeds up
significantly, especially when there is a large flow of liquids towards the end of
cast. This may occur to the extent that the charging system is unable to keep up
and a lowered stockline is the result. The example is extreme, but similar effects
can occur in alternatingly casting furnaces. Generally, with increasing liquid
level, burden descent slows and blast pressure increases, even when the liquid
level is still far below the tuyeres. How is this possible?
A major effect of a high liquid level is that it influences burden descent and blast
pressure by changing the flow of coke towards the tuyeres. For the reference
furnace, about 10–20 lumps of coke are drawn into the raceway per second. The
cokes moves over the tuyere body and nose as indicated in Figure 9.3a, because
of the low pressure at the nozzle caused by the high wind velocity. Tuyeres
work like an ejector, with low pressure and coke ingress being found around the
tuyere nose circumference. It was found experimentally that coke even 2 meters
below the tuyere is drawn to this “tuyere injector”.
Figure 9.3a Coke flow towards the raceway during normal operation
– Tuyeres work like ejectors
– Local low pressure at the nose draws coke into the raceway, all around
tuyere nose
– Below the raceway there is a border between “stationary and moving”
coke (dotted line), the border of the deadman
– Coke is supplied from the active coke zone (arrow)
Figure 9.3b Coke movement with increasing slag level

– The stationary coke comes in the moving coke zone and starts to flow
towards the tuyere
– Coke flow from the active coke zone is impeded. Slower burden descent
– More voidage in hearth (coke is not replenished)
Figure 9.3b explains how this mechanism changes when hearth liquid levels
are higher. Coke submerged in hot metal and slag tries to float and thus exerts
an upward force, which increases as the liquid level rises and decreases with
falling liquid level (Section 7.2). The upward force of the submerged coke presses
from below through the coke grid on the cohesive zone, which is compressed
from above by the weight of the charge. Softening materials, in particular, are
compressed, and the voidage in the cohesive zone decreases. This makes the
burden less permeable, so hot blast pressure will increase. It is important to
note that the tuyeres opposite the operating taphole are the first to be affected
because the slag level is highest opposite the casting taphole, so that the
mechanism leads to asymmetry.
If the slag level reaches the tuyeres, the situation becomes quite severe: dispersed
slag droplets will start to be blown into the active coke zone, gas flow will bend
towards the wall, and heat losses in bosh and belly will increase. If the furnace
has to be stopped, all the tuyeres fill with slag. Although this is poor operation,
furnaces frequently overcome the problems if appropriate liquids are cast in
time.
If the liquid level continues to increase, so that hot metal rises to the
tuyere level, hot metal will flow into the tuyeres and blowpipes and cause a
catastrophic failure. This will lead to a blow–out of coke and slag and cause
a critical emergency stop, with all tuyeres filled with slag, sometimes into the
bustle main. Blast furnaces cannot overcome the problem of hot metal at tuyere
level without major failures.
166 Chapter IX
In summary and in other words: high liquid levels affect blast pressure, burden
descent and heat losses, as is shown schematically in Figure 9.4. High liquid
level affects the coke movement into the raceway burden descent, as well as
cohesive zone shape and permeability. The bosh gas is deflected more towards
the wall, rather than through the center of the furnace. In this instance the
bosh is subject to much higher heat loads than normal, and the root of the
cohesive zone will move downwards. However, at the same time, the cohesive
zone will drop at the center of the furnace, due to the reduction in gas passing
through the center. The blast pressure will also be higher as the resistance in
front of the tuyeres is higher, and the burden descent will slow considerably.
Figure 9.4 Consequences of increased liquid level (red arrows indicate burden
descent rate, the arrow along the wall increasing heat losses)
For these reasons, the liquid level inside the furnace has to be kept as low as
possible. Various casting regimes can be used for this.
“Dry” condition and measurements of liquid level

In the past, observation of a “blow” was the decisive indicator of a dry hearth,
In reality, a blow shows that gas is escaping from the taphole. So, in situations
with cracks in the mushroom or very large flows just before the blow, a very
large amount of liquid can remain in the furnace, so the furnace is not “dry”.
For this reason, various methods have more recently been developed to observe
the liquid level in the hearth more directly. The most applicable methods are:
1. Measurement of EMF (electro motive force) in the furnace: the signal is
sensitive to the amount of liquid within the furnace (Figure 9.5). Operators
tend to rely more and more on it, even though the measurement is affected by
all electrical influences, like hot metal temperature. All these influences have to
eliminated for reliable measurements.
2. Calculating the balance of slag and hot metal from the production level and
the slag and hot metal flow measured. Slag flow can be measured continuously,
depending on type of granulation – from temperature increase of granulation
water, from power required to drive the rotational dewatering of the slag, or
by a weighing system in the conveyor belt that carries the slag to storage. Hot
metal flows can be measured directly with a weighing system in the rails, from
the descending velocity of the springs of the ladles, or from the level in the
ladles used for transportation of the hot metal. An example is shown below in
Figure 9.10.
5.0 #1TH #2TH
4.0
3.0
2.0
1.0
0.0
0 12 24 36 48 60 72
hours
Figure 9.5 Signal of EMF showing the liquid level in the hearth
(Seong–Moon Kim and No–Kun Hur, 2011)
Direct measurements of slag and hot metal flow tend to be more accurate and
more reproducible than indirect measurements, but they are not sufficiently
reliable to indicate that the maximum level in the ladles has been reached for a
timely reaction.
9.3 Monitoring casting regimes
During the cycle of starting a cast, plugging it and opening the next cast, the
liquid level varies inside the furnace. The gap time between casts determines
how high the level in the furnace will be. The duration of a cast depends not
only on gap time, but also on drill diameter, taphole length and the quality
of the clay, i.e. the resistance of clay to wear, especially from slag. Operational
working methods also contribute to variations: is the taphole well plugged,
is it drilled completely through, is there gas leakage in the taphole? All these
influences make cast duration and the moment when slag is cast quite variable.
For this reason, it is important to monitor casthouse operation.
This can be done with the help of a cast tracking chart in the control room to
provide the operator with a quick overview of casting in the last shift or day.
An example of a chart is shown in Figure 9.6 on the next page. On the basis of
a chart like this, the actual situation can be reviewed immediately, and delays,
drilling problems or other problems become clear. The charts are from a well
operated furnace in March 2020.
168 Chapter IX
hot metal
slag
A
west
B
south
night shift morning afternoon
slag delay back to back interrupted
Figure 9.6 Cast tracking charts – 24 hour overview for three different days in March
2020 (2700 m³ IV blast furnace with two tapholes).
Issues like slag delays or back–to–back casting are immediately clear from a
tracking chart like this. In this example:
A. is a day of good casting,
B. is a day with some problems,
C. shows a day of single taphole operation with slag delays on nearly every cast.
The effects of casting parameters like gap time, drill diameter, taphole length
and clay quality can be analyzed with a computerized model, which is described
in Example 9.2.
For high productivity and high slag volumes, large modern furnaces sometimes
operate three tapholes. An example is shown in Figure 9.7. The furnace is
casting 41 hours per day with 96 % slag coverage.
Figure 9.7 Triple taphole operation in a blast furnace (4400 m³ IV, production 2.2
tHM/m³IV.24hrs, slag rate 500 kg/tHM)
Example 9.2 Simulation of cast duration and hearth level

Cast duration and hearth level can be estimated with a simulation model
(originally developed by Molenaar and published in Opbergen and Toxopeus,
1999). In the simulation model, the effects of taphole length, drill diameter,
clay wear and gap time are considered. The model has been calibrated with
observations at numerous furnaces. It provides very reliable results and helps
to understand the effects of various measures. In the present example the
production data of the reference furnace are used. The furnace is casting
excellently with a wear resistant clay.
A
good casting
B
30 min gap time
same drill
C
drill diameter
60 mm o 70 mm
D
lap cast
Figure 9.8 Effect of gap time on cast duration, liquid level and slag coverage
Figure 9.8 shows the effect of a 30 minute gap time. Figure 9.8a shows good
casting with duration of 150 minutes, a highest level of liquids of 1 m above
the taphole and nearly 100 % slag coverage (time slag/time hot metal). If after
a 30 minute gap time, the cast is opened with the same drill diameter, casting
gets out of control as shown in Figure 9.8b: slag tapping starts after 200
minutes and the slag is already above tuyere level. There are three options for
remedial action:
1. Increase drill bit from 60 mm to 70 mm, as shown in Figure 9.8c.
Nevertheless, the highest level in the furnace will rise to 2 meters and slag
coverage will be reduced to 80 %. This can be only be done before the taphole
is drilled. If not:
170 Chapter IX
2. Open a lap cast: open another taphole to enforce liquid drainage (Figure
9.8d). This is normally done when there is no slag flow for 60 minutes after
closing the previous taphole. Liquid level reaches 2.5 meters above taphole
level. If this is not possible:
3. Reduce production rate, while keeping maximum top pressure.
If a furnace is casting on one side only, with 30 minutes gap time, the model
results are identical to those in Figure 9.8c – an increase in the highest
level from 1 to more than 2 meters above the taphole and a decrease of slag
coverage.
Estimations can be made on the basis of the model of cast duration as being
dependent on various casting parameters. Another aspect is wear of clay,
because taphole diameter increases during slag casting, in Figure 9.8 good
clay was used. Comparison with slightly poorer clay is shown in Figure
9.9. The cast with poorer clay was opened with a smaller drill diameter to
achieve a similar duration. Although the casts look very similar, there are
large differences: with poorer clay the hot metal and flows are smaller at the
beginning of a cast. With poor clay slag flow almost disappears during the
cast, and finally the slag flow is much larger at the end. This large flow at the
end of a cast leads to more slag being retained inside the furnace at a blow.
The maximum level increases from 1 meter for good clay to 1.5 meters for
poorer clay, which is increased by an unknown amount because more slag is
retained.
Figure 9.9 Effect of clay quality on casting

Left: good clay, wear 0.15 mm/min
Right: slightly poorer clay, wear 0.3 mm/min
A special situation occurs when two operating tapholes have a different

taphole length. Due to the fact that a taphole has an inclination of 8–12
degrees, a difference in taphole length causes casting at a different level in the
furnace. The shorter taphole has a higher slag/hot metal ratio than the longer
one. Also, the cast duration is enlarged at the longer taphole. Consistency
is important: if taphole length decreases, less liquids can be cast, if taphole
length increases, more liquids can be tapped and there will be a slag delay.
And adjusting from short to long taphole by plugging more clay is very slow
and not very predictable.
Real flows of hot metal and slag as measured for hot metal from the
descending velocity of ladles and for slag from the temperature increase of the
granulation water are used to calculate the liquid level in the furnace as in
Figure 9.10. The casthouse is very well operated. Note the increase in slag and
hot metal flows from the beginning of a cast onwards. This type of data was
used for developing the model mentioned here.
TH 1
TH 2
TH 3
16
hot metal
(t/min)
8
6
(t/min)
slag
0
(m above taphole)
level in hearth
Figure 9.10 Typical casting with a dual taphole operation in a three taphole
furnace, showing hot metal and slag run times at top and hot metal
flow, slag flow and hearth liquid level (4400 m³ blast furnace)
172 Chapter IX
9.4 The taphole

The tapholes are the most vulnerable areas of the blast furnace due to the
constant wear and tear and reliance on consumable materials, equipment, and
manual intervention. If any of these factors are performing less than optimally,
then deterioration in the taphole performance is the likely result. Taphole
heights are drawn at the outside, but tapholes are drilled under an angle sloping
8–12 degrees downwards. A slope of 10 degrees means that for every meter drill
length, the taphole enters 17 cm lower inside the furnace. The larger the angle,
the deeper the furnace is cast, and the more hearth volume is available, but the
more the liquid splashes into the runner system, which causes wear and affects
hot metal–slag separation.
The taphole itself has to fulfill the following conditions (see Figure 9.11):
1. Consistent in length, since length affects casting behavior.
2. Consistent in diameter for reliable consistent volume flow. Minimum wear
of diameter during cast discussed above, although some minor wear helps to
secure a dry hearth.
3. Smooth inner surface since a coarse inner surface causes additional turbulence
in the taphole which leads to more resistance and then drill diameter will
become larger than required.
4. No gas leakage, as gas leakage will cause spraying casts and jeopardize
the “blow” indication. Cracks in the refractory material of the taphole are
unavoidable. For this reason, the working methods and taphole clay have to be
designed for “crack–healing”.
5. No “secondary” cracks with liquid connection: a liquid connection in the final
1.5 m of a taphole will lead to false blows.
6. No hot metal and slag left in taphole after plugging, because this makes drilling
the next cast difficult and may lead to oxygen lancing of the taphole.
In order to meet all these conditions, the total system of taphole length, clay
quality, plugging, taphole face, drilling and working methods have to be
considered.
Taphole length
The taphole end has to be located at a point where flow of liquids towards it is
assured, so it has to be in a pool of liquids with good permeable coke. It is not
too far inside, because the deadman has little coke voidage, and it is not too
close to the wall, since hot metal flow along the wall wears the refractory. The
location has to be found empirically, but will probably be somewhere in the area
of the end of the raceway above it. Typically, the raceway is 180 cm and tuyeres
are 40 cm inside, to which have to be added the thickness of the refractory (for
example 70 cm plus shell and chapel, for example 50 cm), so in total somewhat
less than 340 cm from the taphole front. But note, taphole length is determined
by furnace design and chapel design, so the optimum is furnace–dependent. A
longer taphole can have a higher resistance and may be opened with a slightly
larger drill diameter.
A sketch of an operating taphole is shown in Figure 9.11. In the sketch, a

number of taphole issues are indicated, from the inside of the furnace to the
outside the following should be mentioned:
a. The taphole is drilled through solid material, basically skull with some
refractory in it. It is a relatively unstable structure so cracks can form. When a
crack forms, it starts to drain liquid from a higher level – a “secondary taphole”
is formed. This will give a false blow as soon as gas reaches the secondary
taphole. As a consequence, the next cast on another taphole will cast hot metal
only for a prolonged period and thus with poor slag coverage.
b. The taphole passes through the hearth wall refractory. These refractories can
become slightly porous and can lead to gas leakage into the taphole, because
pressure in the furnace is much higher than the pressure in the taphole.
Especially along the refractory boundaries, gas leakage can occur. The taphole
wears during casting, which means increases in diameter, and gas leakage can
increase during casting. Gas leakage into the taphole causes additional wear,
reduces the casting rate, and makes the cast spray.
c. The refractories are cooled and thus the curing of taphole clay on this position is
impeded. This is the place where operators often find “wet spots”.
Figure 9.11 Taphole in operation, showing gas leakage into the taphole and
secondary taphole via crack. High and low temperature refer to refractory
temperature – low temperature results in slow curing of clay.
Taphole clay
Taphole clay has to be gas–tight, resistant to wear, mainly by liquid slag,
capable of filling cracks and capable of pushing hot metal and slag from the
taphole when plugging. Example 9.3 describes the components of taphole clay.
A dense material is required. It has to be hard clay (which means plasticity
as low as possible), so that the gun can just push the clay into the taphole at
174 Chapter IX
maximum compression. This has as the consequence that taphole clay has to be
used at a constant temperature. This means: storage in stock as well in the clay
gun at constant temperature by water cooling/heating. By doing this, the effects
of the parking position of the gun as well as seasonal influences are eliminated.
Tar–bonded and resin–bonded clays have different properties. Resin–bonded

clay cures 10–20 times faster than tar–bonded. This is an advantage for having
the taphole available quickly, but also a disadvantage since post–pressing is not
readily possible, so that tar–bonded may lead to better gas–sealing by improved
penetration into cracks.
Plugging
Plugging is pushing the taphole clay into the taphole and in so doing removing
all hot metal and slag from it. Any remaining drops of hot metal will jeopardize
proper opening of the next cast. In order to remove the liquids, the refractory
material has to be pushed in in a controlled way, so with a constant volume/
second which can be realized with automated control of the pressure of the
clay gun piston. Best results are obtained with a very smooth taphole interior
without any irregularities.
A major problem of plugging is clay spillage. As soon as clay is spilled, the

internal compression of the clay is lost and plugging is poor. In order to prevent
clay spillage, the following conditions have to be met.
1. Flat and smooth taphole face perpendicular to the taphole.
2. Adequate sealing of the furnace to the face by having sufficient pressure from
the slew mechanism of the gun on the taphole nozzle towards the furnace.
Example 9.3 Taphole clay composition

Taphole clay is a mixture of a number of different materials with varying size
distribution. Since taphole clay has to be dense, the aggregate particle size
distribution of the blend of different materials has to fulfill the requirements
of the Fuller distribution for maximum compactness. Moreover, the largest
particles in the blend still have to be very fine (smaller than 0.5–1 mm), so
that material can be pressed into cracks.
Taphole clay contains:

1. SiO2 – is a filler material, that can be minimized because it dissolves in slag.
2. Carbon–bonding – tar and/or resin can be used for gluing the particles
together. Use a minimum, since excess leads to increased wear.
3. Al2O3 – three qualities of Al2O3 are available normal, fused and sintered; they
all can be used, with the sintered giving the best results and being the most
expensive. Al2O3 lends mechanical strength to the clay.
4. Ferro Silicon Nitride (FeSi3N4) – facilitates taphole stability, does not react,
and its quality can be quite variable. It is expensive – can be 30 % of volume.
5. SiC – reacts with FeO in slag, expands and facilitates lower porosity.
Figure 9.12 Hydraulic pressure pushes the nozzle to the furnace, the piston
compresses the clay into the taphole causing a counter force caused by
friction and the pressure of gas and liquids
This overcomes the counterpressure of the clay that pushes the gun from the
furnace. (Figure 9.12). To this end, the width of the steel of the nozzle of the
gun should not be too large, typically not more than 3 to 4 cm. Sealing requires
a compressible surface between taphole face and nozzle of the gun for which a
wooden laminated ring can be used mounted on the gun.
Clay spillage prevents proper plugging. But in addition: the taphole face can be
damaged when the spillage is removed by oxygen lancing. This is costly, difficult
to do, causes additional damage and usually costs production time.
When the cast has to be plugged, the clay gun is moved in front of the
taphole. The movement of the gun has to slow down to prevent collision with
the taphole face, but as soon as it touches the taphole face the clay has to be
pushed into the taphole. The next step is to push a defined amount of clay
into the furnace after which the piston movement of the gun stops. Shortly
thereafter the compression in the clay decreases. In order to fill small cracks in
the refractory, postpressing has to be applied: this is a very short push of the
piston, so that only the pressure increases, filling the gaps, but the clay does not
move. Movement will destroy the clay structure. Proper post–pressing has to
be automated, since optimization of manual postpressing by an operator is not
feasible. While the clay is curing in the taphole, there are three zones: in the
furnace the temperature is high and curing is fast, in the cooled refractory the
temperature is lower and curing is slow, and in the chapel the temperature is
slightly higher again and the curing is good, but slower than inside the furnace.
Plugging is summarized in Figure 9.13 on the next page.
176 Chapter IX
Figure 9.13 Clay being pushed into the taphole
Taphole face
The quality of the taphole face and the pressure of the clay gun determine clay
spillage. The taphole face has to be strong, smooth and have a long service
life, so that it does not need intermediate repairs. Taphole face can easily be
damaged by the collision with the clay gun the moment it touches the face.
For this reason, the movement of the gun towards the taphole face should slow
down for the final 10 cm and the pressure should increase the moment the gun
touches the face. At the very same moment the clay has to be discharged into
the taphole. This has to be an automatic control. Only the very best castable
refractories are suitable for the taphole face.
Drilling
Cured taphole clay is brittle material, which every operator can check by
himself. Smooth drilling in brittle material requires a slow and constant forward
movement. This has to be (automatically) controlled in the drill system. Fast
forward movement on a certain position, or very slow movement, will damage
the smoothness of the taphole. If properly plugged, hammering is not needed.
Hammering will severely damage the taphole wall.
An appropriate drill bit is required to have a straight taphole. The drill bit has
to guide itself through the taphole as in figure 9.14b. Drill bit A has no self–
guidance and bending of the drill shaft will cause a change in direction of the
taphole. The drill will find a path with least resistance in the cured clay, which
can deviate from the intended taphole direction. Drill bit B has self–guidance,
that together with a stiff drill shaft will follow the path of the intended taphole.
During drilling wear of the drill bit has to be prevented. To this end the drill bit
has to be cooled, but cooling with air will burn the bit because of the oxygen,
so cooling has to be done with nitrogen, which has increased heat capacity
by adding water mist to it. The water addition has to be stopped as soon as
nitrogen flow stops to prevent water in the taphole.
Figure 9.14 Drill bits used for opening taphole
As taphole lengths are around 3.5 meters, drill rods will have a length of around
4 meters. During drilling, force is used to push the drill forward. This will lead
to bending of the drill rod which prevents the drill from following a straight
path into the furnace, causes the drill to oscillate and creates a larger taphole
diameter than the drill bit size. To avoid this, the drill machine has to be
equipped with a guiding system at the front and halfway along the sledge of the
drill machine.
9.5 Single taphole furnaces
Furnaces with only one taphole are of course optimized for tapping single sided.
These furnaces are in general small furnaces and in practice these furnaces are
less vulnerable to increased gap times. This can be explained by the relatively
much smaller deadman and the smaller distance to the other side of the
furnace, so the slag liquid inclination in the furnace is more level. Moreover,
these furnaces have a larger average voidance in the hearth than large furnaces
due to the stronger vertical movement of coke in the hearth. For single taphole
furnaces, the clay type has to wear more quickly, because the wearing of the clay
makes the casting flow, especially for increased slag, and is a kind of safety valve
to ensure that liquids are removed.
In single taphole furnaces the minimum gap time is often dictated by the curing
time for the clay. If the taphole is opened before the clay has hardened, much
of it will easily wash away, which will quickly erode the taphole mushroom and
expose the taphole refractory block itself. For this reason, many single taphole
furnace use resin–bonded type clay that hardens quickly.
178 Chapter IX
9.6 Casting problems and remedial actions

General remarks
In the event of casthouse problems, use is sometimes made of a larger drill bit to
open the taphole. It should be kept in mind, that recuperation of the originally
smaller diameter takes time. For example, from experience: after drilling a
10 mm larger taphole, so 60 mm for instance instead of 50 mm, the taphole
diameter recuperates only 1 mm at the taphole wall per cast, so 2 mm in
diameter. So, it takes five casts to come back to the original drill diameter.
A second remark on production level: When the casthouse can no longer cast
all liquids, the production has to be decreased by reducing blast volume. In
that situation, the top pressure should be retained on the furnace in order to
keep pressure on the taphole as high as possible for liquid flow. Note, that
in the event of external demands to lower the production, the operator has
to try to keep the gas flow through the furnace consistent. This can be done
by decreasing top pressure. In practical terms: decrease the top pressure to
the extent that the wind velocity in the tuyeres is constant. So, handling top
pressure in case of delays depends on the particular problem.
Casting delay
In the case where the operator is faced with a casting delay, different actions
may be taken depending on the current condition of the blast furnace. The next
cast can be opened with a larger drill diameter or the production rate can be
decreased. See Example 9.2 above.
No slag casting
As soon as slag is not draining properly from the furnace, the following
measures can be taken:
– Using a larger diameter drill bit on the next cast will increase the flow, and may
improve the situation.
– Changing to the other taphole may improve the situation inside the furnace.
– Opening the second taphole, so two tapholes are open at the same time (overlap
casting), should be done after a defined period of no slag casting as specified
in the standard operating procedures for the plant, mostly 60 minutes after
plugging the previous cast.
– Reduction of production level by decreasing blast volume and/or oxygen
enrichment.
– Shorter gap time.
Not dry casts

A cast that has ended before all the liquids have been drained from the hearth
is described as a not dry cast. Depending on the cause of the not dry cast,
slightly different reactions may be appropriate. Where the not dry cast is known
because the taphole is closed for operational reasons, the second taphole should
be opened immediately with a larger drill bit. Where this is not possible, the
oxygen and then wind rate should be reduced and the original taphole re–
opened as soon as possible. Where this is not possible, the decision to close the
taphole should be delayed as much as possible, with wind rate being reduced
as far as liquid levels, as seen at the tuyeres, will allow. At this point there is a
balance between how much damage is being caused outside the furnace due,
for example, to molten metal spill, compared to the danger of flooding tuyeres
with slag and hot metal. Timely action for wind reduction is required in order
to prevent major damage.
In cases where the taphole has shown signs of the hearth being empty, but it is
thought from the casting times and amount of slag cast that this is not the case,
then there are a few different actions that may be considered. If there is a second
taphole available, then it may be opened prior to the first taphole being closed.
Once this is safely open, the first one may then be closed, known as overlap
casting. Alternatively, the normal gap time between casts may be reduced to
zero, so the second taphole is opened immediately after the first is closed. It is
important to ensure that both tapholes do not finish casting at the same time
as this will introduce a necessary gap time when there are two clay guns in
the taphole waiting for the clay to harden. So, once slag appears at one of the
tapholes, it should be closed to allow the other to cast normally. This technique
of when to open and when to close a second taphole should be included in
the standard operating procedure for casting to ensure that the best sequence,
proven in practice, is followed by all operators.
These same actions may also be taken if the blast pressure is affected by a
possible build–up of slag in the furnace. At the same time, however, other
causes of increasing blast pressure should also be investigated.
One–side casting
The single most important effect of single taphole casting compared to
alternate casting is that of the gap time. During the gap time the furnace is
still producing liquids but not casting. Ideally the gap time is calculated as
the optimum to allow liquid accumulation for a smooth cast of the expected
duration with controlled liquid level. However, the gap time can also be affected
by external factors such as how long it takes to change torpedo ladles, clay
curing time, maintaining and cleaning the runner system. It is very important
to remember that the furnace is still producing liquids at the same rate, unless a
change is made to slow down the production.
If a furnace must switch from alternate to single–sided casting, gap times will
vary considerably between the two practices. If alternate casting requires a gap
time shorter than the time it takes for the clay to harden, then single casting
will require a change in practice. If faster–curing clay is available, then this may
be applied, but caution should be used during the transition as the clay already
in the hole may not combine well with the new clay.
If there is a significant difference in the gap time, the production rate has to
be reduced to minimize the fluctuation in hearth liquid levels. Experience
has shown that single taphole operation can sustain production levels of up to
180 Chapter IX
5500–6000 tHM per day for a 2800 m³ IV blast furnace and 8000 tHM per
day for a 4400 m³ IV blast furnace when operated at low slag volumes (below
250 kg/tHM). This is often a significant reduction compared with what the
furnace is usually producing. The reason is, that the production rate may not
exceed the drainage capabilities.
Oxygen lancing
On occasion, opening the taphole using oxygen lancing is unavoidable. This
practice should be considered a last resort as it is extremely damaging to the
taphole refractory. Where the use of oxygen lances is increasing, the situation
should be investigated very closely to identify and solve the root cause. Where
oxygen lancing is unavoidable, it should only ever be done by experienced
casthouse workers, following the pre–drilled hole to ensure that the lance is
burning in a straight line down the center of the taphole. If more than one lance
is required, the interval between the two should be as short as possible, with
the practice continuing until the taphole has been opened. Where this causes a
long delay to the cast, alternative or additional actions such as opening a second
taphole or reducing wind rate should be considered at an early stage.
Repeated use of oxygen lances to open the taphole is likely to cause irreparable
damage to the taphole area, and may even pre–empt a taphole break–out or
necessitate an extensive taphole repair to avoid such a break–out. There is a very
large risk associated with using oxygen lances as it is very difficult to ensure
that the lance is burning in a straight line. Damage to the taphole block or to
taphole staves are the biggest concern. Operators can practice oxygen lancing in
a block of cured clay.
9.7 Hot metal and slag separation
Hot metal and slag separate outside the furnace in the runner system as
illustrated in Figure 9.15. The trough or main runner will still hold liquids from
the preceding cast, so when the hot metal from the next cast starts flowing,
it will then increase the level in the runner. The hot metal already under the
skimmer will also increase in height and start flowing again over the hot metal
dam. This hot metal will then flow to the tilting runner and into a torpedo
ladle. Once the ladle is full, the tilting runner will be repositioned into a
torpedo ladle which is parked parallel to the full one, so this second ladle can
also be filled. The full ladle will be removed and replaced by an empty ladle, so
that the cast is not interrupted. This alternating between ladles via the tilting
runner is done until the hearth is empty and the cast is complete.
The slag sits on top of the hot metal, so it does not flow under the skimmer as
long as the separation remains good. Once it has reached a certain level in the
trough it will flow over the slag dam and either to a slag granulator or to a slag
pit or ladle. It is very important that hot metal is not allowed to go over the slag
dam as this can result in explosions in the granulator, burning a hole in the slag
ladle or difficulties in emptying the slag pit. For yield reasons, slag going into
the torpedo ladle has to be prevented. Slag in the ladle can result in a solid slag
skull at the mouth of the ladle, resulting in difficulty in pouring the hot metal
at the steel plant. Best separation is reached by
– Laminar flow in the main runner, which means a cast rate just above the
production rate. Flow in the main runner changes from turbulent to laminar as
indicated in Figure 9.15.
– Properly designed levels of runner bottom, skimmer, hot metal dam and slag
dam, which are maintained at design specification. Note that the hot metal
slag interface moves up and down in the trough depending on relative flows.
This separation line has at all times to be 10 cm above the level of the skimmer
opening and 10 cm below the slag dam level.
The more stable the cast flow, the better the separation.
Improved separation leads to cost savings because of improved yield. Controlled

casting with a good slag coverage leads as well to improved hot metal sulfur,
since desulfurization of hot metal to a large extent takes place in the taphole
(Chapter VIII).
Figure 9.15 Slag and hot metal separation in the hot metal runner, or trough
X Control of Blast Furnace Operation
and Burden Selection
10.1 Production level and reductant rate
The actual production level of a blast furnace is derived from the number
of charges per hour and the hot metal quantity per charge. However, the
production level has a delay compared to the momentary value, because
monitoring charges per hour takes a few hours. In order to take into account
short term variations in hot blast volume, oxygen enrichment, moisture and
injectant rate, the real time production level has to be calculated from oxygen
input into the tuyeres from air, enrichment oxygen, blast humidity and coal.
This method assumes a fixed amount of hot blast oxygen per tonne of hot
metal, called the specific oxygen rate. The specific oxygen rate is periodically
updated based on charges per hour production rate and tuyere input oxygen.
This method is not sensitive to shorter term changes in specific oxygen rate
due, for example, to increased direct reduction; in this situation the calculated
production rate will be incorrect. With a high accuracy top gas analysis system,
it is possible to calculate the real time production rate from the oxygen balance
from minute to minute. This gives a faster indication of when the production
rate is changing. Constant monitoring of “expected” production and “real
production” is required to find the reasons for changes, like an increased
production rate caused by direct reduction, so that timely actions can be taken.
Using charging rate works well for small changes in burden, but for large
changes the delay between charged at the top and processing of the charged
material 5–8 hours later has to be taken into account. Burden tracking in the
furnace is helpful and is a standard part of most modern Level 2 monitoring
systems
The operator needs to understand:

– production comes from oxygen blown into the blast furnace that gasifies about
340 kg/tHM of reductant; the 160 kg difference with the 500 kg reductant
rate is for direct reduction (around 110 kg) and carbon dissolved in hot metal
(around 50 kg.
– production decreases when less oxygen is blown into the furnace.
– production decreases when more reductant per tonne is used, for instance by
increasing coal injection or charging a coke blank.
– production increases when direct reduction increases.
184 Chapter X
In general, the oxygen input through the tuyeres is closely monitored, although
it is very difficult for the operator to get to know the real input of oxygen, as
oxygen comes from hot blast, the oxygen flow for enrichment, the moisture in
the hot blast (includes ambient moisture) and oxygen from coal. A standard
Level 2 system will calculate this on–line by adding all oxygen input together,
for instance in m³ STP/min. It is also helpful if the control system of the
furnace uses oxygen in a ratio to blast volume, so that a constant enrichment
percentage is maintained when blast volume is reduced. Coal and gas injection
should be controlled as kg/tHM.
It is already difficult for the operator in the control room to observe the oxygen
input and thus the production level. It is even more difficult to observe the
actual fuel rate used to produce a tonne of hot metal. The charged coke rate
and injection rate are known, but coke used for direct reduction is difficult to
observe when it is expressed as a direct reduction percentage. It is more helpful
to express it in kg/tHM of coke. Rules of thumb can then be used to translate
direct reduction, heat losses and changes in gas utilization to the required fuel
rate. Or a mass and heat balance can be used for online calculations.
The blast furnace operates with aims for the most important process variables:
– Input through tuyeres: blast volume, hot blast temperature, oxygen enrichment
of the hot blast, hot blast moisture (in g/m³ STP), PCI rate and gas rate.
– Input at top: charge weights of the various materials (coke, sinter, pellets, lump,
miscellaneous materials).
Since furnace efficiency and raw material quality are variable, production level
and reductant rate have to be reviewed every shift. Remedial actions in case of
deviations are documented in a system of standard operating procedures. This
will work well as long as a furnace is under control.
Blast furnace operators have different practices with respect to the consistency
of the production level. Most companies pay a lot of attention to keeping the
blast volume constant for weeks. These companies have done, and are doing, a
lot to keep the quality of raw materials consistent, to eliminate all mechanical
delays and to automate working methods. It is certain that this practice leads
to very good operational results. Different companies are more flexible and
change blast volume if and when required several times per week. They will
allow burden quality to change if there are opportunities in procurement, will
try to reach very low coke rate and will change to a slightly higher coke rate
if permeability deteriorates. Wind volume aims at production, that the steel
plant can process. These companies have standard operating procedures for 10
or 20 % lower production rates and feel confident in stopping a furnace; the
blast furnace is seen as quite forgiving. Besides technical reasons between these
two modes of operation, like the capacity of the steel plant, there are as well
cultural differences. But note that some of the more flexible companies are also
performing excellently and can reach considerable cost advantages by making
use of procurement opportunities and avoiding plating of hot metal.
Control of Blast Furnace Operation and Burden Selection 185
10.2 Under control and out of control

Blast furnace operation can change in a short timespan to a situation where
the process is no longer under control and actions for returning the furnace to
normal operation are required.
For instance, the thermal state of a blast furnace can change dramatically in
few hours. An example is shown in Figure 10.1, where hot metal temperature
decreased from 1440 °C to below 1390 °C in less than 10 hours. The operator
may well have reacted according to the standard operating procedures, but still
the furnace chilled. Most important is, that an operator recognizes the situation
from burden descent (slippery), sudden high cooling losses (double normal
values, localized peaks) or very low gas utilization (some 4 % below average) or
very low hot metal temperature or silicon. Remedial actions are:
– Continue casting, high residual level can cause the furnace to get out of control.
– Decrease blast volume to the minimum where fuel injection can be maintained
on the furnace. Being out of control is always related to poor melting of the
ferrous burden and, by lowering the blast volume, there is more time for melting
and improved heat transfer from bosh gas.
– Lower basicity. A lower basicity makes the slag liquid at lower temperatures. A
lower basicity may be more important than the next point.
– Increase reductant rate by increasing coal injection (around 20 kg/tHM), while
preventing too low Tflame.
– Increase reductant rate by charging more coke. Charging coke blanks is to be
preferred over increasing coke rate, because it reaches the cohesive zone faster.
Most plants have ‘red button’ procedures to take these actions, which are
considerably more aggressive than the standard thermal control procedures,
to prevent such situations as above from occurring, if there are indications of
severe cooling. The objective is to provide a significant boost to heat in the bosh
gas, thus an immediate boost to heat in the lower zone, to bring the furnace
back to a safe state by replacing the lost reserve of coke below the melting zone,
and to reduce slag liquidus in case the hearth cools.
0.9
1480
Hot metal temperature (°C)
Hot metal Silicon (%)
1440
0.6
1400
0.3
1360
1320
0
0 12 24 36 48 60 72
Time (hrs)
Figure 10.1 Thermal state of a blast furnace during an upset (the blue line represents
the hot metal temperature, the red line hot metal silicon level)
186 Chapter X
10.3 What to control in daily operations?

The blast furnace must be constantly monitored to ensure that the fuel input
is correct and that the product is within the required quality parameters. Key
areas of control are:
1. Thermal control: temperature and/or silicon of the hot metal.
2. Control of slag composition, appropriate inputs to be selected for desired slag
composition.
3. Control of hot blast parameters and injection at the tuyere level.
4. Gas flow control by burden distribution
For some control parameters automatic control is the best (or only) possibility.
Examples are given in Section 10.8 of this chapter.
10.4 Thermal control

The thermal state of the furnace is adjusted on the basis of hot metal silicon
and/or temperature. Different companies use different methods, but in general
there are two casts below or above an acceptable range required in order to
make adjustments. Adjustments can be made by changing PCI/injectant rate
or coke rate at the top. If adjustments are based on two casts, average response
times are more than 7 hours if done with PCI and more than 12 hours if based
on charged coke weight. This is a slow reactive adjustment process.
Online and continuous hot metal temperature measurements are used to

speed up the thermal adjustments. Example 10.1 shows continuous hot metal
temperature measurements.
More proactive thermal control is based on the observation that thermal

changes are often caused by a change in direct reduction. Online monitoring
of direct reduction (or the heat balance over the lower furnace) can recommend
thermal control actions. For example: if direct reduction increases and the
burden descent is higher than expected on the basis of the oxygen input, the
furnace will cool down. There is no need to wait for the hot metal temperature
to decrease before taking appropriate action. So, immediate actions like
increasing the PCI rate by 10–20 kg/tHM (preferably at constant Tflame and
oxygen input through the tuyeres) can be taken. Systems that calculate the
heat balance over the lower furnace while on–line address the same subject
and provide very reliable results in preventing thermal upsets. Unfortunately,
companies express heat shortage in the lower furnace differently. Among the
indicators used are: direct reduction percentage, gas reduction percentage,
direct reduction in kg C/tHM and factor Ω (the difference between the ηCO on
FeO level in operation compared to thermodynamic maximum). In some cases,
the recommendations are implemented automatically.
Example 10.1 Continuous measurement of hot metal temperature

Hot metal temperature is generally measured with a thermocouple dipped in
the iron runner directly after the skimmer. The temperature measurement has
a delay to reach the value inside the furnace since the runner system has to get
warm. Especially when casting alternatingly on two sides, the runner cools
down for 2–2.5 hours between casts.
A relatively new development is continuous measurement (Figure 10.2).

This can be done by observing the hot metal temperature after the skimmer
by optical methods, but also by measuring with the optical methods right
at the taphole. A thermocouple in a refractory based tube, installed in the
center of the skimmer, can also be used, but has to be changed every few
weeks. If hot metal temperature is measured at the taphole, the correct hot
metal temperature is measured right from opening the cast. The temperature
difference with measurement in the skimmer is caused by hot metal and slag
cooling down in the main runner after closing the preceding cast.
1575
taphole
Hot metal temperature (°C)
1550
1525
skimmer
1500
1475
1450
9:23 9:34 9:44 9:54 10:04 10:15 10:23 10:33 10:44 10:55 11:05 11:15 11:24 11:34
Time
Figure 10.2 Example of continuous hot metal temperature at skimmer and taphole
at a 3200 m³ blast furnace
10.5 Control of slag basicity
The chemical composition of slag is controlled by varying the burden input in

order to reach the aim for slag basicity.
Changes have to be made:

– when raw material quality changes, for instance after receiving a new average
chemical composition of sinter;
– when raw material quantities are changed;
– when a stop is planned;
– when calculated and measured basicity have too large deviation;
– when hot metal silicon changes to a new level due, for example, to a significant
change in target production level.
188 Chapter X
Required basicity changes can be made by changing the charge weight with one
or more materials. For instance, gravel or quartz can be used to adjust basicity,
(more gravel gives a lower basicity), acid pellets can be exchanged with basic
sinter. Limestone or dolomite can be used as well. However, operators tend to
minimize use of these fluxes, since these materials have a penalty in coke use:
decarbonization costs around 200 kg coke per tonne of limestone and around
150 kg coke per tonne of dolomite). Basicity changes are in most cases done
once per shift and when needed.
10.6 Control of hot blast parameters and injectant

at tuyere level
The raceway conditions are: blast volume and oxygen rate, the hot blast
temperature, the moisture in the wind as well as the type and quantity of fuel
injection (coal, gas).
Blast volume
Blast volume and oxygen rate determine production. Blast volume is to be
decreased if:
– Blast pressure exceeds its maximum allowable value. Blast volume has to
decreased, preferably at a constant oxygen enrichment level. In order to keep
the gas flow through the furnace stable, top pressure has to be decreased as well.
Gas flow in the furnace is constant, if tuyere velocity is maintained.
– Burden descent is poor. A lower ΔP will help to stabilize burden descent. In this
situation top pressure has to be maintained at maximum.
– Production is too high. If for internal (casthouse) or external (steel plant)
reasons, the production level has to be decreased, blast volume has to be
reduced. In this situation it can also be effective for maintaining stable gas flow
in the furnace to reduce oxygen more than blast volume.
– Very cold furnace. If the furnace gets very cold, blast volume has to be reduced
in order to have more time for heat transfer to, and melting of, the ferrous
burden.
Hot blast temperature

Hot blast temperature is the cheapest method for getting energy into the blast
furnace. For this reason, hot blast temperature is normally maintained at a
maximum and there are few reasons to decrease it.
Moisture in hot blast.

Moisture in hot blast uses coke, and is normally kept to a minimum. It is an
efficient method of thermal control, but can usually only be used in the case
of a hot furnace. Since atmospheric moisture varies from day to night and
from season to season, some companies prefer to work with some steam added
to hot blast in order to maintain a constant moisture in hot blast. It can help
in keeping RAFT constant and can be used in case of a cooling trend. Some
companies use steam injection in hot blast only in the event of an emergency.
Coal and gas injection.

The coal and gas injection level is used extensively for thermal control. Besides
needing criteria when to put additional fuel into the furnace, there is also a
need for criteria to take it out again. Companies using co–injection of coal and
natural gas have the option of increasing coal injection at constant Tflame by
exchanging coal and gas injection (Geerdes, 2015).
10.7 Gas flow control
Since gas flows along the path of least resistance and the resistance is defined
by the ferrous layers, ferrous layers should be kept stable as long as possible.
This means that for smaller or larger changes in coke rate, the coke weight
has to be changed and the ferrous base kept constant. An illustrative example
showing a change in coke rate from 350 kg/tHM to 300 kg/tHM is presented
in Table 10.1. The ore base is kept constant and coke base reduced. Experience
has shown that relatively minor changes in burden distribution will be required
for optimization of the central gas flow (i.e. coke distribution). The burden
distribution adjustments can be applied as a second step if required.
Previous Ferrous base Coke base

constant constant
Coke rate kg/tHM 350 300 300
PCI rate kg/tHM 143 200 200
Coke base t 30.3 25.5 30.3
Ferrous base t 136 136 162
Table 10.1 Coke base change when PCI rate changes
For daily gas flow control, it is sufficient to increase or decrease the amount
of central coke. Indicators of too much central coke are a relatively inefficient
furnace (too much gas flow in center) and a low position of the cohesive zone,
as indicated by very low heat losses in the bosh. Too little central coke is
shown from peaks in the heat losses at stable burden descent, which can also
be observed from pressure taps showing the same pressure on various levels
(short circuiting of gas flow along the wall). Table 10.1 gives examples of burden
distribution control schedules. If more central gas flow is required, Coke 3
replaces Coke 2. Replacing Coke 2 with Coke 1 reduces central gas flow.
Position 11 10 9 8 7 6 5 4 3 2 1
Wall Center
Coke 1 Less central – 14 % 14 % 16 % 14 % 14 % 14 % – 6% – 8%
Coke 2 Normal – 14 % 14 % 14 % 14 % 14 % 14 % – 6% – 10 %
Coke 3 More central – 14 % 14 % 12 % 14 % 14 % 14 % – 6% – 12 %
Ore 16 % 16 % 16 % 12 % 10 % 10 % 10 % 10 %
Table 10.2 Bell–less top charging schedules with varying central gas flow
190 Chapter X
10.8 What should be controlled automatically?

There are numerous control loops in a blast furnace, but some controls cannot
be done in manual mode by an operator. There are numerous control loops
already automated. Here a few control loops are mentioned, where automation
may not be self–evident. Important control loops to be automated are:
– Adjustment of material flow gate opening. When discharging material via the
bell–less top, the opening of the material flow gate determines the flow rate. If
something changes in raw material quality, the flows change, which should be
automatically corrected. If not, the amount of coke charged in the center will
fluctuate strongly, affecting central gas flow and furnace behavior.
– Plugging the taphole. Movement of the clay gun into the taphole face,
compression and post–pressing can only be done systematically when
automated.
– Drilling the taphole. Drilling with a constant forward velocity requires
automated control.
– Blast volume and maximum ΔP. Ensuring that a blast furnace will not exceed
its maximum ΔP can be automated. In practice it may mean that the blower is
switched from (usual) volume control to (incidental) pressure control.
10.9 Burden selection
10.9.1 Burden composition international comparison
In expanding production facilities, especially in China, more and more locally

produced pellets are put into the burden. What is the optimum burden
composition? The answer to the question depends on the local cost situation.
Here we address the metallurgical possibilities. In a sense this section covers
many of the aspects in preceding chapters.
100 %
sinter
Asia, Europe,
South America
50 %
sinter
Individual
North America plants
Scandinavia
100 % 100 %
pellets lump
Figure 10.3 International comparison of burden composition

The total burden consists of sinter, pellets and lump ore in varying fractions as
shown in Figure 10.3. The sinter content in the burden varies from no sinter to
90 %. Pellet content can reach 95 %. Lump ore has an acid gangue consisting
of SiO2 and some Al2O3 and is used mostly up to 10 % of the burden with few
exceptions of up to 40 %.
With increasing pellet content, the choices that have to be made are:
– Acid or fluxed pellets, for acid pellets there exists a high MgO variant.
– What basicity of sinter has to be made.
– The type and amount of lump ore.
Optimization requires that the quality of the materials is suitable for high
productivity and low coke rates. To this end fluxes have to be incorporated
within the blast furnace burden materials and not charged directly into the
furnace as limestone or dolomite.. As a next step costs have to be optimised.
It should be clear from this book, that softening and melting behavior of
the ferrous burden is critical for efficient operation of the blast furnace. The
problem is, that softening and melting of blends is not a simple average of the
softening of components. The question of burden optimization in this section is
addressed by discussing the following topics.
– Reduction of burden materials in the “dry” area of the blast furnace, that is
before softening.
– Softening and melting of pellets and sinter.
– Softening and melting of blends of pellets and sinter.
10.9.2 Gas composition in the furnace

Reduction of each blast furnace burden component is characterized by
its “reducibility” (Chapter III). However, since the blast furnace burden is
processed relatively slowly, a final stage of reduction prior to softening of the
burden is reached, which is (almost) independent of the reducibility of the
various materials in the early stage of reduction
Gas composition in a blast furnace can be monitored by vertical sampling in

the furnace, Figure 10.4 on the next page. Temperature is measured with a
thermocouple, gas is sampled and analyzed outside the furnace. Temperature
and gas composition show different behavior: at the wall the temperatures are
lower and ηCO is higher.
The figure shows the thermal and chemical reserve zone in the upper and lower
graphs, especially in the sample 0.3 meters from the wall: for 120 minutes
the gas temperature is 900–1000 °C and ηCO is around 30 %. There is a large
difference in gas composition versus time between various radial positions.
192 Chapter X
Figure 10.4 Vertical probes measuring temperature and gas composition in an

operating blast furnace with 2678 m³ IV, average of several measurements
(Chaigneau et al, 2001)
However, the two graphs in Figure 10.4 can be combined to show how
ηCO varies with temperature as shown in Figure 10.5. For the various radial
positions, the relationship temperature–ηCO is identical, where between
1000–800 °C the line corresponds with chemical equilibrium for the reduction
of wustite. The conclusion is that gas composition ηCO is only dependent on
temperature.
100
80 Fe3O4
60
FeO
ηCO
40
0.3 m 1.3 m 2.1 m
from wall
20
Bosh gas
Fe
0
0 500 1000 1500
Temperature (°C)
Figure 10.5 Vertical probes measuring temperatures and gas composition showing the
relation temperature–ηCO, the arrow indicates bosh gas entering into the
system
10.9.3 Reduction of the ferrous burden

The preceding section was about gas composition inside the furnace. What
about the reduction of the burden materials? The behavior of the burden
materials is analyzed in various reduction–softening–melting tests, which vary
greatly in purpose and methodology. Here the behavior of burden components
and the total burden is described.
The behavior was studied by reducing, softening and melting burden

component and the complete burden in a test rig. A sample was reduced with
a predefined gas composition/temperature pattern. Figure 10.6 shows the
reduction behavior of fluxed sinter (B2 = 1.9), acid pellets and two types of
lump ore. In the early stage of the reduction there is a large difference in the
reduction of the raw materials. Acid pellets and lump ore 2 are reduced fast,
sinter somewhat slower and lump 1 lags behind in reduction. However, as soon
as the temperature approaches 950 °C, the reduction of various materials is
equal and the reduction reaches the wustite level (FeO1.05).
The burden (60 % sinter, 20 % acid pellets, 10 % lump 1, 10 % lump 2)

showed a weighted average behavior up to that temperature. The conclusion
is that various types of materials show the same degree of reduction in the
range of 950–1100 °C, even poorly reducible material like lump ore 1. At
higher temperatures, differences are observed due to differences in softening
temperatures, where the reduction rate of sinter is highest due to its higher
softening temperature.
1.50 sinter
1.25 acid pellets

lump 1
1.00 lump 2
O/Fe
burden
0.75
FeO
0.50 Fe3O4
0.25
0
400 600 800 1000 1200 1400
Temperature (°C)
Figure 10.6 Reduction of fluxed sinter (B2 = 1.9), acid pellets and two types of lump
ore in a reduction test (Brass test, data from Chaigneau et al, 2005)
10.9.4 Softening and melting of sinter, pellets, lump

Softening temperatures are the temperatures where burden or its components
tend to collapse. It can be measured from bed shrinkage or increase in gas
resistance. No standard tests are available, but an impression of softening
temperatures for various materials is provided in Figure 10.7 on the next page.
194 Chapter X
From high softening temperature to low:

– Super–fluxed sinter, softening at temperatures above 1330–1400 °C. After
a melting test with a temperature above 1500 °C super–fluxed sinter is not
completely molten due to the slag formed by its gangue.
– Fluxed sinter.
– Fluxed pellets.
– Olivine pellets.
– Acid pellets.
– Lump ore: lump ores have a very wide range of softening temperatures from
below 950 °C to above 1300 °C due to variability in chemical composition and
physical properties.
1500
Softening temperature (°C)
1300
1100
900
sinter sinter fluxed olivine acid lump
(B2 > 2) (B2: 1.6–1.9) pellets pellets pellets ore
Figure 10.7 Softening temperatures of various types of burden materials compiled

from various sources (Chaigneau et al, 2001, 2005; Liu et al, 2018,
Kemppainen et al, 2015, Etchevarne et al, 2018)
10.9.5 Softening of the burden as a blend of sinter, pellets, lump

Softening and melting of a blend is different from the softening of the
components. As soon as material starts softening in a blast furnace, the internal
structure collapses. However, the stronger material prevents the softening of the
weaker materials by taking away the physical pressure from the burden on top.
3.0
(mm water gauge/mm height)
sinter
2.5
lump
2.0 60/40 blend
weighted
ΔP
1.5 blend
1.0
0.5
0
900 1000 1100 1200
Temperature (°C)
Figure 10.8 Softening of sinter and lump and its blend (60/40) as tested with a
softening/melting test (Chaigneau et al, 2005)
The example of Figure 10.8 shows a sinter/lump ore blend (60/40 %), where
sinter as the stronger material forms a skeleton. It prevents collapse of the
cohesive zone and keeps the structure open and permeable for gas promoting
reduction of the blend. A blend has softening properties similar to the stronger
burden component. The unexpectedly good softening properties of blends
compared to weighted averages were observed at numerous research projects.
10.9.6 Optimizing burden: sinter basicity and pellet type

Requirement for additional iron units comes especially from Chinese iron and
steel industry, where additional pellets are produced. What type of pellets can
be considered? Acid, acid/olivine or fluxed pellets and what sinter basicity?
What are advantages and disadvantages?
Very high pellet rates (above 80%) require a proportion of fluxed pellets. This is
the situation in North America and Scandinavia and specific plants elsewhere.
High sinter rates (above 70%) can be combined with acid pellets, because it
brings the sinter basicity into a good range. Sinter basicity is important for its
metallurgical properties and also for sinter plant productivity. For high pellet
operation in the range of 40–65 % pellets and 30–50 % sinter, options are:
– Acid pellets/super–fluxed sinter
– Olivine pellets/super–fluxed sinter
– Fluxed pellets/fluxed sinter
An analysis of these options starts with the mass balances, then the effects of
costs have to be evaluated. It is clear from operating experience that the three
options above are feasible, since companies with excellent performance for all
variations exist. A simplified series of mass balances is based on starting points
shown in Table 10.3. For optimization the slag, B3 (CaO+MgO)/SiO2 was kept
constant. The table also shows the scenarios for the use of acid, olivine and
fluxed pellets.
kg/tHM CaO/SiO2 Fe SiO2 CaO MgO Al2O3 B3

pellet acid 785 0.30 65.0 4.0 1.2 0.4 0.6
sinter 785 2.15 55.7 5.5 11.8 1.8 1.5
slag 247 1.19 35.4 42.0 7.3 12.3 1.39
pellet olivine 783 0.08 64.6 4.0 0.3 2.0 0.4
sinter 783 2.09 56.2 5.5 11.5 1.4 1.5
slag 244 1.08 35.7 38.5 11.1 11.7 1.39
pellet fluxed 784 0.93 63.0 4.0 3.7 1.0 0.3
sinter 784 1.66 57.9 5.5 9.1 1.4 1.5
slag 245 1.17 35.8 41.8 8.0 11.4 1.39
Table 10.3 Comparison of pellet burdens and consequences for sinter composition.
Starting points: 50/50 sinter/pellet burden, B3 = 1.39, pellets contain 4 %
SiO2, sinter ferrous content corrected for CaO content, 300 kg/tHM coke
rate and 200 kg/tHM coal rate.
196 Chapter X
Sinter basicity for a 50/50 sinter/pellet ratio varies from 1.66 when using fluxed
pellets to 2.15 when using acid pellets. Optimum slag MgO content is 8–10 %
and all types of slag are acceptable.
What to choose? Most operators prefer a similar basicity for sinter and pellets
and prefer to use fluxed pellets when the sinter rate decreases to below 60 %.
There are, however, plants operating with super–fluxed sinter and acid pellets.
Olivine pellets are metallurgically improved acid pellets. The advantages of the
super–fluxed sinter with olivine or acid pellets are:
– Increased productivity of sinter plant (Chapter III) as well as pellet plant, since
production of fluxed pellets costs energy for the decomposition reaction of
carbonates driving off CO2.
– Acid pellets have a relatively low softening temperature, which can be improved
by producing olivine pellets.
– Good permeability of the cohesive zone, since super–fluxed sinter softens at
high temperature.
10.9.7 Burden distribution for high pellet operation

A high pellet percentage also requires consideration of burden distribution.
The major effect of a big change in pellet percentage is the effect on the angle
of repose: pellets have a shallower angle of repose than sinter. If a burden
distribution is used that is insensitive to angles of repose, like the ideal burden
distribution in Chapter VII, then the furnace will accept a different pellet
percentage. If however, burden distribution makes use of angles of repose, like
the burden distributions used for double bell furnaces in the past, which are
sometimes known as “platform–V”, then changes in pellet percentage will be
difficult to accept, since the increase in pellet% will lead to more pellets rolling
into the center. This will block central gas flow. In such a situation the furnace
will have to be operated on lower productivity and will operate with higher heat
losses and lower PCI rates. This is explained in Figure 10.9.
Figure 10.9 Left: effect of increasing pellet percentage on the ore free center, which
is filled since pellets have a shallower angle of repose (ideal burden
distribution as in Chapter VII for comparison)
XI Operational Challenges
11.1 Raw materials from stockyard: fines and moisture
The raw materials for an operating furnace are ideally freshly produced, skipped
“hot” to the blast furnace and fulfil all quality requirements. Blast furnace
operation, however, is often confronted with variability of the materials input.
The percentage of various burden materials like sinter, pellets and lump has to
change, or part of the materials comes from the stockyard. It has to be pointed
out, that stockyard materials have different properties from freshly produced
materials, especially with respect to the amount of fines and moisture. If the
operator uses the ideal burden distribution as pointed out in Chapter VII, then
the furnace is more robust in coping with raw material variation. Nevertheless,
an operator should take appropriate measures if confronted with second
grade raw materials. For instance, if a large percentage of sinter from stock is
being used, the furnace cannot be expected to continue to run at maximum
productivity, and timely reduction of blast volume is indicated. A large load of
fines into the furnace may also come from direct charge materials like pellets
when finishing a stockpile, or from very wet lump res, which are impossible to
screen properly. In the present section, the effects of fines on the ore burden are
discussed, as well as high moisture input, for instance caused by local heavy
rain.
11.1.1 Fines in ore burden

For the blast furnace operator, one of the main raw material concerns is that
of the fines loading. This is the proportion of undersize material that is in
the furnace. The direct effect of a high fines loading is that it will affect the
permeability of the furnace. The permeability of the ore layer is determined by
the amount of fines (smaller than 5 mm) in that layer. Unfortunately, when
bulk material is handled, fines are generated. For this reason, coke and ore
burden are normally screened before being charged into the furnace. But, for
example, when emptying a pellets stockpile or using sinter from stockyard,
high fine loadings cannot be avoided. Especially if the material is wet, proper
screening is difficult or even impossible. Wet lump ores can carry huge amount
of “piggy–back” fines that adhere strongly to the larger particles and cannot be
screened. In Chapter VII it was already noted that fines should be distributed
over the furnace radius by using a large number of chute positions (7 or more)
for the ferrous burden.
198 Chapter XI
Fines tend to segregate. When material is put into the stockpile the fine material
remains on the point of impact and the coarser material rolls outwards, the
size segregation phenomenon. This effect is seen wherever granular material is
handled. So, when reclaiming material from stock, it is important to avoid high
amounts of fines being reclaimed and sent to the furnace without screening.
Similar segregation can take place while charging the furnace, and can impact
the furnace process. Fines in general are undesirable, due to the blocking of the
spaces between the larger particles, however due to the flow characteristics of
fines, they can also deposit preferentially in certain areas. The impact of this is
particularly noticeable with bell–charged furnaces, where the fine particles will
drop directly down onto the stockline, and the large particles will flow a little
more outward and deposit at the wall (see Figure 11.1). If material hits the wall
before it reaches the burden level, the fines will accumulate close to the wall and
the coarser material will flow inwards more.
This segregation effect also occurs when filling a bunker. Be it in the stockhouse
or on the bell–less top, segregation will always take place. When material is
charged into a bin, the fines remain on the point of impact, the coarse material
runs “downhill”. If material is required from a bunker, it starts to deliver the
material that is located in the center: this being the fine materials, while later
the coarser materials from the sides of the hopper begin to flow. Finally, also the
chute segregates coarse and fine materials with fines on the inside of the falling
curve and coarse on the outside (Figure 11.1).
Figure 11.1 Segregation of fines during charging with a bell and bell–less top
charging system. Right figure: coke pushed towards center.
A concentration of fines close to the wall can have a negative effect on the
reduction and melting of the ore, as it forms a blockage for the hot reducing
gases. When the fines reach the cohesive zone and melt, the remaining oxygen
is removed by direct reduction. This results in a higher fuel rate or cooling of
the furnace. If the material does not melt in due time before reaching the tuyere
zone, the “non–molten” material is observable as scabs through the peepsights.
Operational Challenges 199
Note that there is a difference between the path travelled by the coarse materials
and fines. When the burden descends through the furnace, the fines fill the
holes as soon as they are formed, while coarse materials follow the wall. Fines
travel more vertically and faster towards the cohesive zone (see Figure 11.2). It
is possible to deflect the fines with a bell top arrangement by using the furnace
movable armor as a deflector, and with a bell–less top by charging from the
outer to the inner position.
Figure 11.2 Fines charged at wall migrating through the furnace and appearing as
‘scabs’ in front of tuyeres
Stockhouse management
An additional source of fines that can be avoided through slight modification
in stockhouse practices is that of bin management. The bin fill levels may drop
if there has been an interruption in the supply of materials. Note that bins are
not to be used for logistic optimization; bins have to be full above 80 % all the
time to prevent degradation. Nevertheless, occasionally the bin level may have
dropped so low as to require a stop of the blast furnace, or supply may have
resumed in time for the blast furnace to remain in operation. In either case
the refilling of the bins should be managed to avoid excessive breakdown of
the material as it falls a longer distance. It may be tempting to try to increase
the bin levels at all of the bins by filling them all up gradually. However, this
method will result in more material falling from a greater height as the bins will
continued to be emptied as they are being refilled. It is preferable to work with
fewer bins, so that the levels can be increased more quickly, even though they
are being emptied at the same time. Once the bin levels are up at 80 or 90 %,
the excess feed material can be diverted to a low–level bin, which remains out–
of–service. The fuller bins are then topped–off when they drop to their usual
target level of 65 to 70 %, giving some head–room for the supply to be diverted
to the less full bins (Figure 11.3 on the next page).
200 Chapter XI
Refilling the bins by this method will reduce the amount of material that has
to drop from a greater height, as there will be less transfer time and more filling
time, and the bins that are not being used while they are being filled will fill
more quickly. Once each bin is back to normal fill levels they should be brought
into operation. This will result in each bin coming back into operation one at
a time. This staggering of the restart of the bins will avoid the material that
has been dropped the furthest distance in each bin being charged to the blast
furnace at the same time.
Figure 11.3 Effect of staggered restart on stockhouse bins: by using less bins
11.1.2 Moisture in coke and burden
Moisture input and elimination

The moisture charged into the furnace with the coke and ore burden must be
removed before the process can start. This takes place in the upper part of the
furnace. The amount of water with the charge at 1 % ore burden moisture and
5 % coke moisture is about 33 kg/tHM of water. The center dries very quickly,
but in the wall area it can take much longer – up to 40 minutes, as shown in
Figure 11.4.
Figure 11.4 Temperatures and height in a blast furnace, the red dotted line indicates
“dry” 100 °C
If the moisture input increases, it will take longer for the material to dry, and
the isotherm where the reduction process will start will descend downwards.
As a consequence, gas reduction will be less efficient. More oxygen will remain
bound to iron, and this oxygen has to be removed by direct reduction in the
lower part of the furnace. This consumes coke which was not charged and cools
the furnace.
Most companies are equipped with moisture gauges for coke bins, so that
variation in moisture input in coke is compensated to ensure that the dry coke
rate is constant. Note that this is only a minimum correction to maintain
the current thermal state. If the furnace is already operating with low top
temperature (below 100 °C) the compensation with coke moisture gauges will
not be sufficient to compensate for the decreased efficiency of the reduction
process.
Where moisture is added in place of coke, the furnace cools and so the normal
thermal control procedures will be activated, usually calling for additional fuel.
If the moisture level then reduces again, the furnace will warm up, triggering
another set of actions. If this is allowed to continue, the furnace will enter a
thermal cycle that will in turn consume more fuel than required. Extreme coke
moisture variation typically occurs when the coke supply is switched from wharf
coke to stockpiled coke that has absorbed more moisture due to rain or a storage
field that is not well drained. If coke moisture readings are not correct, it can
cause large swings in the thermal state of the furnace. Another cause of major
swings in moisture input is the use of water sprays to prevent dust emissions.
This effect is just as important with pellet moisture, especially where pellets
have been shipped or stored under damp conditions. They can contain up to
6 % water. When a batch of these pellets is charged to the furnace, the top
temperature will decrease with the additional moisture. The furnace will start
to warm up due to the fact that a lower amount of iron is being charged to
202 Chapter XI
the furnace with the higher moisture pellets, provided that there is sufficient
gas available to eliminate the water. Coke rate changes will normally be made
to correct for this warm up. However, once this batch of wet pellets has been
consumed it is very important to realize that the furnace will then cool down,
due to the additional iron that is being charged with low moisture pellets. If
this is not anticipated, the furnace can cool down very quickly, so it is better
to anticipate this change by increasing coke rate when it is known that the
wet pellets have been consumed and dry pellets are soon to arrive. Some of
the larger blast furnaces also have pellet moisture measurement instruments
installed on the stockhouse bins.
Ideally, coke and pellet moisture gauges can be installed to monitor and correct
for any changes on–line. These moisture gauges take regular readings of the as–
charged moisture levels for coke and pellets and will make corrections for the
weight, so that the required quantity of the material is charged.
The recommended approach is that the top temperature is not allowed to fall for
a prolonged period (8–16 hours) below dew point temperature. Some companies
are able to run the top gas temperature at low average levels, well below 100 °C.
In these situations, monitoring the temperatures in the wall area (3–5 meters
below the burden level) to determine whether or not the burden is dry ‘on time’
is recommended.
Moisture condensation and recirculation

Since burden materials are charged well below the dew point of the moisture
in the furnace gas, the moisture in the gas will condense on the colder parts,
frequently the cooled staves and freshly charged materials. This applies not only
to the moisture which is charged (typically 30–40 kg/tHM), but also to the
moisture which is generated in the process. There is more water generated when
there is more hydrogen produced from the injectants at the tuyeres. At a coal
injection level of 200 kg/tHM and 5 % hydrogen in coal, about 36 kg of water
per tonne of hot metal comes from the coal and has to be eliminated with the
top gas. The amount of water generated from natural gas is even higher: at an
injection level of 80 kg/tHM, about 65 kg water comes from the natural gas. All
the water has to be eliminated with the top gas.
If this is not the case, a condensation–evaporation cycle will start. Evaporation

will take place more towards the center and lower down in the furnace, while
condensation takes place closer to the wall as shown in Figure 11.5. In the
course of this cycle, the isotherms where the reduction process starts will
descend lower and lower in the wall area of the furnace, reducing the efficiency
of the process and creating a risk of tuyere failure and upsets. This situation can
be very serious if water starts to flow between the staves and shell downwards.
Skulls can be created on the furnace wall, and water has been observed coming
from the tuyere seals and even the taphole. Remedial actions have to increase
the gas volume per tonne of hot metal, since water is eliminated as a vapor.
For instance, extra coal can be injected. Note also, that the water elimination
depends on the top pressure, since at a lower top pressure the top gas can
contain more water vapor at the same temperature. Small furnaces with low top
pressure are much more forgiving for high water loads than large, high–pressure
furnaces.
Figure 11.5 Effect of evaporation–condensation cycle of moisture, coming from

charged materials and generated in the process. The gas flows from inside
out reaching lower temperatures and water will condense. The 100 °C
isotherm is indicated in white. The mechanism proceeds as follows:
Phase 1 – Normal operation
Phase 2 – When the water input is high, vapor condenses where the gas
temperature is below 80 °C; this ccurs in the wall area, especially on the
freshly charged raw materials. The 100 °C isotherm is driven downwards.
Phase 3 – As the evaporation–condensation cycle continues, the 100 °C
isotherm is driven further and further downwards. The metallurgically
active area near the furnace wall becomes smaller and smaller, impeding
gas reduction in the periphery. Direct reduction levels increase, the
furnace becomes inefficient with low ηCO.
11.2 Burden descent
11.2.1 Variability of charging and production rate

Most operators observe the charging rate in a furnace as defined by the amount
of charges put into the furnace per hour. If the charging rate increases while
tuyere conditions are unaltered, the furnace will fall short of heat. Simply put,
with the same amount of heat and gas produced at the tuyeres, more hot metal
is made, so the furnace will chill. The reasons for this happening can be various.
– fuel shortage as a consequence of an increase in direct reduction,
– too low coke input for instance by incorrect compensation,
– too high input of ferrous material (e.g. when changing from ‘wet’ pellets to dry
pellets),
– by changing process conditions.
204 Chapter XI
Example 11.1 Monitoring drying capacity and remedial actions

The drying capacity of the gas determines how much water can be eliminated
from the furnace. How to determine whether or not the burden has become
“dry” sufficiently quickly? This can be done by
– Average top temperature, which is a relatively insensitive measurement.
– Skinflow measurements (skinflow are temperatures a few centimeters inside
and 1–5 m below the stockline).
– Short in–burden probes. A temperature measurement around 60 cm inside
the hot face, around 3 m below the stockline, Figure 11.6.
A statistical analysis has shown that the short in–burden probes are the best
tool (Geerdes, 2018).
An operational example is shown in Figure 11.6. Heavy rain caused an

increase in moisture input, which caused top temperature and in–burden
probe temperatures to decrease. The operator reacted by reducing oxygen
enrichment by 2 %, which also resulted in a temporary reduction of
production by 7 %. As a consequence of the actions, the process remained
stable (ηCO, Silicon) and was normalized as soon as the rain stopped.
Figure 11.6 Short in–burden probe (left) and monitoring during heavy rain, while
operator reduced oxygen enrichment by 2 % in order to keep probe
temperatures above 300 °C
Here we refer to increased direct reduction. In some situations, the gas

reduction of the burden does not progress sufficiently. This can be caused by
– Too much water input, lowering the isotherms within the furnace and
shortening the process height of the furnace, especially at the wall.
– Large amount of fines charged into the furnace resulting in poorer pre–
reduction and increased direct reduction.
– Irregular burden descent, causing mixed layers.
– High liquid residual level which affects the normal gas flow through the
burden.
– Charging delays causing the newly charged material to see shorter process
height and altering burden distribution.
The resultant material with insufficient pre–reduction will in any case continue
to descend to the high temperature region above the tuyeres. When this
material starts melting, all the oxygen will participate in direct reduction.
This consumes coke, makes the furnace descend faster and drives the cohesive
zone downwards. This is a self–propagating effect, and when the cohesive zone
reaches the tuyere level, it will chill the furnace within hours.
Experienced operators equipped with the right tools can observe the increased
direct reduction long before the casthouse gives warning of low hot metal
temperature. The method to correct the incident is to bring the cohesive zone
back to its previous position. This can be done with extra fuel injection and/or
lower blast volume, and by maximizing heat input into the furnace (maximum
hot blast temperature and no blast moisture) or, in serious situations, by
charging extra coke from the top.
An example of the use of direct reduction as an early warning signal for a

cooling trend in the furnace is shown in Chapter VI. The earliest signal of a
cooling trend comes from the increase of direct reduction followed some time
later by an increase in charging rate. The hot metal silicon reacts a few hours
later, because the colder material has to percolate through the hearth.
11.2.2 Hanging and slipping

The burden descent sometimes becomes erratic (see Figure 7.5). What happens
in the furnace if it hangs and slips? The mechanism of hanging and slipping is
illustrated in Figures 11.7–11.9.
First, the furnace hangs because at the cohesive zone, bridges of melting ore
burden are formed. “Bridge formation” is the phenomenon where solid materials
can be piled upon each other and will not collapse into a hole: see Figure 11.7
for a bridge formed from marbles.
Second, while the furnace hangs, the process continues: coke is consumed and
ore burden melts. Voidage then arises in the active coke zone, which is below
the cohesive zone.
Figure 11.7 Bridge formation illustrated by a theoretical experiment with marbles

206 Chapter XI
Figure 11.8 Creation of voidage below bridges and consequential collapse
Third, when this voidage becomes too big, it collapses: the furnace burden
slips (Figure 11.8). The layer structure is completely disrupted and the gas flow
through these layers is impeded. This leads again to areas in the furnace where
ore burden is insufficiently reduced and remains in a cohesive state for too
long. These areas will form the bridges for the next time the furnace hangs.
The problem can only be solved by re–establishing the layer structure within
the furnace, which means that the complete content of the furnace has to be
refreshed: the furnace has to be operated on reduced blast volume for five to ten
hours to prevent a continual cycle of hanging and slipping.
After a slip, the layer structure in the furnace is disrupted, impeding the contact
between gas and burden (Figure 11.9). As a consequence, the gas reduction
reactions slow down, and extra direct reduction will take place in the hearth:
the furnace will chill. The process will recover when a normal layer structure is
restored. It takes 6–8 hours to refill the furnace on a decreased wind volume.
Figure 11.9 Disrupted layer structure and impeded gas flow

11.3 Channeling
Channeling in a blast furnace is a local, preferential gas flow through an area
with the lowest gas resistance. Channeling makes a blast furnace less efficient.
Channeling is often observable from sudden fast drops in ηCO. An example is
shown in Figure 11.10, where ηCO decreases from 49 % to 36 %. This means
that a large percentage of the bosh gas passes through the furnace unused.
Channels normally come and go in a few hours. As soon as a channel is formed,
the gas finds a preferential path and blast pressure drops as well. Severe channels
can lead to drops in ηCO of 5–10 % or more, as in the example.
The major impact of channeling is the loss of chemical and physical energy in
the gas. As an example, a sustained reduction of 3 % in ηCO and increase of 100
°C in top gas temperature was seen to reduce the hot metal temperature by 80
°C. The increase in top temperature is visible in the example as well.
Figure 11.10 Typical example of channeling
Most of the channeling starts along the wall of the furnace. Because of the
inverse conical shape of the shaft, a gap is easily formed when burden descends.
Channels have a self–enhancing effect: a strong gas flow blows away materials,
especially fines, and prevents materials from collapsing in the channel. Gas
flows can be so strong that fluidization starts. Gas flow through channels
along the wall cause the staves to heat up. This means that channels can also
be observed from local heat losses and/or stave temperatures. Pressure taps
can show a very small pressure difference at different levels, which can be an
indication of a channel.
Channels are caused by the difference in permeability around the circumference

and along the radius. Experience shows that channels can be caused by irregular
burden descent, since fast descent creates gaps between wall and burden.
Irregular burden descent is caused by uneven melting of the cohesive zone.
208 Chapter XI
This can be caused by

– Concentration of fines charged or created by reduction disintegration, since
fines create an area with high resistance to gas flow. Channeling frequently
coincides with charging stockyard materials.
– Drainage of primary slag because of poor melting. This can be caused by
insufficient compensation for direct reduction or a high local basicity.
– Formation of scabs in the cohesive zone in the wall area.
– A special situation is that a charge of 100 % pellets at the wall may cause
the pellets to glue together, and this leads to channeling. Lump ore with low
melting temperature can behave similarly.
– Unstable burden descent: there are many reasons for unstable burden descent,
such as poor casthouse operation and operating the furnace at too–high blast
pressure.
Remedial actions for channeling are the following.

– In the very short run an operator will try to cause the channels to collapse. In
the first instance by checking the furnace at the end of a cast. Checking may fill
the gaps. If the channel reappears after the checks, then the furnace has to be
operated on a lower blast volume.
– In the long run, the occurrence of channels has to be prevented by having
stable burden descent. Channels can be prevented by using the ideal burden
distribution as in Chapter VII. Proper blending is essential as well as preventing
the formation of thick, cohesive ferrous layers.
11.4 Recirculation of alkali and zinc
In furnaces operated with a central gas flow, the top gas temperatures in the
center increase to such a level that part of the alkalis and all the zinc leaves the
furnace as a vapor with the top gas. If top gas temperatures are low, the alkalis
and zinc may accumulate in the furnace. The zinc normally condenses on the
refractory.
11.4.1 Alkali in the blast furnace

Alkalis, which are sodium and potassium salts, enter the blast furnace with the
ferrous burden and coke and coal. The alkalis takes part in a cycle of reduction–
oxidation as well as vaporization–condensation.
The alkali input in a blast furnace is typically 1.5–5 kg per tonne of hot metal.
In local situations, where companies have access to local ores with relatively
high alkali content, the input can reach higher levels. In this situation, high
productivity is normally no longer possible, but the competitive edge is gained
from the cost advantages of the use of cheap local ore.
From measurements of chilled blast furnaces, it has been observed, that the
amount of recirculation is about 3–10 times the input.
Alkalis in the blast furnace have various effects on operation. From a chemical
point of view, alkali acts as a catalyst for solution loss (C + CO2 2 CO). This
means that, at high alkali input, the fuel rate is slightly higher. This effect is
generally rather small; reported values are 6–11 kg coke per kg alkali in the
burden. The additional fuel can be added by increased PCI, leading to more top
gas energy from the furnace.
A second effect is, that degradation of coke and ferrous materials is promoted,
which leads to poorer permeability and can affect productivity at high
production rates.
Moreover, alkali can lead to the formation of scaffolds in the furnace: solid
material adhering to the wall in the stack of the furnace. The effect of the
formation of scaffolds is that the burden descent deteriorates, in extreme cases
leading to hanging and slipping. High levels of alkali have also been observed in
scabs, formed from cohesive material in the bosh/belly area.
Finally, the refractory materials, especially the carbon–based refractories, can

be attacked by alkali, affecting campaign length. For this reason, alkali control
in the furnace and prevention of scaffold formation require attention from the
blast furnace operator. Blast furnace operators should monitor alkali and zinc
loading and removal to have an idea of the recirculation of these undesirable
elements inside the furnace or calculate percent removal.
Sodium and potassium

The two types of alkali, sodium and potassium, behave differently. Sodium
is more easily removed from the furnace with slag as well as via the top gas.
Moreover, in most furnaces the potassium input is higher than the sodium
input. So, for our purposes it is sufficient to observe the potassium balance.
Only in situations where the sodium input is higher than the potassium input,
do both have to be observed. The alkali leaves the furnace mainly via the slag.
Some of the alkali leaves the furnace via the top gas and top gas dust.
Recirculation of alkali is shown in Figure 11.11 on the next page. Alkali

entering the furnace as silicates are first dissolved in the primary “melt” formed
in the cohesive zone. This melt has a very high level of FeO. As soon as the iron
oxides have been reduced out of the primary slag the final slag is formed. As
indicated below, the alkali can be reduced to elemental sodium and potassium,
which do not dissolve in hot metal or slag. Sodium and potassium are gaseous
and ascend with the gas in the furnace. In the high temperature area (around
1400 °C), cyanides are formed that bind with the alkali and ascend with the
gas. As soon as the oxygen potential rises, the cyanides are no longer stable and
disintegrate to alkali oxide and alkali carbonate (around 1100 °C). The material
is absorbed into the solid material (coke and ore burden) and starts descending
again with the melting ore burden. The carbonates will dissociate to CO2
and alkali oxide again. The alkalis recirculate in the area where the ore starts
softening (around 1200 °C) to where the slag is free of FeO (1400–1450 °C).
210 Chapter XI
Figure 11.11 Recirculation of potassium (a similar cycle applies to sodium)
The retention of alkali by slag is influenced by the slag basicity, slag composition
and the temperature of hot metal and slag. The lower the basicity, the more
easily the basic K 2O and Na 2O are removed. The lower the temperature, the
better the alkali absorption capacity of the slag.
In operational practice this has interesting consequences. If a blast furnace is

stopped, the slag basicity is normally decreased by 10–20 % in order to restart
the furnace with a slag with lower liquidus temperatures. After the restart, the
slag basicity is often low and the temperature is low. So, analyzing the K 2O in
slag before and after a stop, the operator can get an impression of whether or
not a high amount of alkalis is circulating in the furnace. The slag composition
(especially MgO and MnO content) also affects alkali removal capacity.
Some alkali is removed with top gas. The top gas temperature varies over the
diameter of the furnace throat. Furnaces operating with a central gas flow have
high temperatures in the center (average > 600 oC). The alkali is removed with
the central gas flow. It is generally accepted that operation at high productivity
and low coke rate (high PCI rate) requires a central gas flow. This facilitates the
elimination of alkalis.
11.4.2 Zinc
Zinc enters the blast furnace via the sinter, mainly through recycling of blast
furnace top dust, blast furnace sludge and dust from the secondary dedusting.
Interestingly, a relatively large part of the final zinc input comes indirectly from
the type of scrap the steel plant is using, especially scrap from galvanized steel.
Zinc enters the furnace as an oxide, silicate or ferrite. It is reduced in the lower
part of the furnace to elemental zinc. Since it has low melting and boiling
points (419 and 907 °C), it ascends through the furnace as a vapor and is then
reoxidized by CO2, condenses to burden and coke and participates in a similar
recirculation pattern as potassium. Zinc is eliminated from the furnace with
the dust from the top gas, especially from the gas at the furnace center. Zinc
adheres to the surface of particles. As a consequence, the finer the particles, the
higher the zinc content. This means that zinc can be recaptured by segregating
the fine fractions in the wet sludge. The fines can be upgraded to higher zinc
concentrations in a specialized blast furnace or a rotary hearth furnace.
Zinc oxide can condense on the furnace walls and on/in the coke lumps. Zinc
can form scaffolds in the furnace, thus leading to deteriorating operational
results. Occasionally the scaffolds melt when a tuyere is changed and liquid
metal (zinc) drops from above. When zinc is deposited in the pores of the
ferrous materials, it can impede the gas reduction, leading to higher reductant
rates. Zinc also deposits in carbon–based refractories causing them to become
brittle and crack, which reduces lining life. Zinc control is mostly done by
controlling input levels. Typically, 100–200 g/tHM is used, but higher values
for specialized operations are also known.
11.5 Adherences to the furnace wall
Solid material can adhere to the furnace wall. Scaffolds consist of materials that
adhere to the wall, while scabs are materials that are not yet completely molten.
Different locations for adherences are shown in Figure 11.12.
Figure 11.12 Positions of scaffolds in blast furnaces (photograph shows upper furnace)
Scaffolds in the upper part are often caused by high alkali loading, high zinc
loading, high fines input and low top temperatures. A number of companies
have experienced high alkali input due to their local raw material supply.
212 Chapter XI
Methods have been developed to control the amount of circulating alkali in the
furnace by various methods, like coke charging at the wall, cleaning operations
at low basicity and/or low burden level. A company operating at alkali levels up
to 4.5 kg/tHM (2–2.5 kg K 2O/tHM) controls scaffold formation by:
– Eliminating as much fines as possible from the burden. In doing so the chance
on scaffold formation becomes very much lower.
– Cleaning action. As soon as indications of scaffold formation are manifest from
the unstable burden descent and high pressure, a cleaning action is carried out.
This cleaning action consists of lowering the burden level to 14–15 m above the
tuyeres (9 m below stockline), while keeping the top temperature under control
with water sprays. During the period of low burden level, the scaffolds will fall
into the furnace and the normal operational conditions are restored as soon as
the burden level has been restored.
Scaffolds in the lower part of the furnace in the bosh/belly area are re–solidified
slag. The scaffolds have a strong effect on burden descent as can be understood
from Figure 11.13. Since the scaffolds tend to grow, operation deteriorates.
Figure 11.13 Scaffolds in lower part of the furnace
This type of scaffold can only be melted away. In a typical example, in total
20 % of the working volume was charged as coke blanks in three batches
on successive days, while the basicity was decreased by 20 %. After that, the
scaffold disappeared.
11.6 Tuyere blockage
With coal injection it is very important that the tuyeres are clear and open,
allowing the coal plume to flow into the raceway to optimize combustion. If
the tuyere should become blocked, or a blockage in front of the tuyere occurs,
the coal must be removed immediately. If it is not, then the coal will be forced
backwards into the tuyere stock and can ignite further up in the connection
with the bustle pipe (see Figure 5.9). This can cause serious damage or even
explosions. It happens especially within an hour of the restart of a furnace,
since unmolten material blocks the tuyere gas flow. The phenomenon has also
been observed with natural gas injection. In many cases this has been a cause
of a chilled hearth since the furnace is shut down in an unprepared state for
extended periods to replace tuyere stock or repair the bustle main.
To prevent an incident, a light sensor may be fitted in front of the peepsight to

detect a blockage at the end of the tuyere, or the ΔP can be measured over the
tuyere stock to detect gas flow into the furnace, if there is no flow, this indicates
that a blockage is present. The coal to that tuyere is automatically switched
off and restarted only once an operator has checked tuyere condition. Many
furnaces using natural gas injection rates over 90 kg/tHM are also installing
the ΔP blocked tuyere detection to shut off the natural gas to prevent back–ups.
Tuyere cameras allow the operator to monitor blockage from the control room.
11.7 Tuyeres: failures and water leakage
The tuyeres are a critical part of furnace equipment, but what can be observed
when walking around the tuyere platform? What can happen to the tuyeres?
There are two types of tuyere – single and double chamber. The single–chamber
tuyeres have one cooling channel, whereas the double–chamber tuyeres
have a separate nose cooling pipe and a body cooling pipe. The advantage
of the double–chamber tuyeres is that the body can continue to be cooled
after the nose has been burnt, and changing the tuyere can wait to the next
planned furnace stop. If a body cooling circuit in a 2–chamber tuyere or the
water circuit in a single–chamber tuyere is burned, a furnace stop is required
immediately to change the tuyere to prevent large quantities of water from
entering the furnace hearth resulting in furnace cooling.
11.7.1 Water Discipline

A duller light in front of the tuyere can be an indication that water is present
at or around the tuyere. This water could be coming from the tuyere itself, the
tuyere next to it, or a cooler in the region above the tuyere.
Other signs to look out for when water is suspected are:

– Water visible under tuyeres, at grouting nipples or at thermocouple sockets
– Top gas hydrogen increase. However, when hydrogen in the top gas indicates
water leakage, there is a very large water flow going into the furnace.
– Smell of ammonia
– Unusual thermocouple activity
– Increase of make–up water frequency in closed cooling systems
– CO gas detected at outlet of open water system
– Cooling furnace trend, less reaction to additional fuel than expected
– Water flame at taphole, a yellow or orange flame instead of a blue flame
– Short or spitting tapholes
214 Chapter XI
When water is suspected it is very important to locate and isolate the source
as quickly as possible to limit the quantity of water entering the furnace.
Depending on the location of the leak, a furnace stop may be required to find
and isolate the leak.
Water inside the furnace is extremely damaging to both the process and the
equipment, and the consequences of allowing leakage to continue are extremely
serious for both. The major consequences are:
– Attack of refractories in the hearth and/or the taphole.
– Local heat shortage in the process leading to inefficient process, scabs, more
burnt tuyeres and an increased requirement for reductant/fuel.
11.7.2 Burnt Tuyeres

The ideal scenario is to change tuyeres because the working lifetime has been
reached, or will be passed before the next furnace stop. This preventative
changing of tuyeres is usually done with a target lifetime of two years for double
chambered tuyeres. However, many companies need to change their tuyeres
much more frequently than this, with a tuyere lifetime of less than 6 months
considered to be poor. Other than preventative changing, the most frequent
reason for changing tuyeres is burning (Figure 11.14). As the tuyere coolers are
made of copper, which has a melting temperature of 1083 °C, water cooling
is absolutely essential to protect the tuyeres, and even then they may not be
protected against a direct ‘strike’ with hot iron.
Burnt tuyeres may occur due to a number of different mechanisms, these being:
– Hot metal attack by low position of root cohesive zone. In order to prevent this
a protective layer is sometimes made on the upper part of the tuyere.
– High hearth liquid levels
– Tuyere condition: inadequate cooling or casting defects during production of
the tuyeres
– Presence of aggressive chemical compounds in the blast furnace, primarily
compounds containing chlorine.
Sometimes tuyeres are damaged and start leaking if the coal lance is not
maintained in the center of the blast flow, and a situation is permitted where the
coal causes erosion of the copper all the way through to the cooling channel.
For tuyere burning by the process 2 different mechanisms are suggested.
1. Tuyere burning at the lower side and nose. This is caused by attack from
circulating coke and sometimes as well by poor drainage of liquid from the
raceway to the hearth. Poor drainage can be caused by coke fines and a high
Al2O3 slag coming from the coke and coal ash (Figure 11.14a).
2. Tuyere burning from the top: hot metal drips down onto the tuyere, causing
local high heat flux so that the cooling water starts boiling, which jeopardizes
heat transfer. Typically, sharp holes “bullet holes” are formed (Figure 11.14b). If
these occurring at frequent intervals, charging more coke at the wall to raise the
cohesive zone can often be successful.
In general: tuyere burning is caused by local poor melting of the ferrous burden.
Methods to prevent it are mentioned throughout the book, an adequate primary
slag basicity is of prime importance (see Chapter VIII).
Figure 11.14 Tuyere failures: Wear at the nose (A) and “Bullet holes” (B)
(A. Yaniga and Francis, 2010)
11.7.3 Tipped Tuyeres

Tipped or dipped tuyeres can be identified by seeing an oval shape rather than
a round orifice when looking through the peepsight. There may also be gas
blowing between the tuyere and cooler seat. In the more extreme cases, the
outlet of the tuyere will be completely obscured. This is thought to occur when
material from above impacts on the tuyere, forcing it downwards as shown
in Figure 11.15. It is also an indication of little support being in place below
the tuyere, to prevent the downward movement. This can either be due to the
design or to wear in the tuyere band. The major cause of tipping is that there
is a big scab of solid burden material resting on the tuyere, which is normally
caused by too low coke percentages at the wall in the throat.
Figure 11.15 Tuyere tipping mechanism (left) and result (right) – upper: tuyere
completely drawn into furnace
216 Chapter XI
11.6.4 Scabs in Front of Tuyeres

It is sometimes reported by the blast furnace operator that scabs are seen
dropping in front of tuyeres. This has long been accounted for as scabs peeling
off from further up the furnace, which may well have been true in the days of
high alkali input to the furnace, but this explanation is less acceptable now that
burden quality has improved.
A more likely explanation is that it is non–molten burden, which was charged a

few hours earlier. There can be several reasons why the burden did not melt in
time: late warming up (caused by low burden level at the time of charging, or
by poor gas permeability caused by lot of fines or by high moisture content), a
very cold furnace (low root of cohesive zone, leading to positioning the cohesive
zone almost on top of the tuyeres), too fast coke usage (when a tuyere has no
injectant).
11.6.5 Blowpipe failures

With coal injection it is very important that the tuyeres are clear and open,
allowing the coal plume to flow into the raceway. If the tuyere should become
blocked, or a blockage in front of the tuyere appears, then the coal must be
removed immediately. If it is not, the coal will be forced backwards into the
tuyere stock and can ignite further up in the connection with the bustle pipe
(see Figure 11.13). This can cause serious damage (Figure 11.16).
The blowpipe can also fail as soon as liquid iron enters it or when the coal lance
breaks. The damage is caused by the hot gas and coke escaping through a hole
in the blowpipe mantle. These situations may cause extreme collateral damage
outside the furnace on the tuyere platform, such as melting wiring, piping and
instrumentation, or even the blast furnace shell.
Figure 11.16 Failed blowpipe

11.8 Stops and starts

11.8.1 Internal state of furnace
When a blast furnace in full operation is stopped, some processes continue.
While the blast is stopped, the direct reduction reactions within the furnace
continue as well as heat losses to the wall. As a consequence, the temperature
of the material in the melting zone is reduced to around 1000 °C, where the
carbon solution loss reaction starts. The decreasing temperature re–solidifies
the melting materials. This means that after a stop, it takes some time for the
burden to start descending. The burden descent restarts as soon as the “old”
melting zone is once again molten (Figure 11.17).
Figure 11.17 Solidified cohesive zone as consequence of a stop
The heat shortage for a stop of a furnace operating with PCI is even worse:
during the stop procedure coal injection into the furnace is switched, and
during the start–up it takes time to restart the PCI. An additional reductant or
fuel shortage results. In addition, after a stop the hot metal silicon sometimes
rises to very high values, especially if during the stop/start procedure the
furnace is operated at a low blast volume. As shown in Figure 8.17, the basicity
of the slag will be affected by the high hot metal silicon and might even solidify
within the furnace. This results in disturbed burden descent. Heating up the
slag is the only solution, which can be achieved by charging extra coke into the
furnace 6–8 hours prior to the stop.
So, in order to compensate for the heat losses during a stop and the risk for high
hot metal silicon, the following measures have to be applied:
– Extra reductant into the furnace. Coke, as well as auxiliary reductants, are
possible compensations for these heat losses. Additional reductant is needed for
the period when the furnace is not operated on PCI.
– Design slag composition for low basicity at high hot metal silicon. Use of a
siliceous lump ore is recommended to reduce the slag basicity. Even if a stop
is unplanned, taking these measures after the stop is worthwhile, since the
cohesive zone will be lifted as soon as the extra coke is processed.
218 Chapter XI
For a blow–in or start–up after a stop, major pitfalls are:

– Too fast blow–in. The solidified melting zone will take time to melt during
the start–up. If the time allowed is insufficient, the pressure difference over the
burden can increase too much, leading to gas escaping along the wall (high heat
losses) and poor burden descent.
– Too fast restart of the PCI. Since the melting zone is solidified, there is a
risk that solid agglomerates will block the hot blast through the tuyere. If
this happens, the coal will still be blown into the blowpipe where it can
cause blowpipe failure. Restarting coal injection only after the burden starts
descending is recommended.
– Too high slag basicity.
11.8.2 Stopping a blast furnace

Stopping a blast furnace means that the hot blast is removed from the furnace.
Since the furnace is connected to a gas flow from stoves via the furnace to
the gas cleaning system, a stepwise approach is required. The objective of the
stepwise approach is that there is no chance that an explosive gas–air mixture
is formed or that slag enters the tuyeres. To this end, the blast furnace has to be
kept on slight overpressure during the stopping procedure.
5500
Reduce blast to min. blower volume at 100 (m³ STP/min)/min
5000 Stop tuyere injectants

Stop charging and check furnace, when furnace empty
4500
Reduce blast to 30 kPa(g) with snort at 150 (m³ STP/min)/min

4000
blast volume (m³ STP/min)
3500
3000
2500
At blast pressure 30 kPa(g) open bleeders
and isolate gas cleaning system
2000
1500 Remove all blast air
1000
Open back draft
500
Open peep sights and back doors
0
0 20 40 60 80 100 120
time (min)
Figure 11.18 Guideline shutdown procedure of a large blast furnace
Step 1 Furnace stopping procedure starts when the furnace hearth is empty. Then blast
volume is reduced, a check may be carried out, and the blast pressure is reduced
to a low level of 30 kPa.
Step 2 Top gas bleeder(s) are opened and the furnace is disconnected from the gas
cleaning system via a goggle valve or water seal – all blast furnace gases are
emitted into the atmosphere. The furnace still has at least one taphole open.
Step 3 Blast volume is reduced to zero, and the backdraft stack is opened in order to
burn gas from the furnace. The tuyeres may be clayed to prevent air ingress and
consequential coke burning during the stop. Some backdoors or peepsights are
open to allow for some air ingress into the bustle main.
11.8.3 Start with all tuyeres

After a stop shorter than 5–7 days, medium to large blast furnace can start–up
with all tuyeres, provided the stop was well prepared. During a blow–in, the
reverse process takes place:
Step 1 Furnace is put on low blast pressure.
Step 2 Furnace is connected to gas grid and bleeders are closed.
Step 3 Hot blast volume and pressure are increased stepwise.
Note, that it takes time for the burden to start descending after a stop and that
coal injection should only be restarted after descent has begun.
5500
During start–up, Δp is kept below
maximum (under 1.2–1.3, to be Increase blast volume at
5000 confirmed during operation) 2.5 (m³ STP/min)/min until
100% of target blast volume
4500 Increase blast volume at 5 (m³ STP/min)/min
until 90% of target blast volume
4000 Increase blast volume at 200 (m³ STP/min)/min
until 80% of target blast volume
blast volume (m³ STP/min)
3500 When burden descends, start PCI and oxygen

At 60% or target wait for burden descent. Start tapping
3000
Increase blast volume at 100 (m³ STP/min)/min until 60% of target blast volume.
2500
2000
Connnect gas cleaning plant and close bleeders and start charging
1500
Put blast on furnace and increase to 30 kPa(g)
1000
500 Prepare casthouse equipment (first casts with diameter +10 mm),
open tuyeres and insert PCI lances
0
0 60 120 180 240 300 360
time (min)
Figure 11.19 Guideline for restart with all tuyeres for a large blast furnace
11.8.4 Blow–down
Blowing down a blast furnace requires operating the furnace without
simultaneous charging of the furnace. All the material charged into the furnace
is then exposed to the same temperatures and reduction processes, as if the
furnace were fully charged.
However, since the temperature of the shaft gas is not transferred to the cold
charge, the off–gas temperature increases, and the gas composition changes.
Since the equipment has not been designed to withstand the high top gas
temperatures, the top gas temperatures are kept under control by spraying water.
220 Chapter XI
The water sprayed above the burden should be prevented from reaching the
burden surface, either directly via descent on top of the burden, or indirectly via
the wall. Special water atomizing nozzles are required. The success of the blow–
down depends heavily on proper spraying. The progress of the blow–down
process can be measured from the burden level, as well as from the analysis of
the top gas composition. Since less and less oxygen is removed from the ore, the
CO2 percentage decreases and the CO percentage increases (Figure 11.20).
0
–4
–8
Stack
–12
–16
Bosh
–20
Tuyere level
–24
40 0.8
35 0.7
CO
30 0.6
25 0.5
20 0.4
15 H2 0.3
10 CO2 0.2
5 0.1
O2
0 0
0 60 120 180 240 300 360 420 480 540 600 660 720 780 840
Figure 11.20 Typical progress of a blow–down
Moreover, generally H2 levels increase as a consequence of the (unavoidable)

contact of spraying water with the hot coke. At the end of the blow–down,
when the level of the coke approaches the tuyeres, the CO2 formed at the
tuyeres has insufficient opportunity to be transformed to CO, and the CO2
percentage in the top gas increases. As soon as half of the oxygen is in CO2 (i.e.
when the CO2 percentage equals half the CO percentage), the furnace should
be isolated from the gas system. Normally, a blow–down takes 10 to 12 hours,
after a preparatory stop, to reach the tuyere level.
Prior to the blow–down the furnace contains coke in the active coke zone and
deadman, and alternating layers of coke and ore in the melting zone and stack.
Since during the blowdown, the coke in the active coke zone and the deadman
will be gasified, there is coke excess in the blast furnace. During the latter stages
of the blowdown reduction reactions have largely stopped, so any auxiliary
reductant injection can be stopped during the early stages of the blowdown. The
moment is indicated by the gas analysis: as soon as the CO2 percentage starts to
decrease to below 10 %, there is little iron oxide left to reduce.
The burden level in the furnace is difficult to measure with standard stock rods.
Mechanical stock rods have to be equipped with chain or cable extensions and
recalibrated for the purpose. The stock rods should be used only at intervals,
since the high temperatures above the burden may cause chain breakage.
Radar level indicators can be used if reliable. Indications from the level of the
burden can also be obtained from:
– The pressure taps.
– The casthouse operation i.e. the quantity of iron cast.
– Calculation of the amount of coke consumed in front of the tuyeres.
Another indication that the blow–down is complete is the collapse of the

raceway (no more coke rotating) and detection of oxygen in the top gas. This is
caused by some of the blast going directly to the furnace top, since there is no
longer a full quantity of coke in front of the tuyeres.
The required condition of the furnace after the blow–down depends on the
purpose of the blow–down and consequent repair. Generally, the walls have to
be clean. Cleaning of the hearth is another important topic. If solid skulls and
scabs are expected in the hearth and have to be removed prior to the blow–
down, the furnace can be operated for a prolonged period on a high thermal
level, relatively low PCI rate, increasing hot metal manganese, increase of
central coke charging and a burden without titanium addition. The full effect of
these measures is uncertain, but hearth bottom thermocouples can indicate that
there is more activity and less of a skull in the hearth. Eliminating nut coke for
a week or two before the shutdown can also be considered.
11.8.5 Blow–in from new or after reline

Blowing in a furnace from new can be considered in two phases:
Phase 1 Heating up the hearth.
Phase 2 Starting the reduction reactions and iron production.
The heat requirement in the early stages of the blow–in is for the following:
– Heat coke in the hearth, deadman and active coke zone to 1500 °C.
– Heat required for evaporation of moisture from the coke.
– Heat required to compensate for moisture in blast dissociating into hydrogen
(H2O + C CO + H2).
– Heat to compensate for loss of heat through the furnace wall.
Many operators have learned from past experience that a coke blank ranging
from 50 to 65 % of the working volume or coke blank elevation of 11 meters
above the tuyere fulfils these heat requirements.
– In the early stages of a blow–in, blast temperature should be maximized and
blast moisture minimized.
– Heating up the hearth requires some 7 to 8 hours after the blow–in. Heat is
generated from coke used at the tuyeres.
222 Chapter XI
Starting the reduction processes

During the early stages of the blow–in while the hearth is heating–up,
reduction of the iron oxides has not yet begun due to the temperatures being
too low. For this reason, the increased amount of direct reduction has to be
considered. The situation may become difficult if the level of direct reduction is
too high, (and gas reduction is low). This situation manifests itself from:
– The gas utilization.
– The direct reduction, as shown by CO + CO2 exceeds “normal” values.
The gas utilization is an indication of the amount of gas reduction taking

place, while the total CO and CO2 percentage is an indication for the direct
reduction. The CO2 percentage in particular indicates if gas reduction is taking
place.
Slag formation
In general, the slag during blow–in has to be designed for high hot metal
silicon. However, with the proposed method the hot metal silicon should be
under control. If we continue to follow the “two–phase” blow–in approach
mentioned here, during the first phase of the blow–in about 350 tonne coke is
gasified in 8 hours and the slag formed comes only from the coke ash. Taking
10 % ash and 30 % of the ash as Al2O3, we get 35 tonnes of high Al2O3 slag
during the first 8 hours. This will not cause a problem in the furnace because of
the small volume. The coke ash can be fluxed with material containing lime.
Hot metal quality during blow–in

As soon as the hearth is heated the hot metal temperature exceeds 1400 °C. As
soon as the top temperature exceeds dew–point, all excess moisture has been
removed from the furnace and the process has started. Only limited heat is
required for heating–up and drying the refractories, if compared with the heat
requirements of the process itself. So, as soon as hot metal temperature reaches
1400 °C and top temperature exceeds 90 °C, the process has to be brought back
to normal operation conditions.
However, in this situation the coke rate in the furnace is still very high and
the hot metal silicon will rise to 4–5 %. The hot metal silicon can be reduced
by putting a normal coke rate into the furnace. The “normal” coke rate at “all
coke” operation is about 530 kg/tHM. A considerable period is needed here
to consume all the excess coke present in the furnace. More rapid decrease of
hot metal silicon can be reached if a lower coke rate is charged and auxiliary
injection is used as soon as required. The injectant is switched on as soon as the
hot metal silicon decreases below 1 %.
Example 11.2 Rapid Blow–in

An example of a rapid blow–in of a furnace with a filled hearth is presented
in Figure 11.21. Initially, the furnace was started up with four tuyeres (of
25). After opening all tuyeres, a “heavy” burden (coke rate 440 kg/tHM) was
put in the furnace 30 hours after the blow–in and coal was injected into the
furnace 40 hours after the blow–in. Hot metal silicon reached the 1.1 % mark
44 hours after the blow–in.
800 25
Tuyeres Opened
Coke rate
700
20
600
500
15
kg/tHM
400
Silicon (%)
10
300
200
5
100
Blast Volume (1000 m³/hr) PCI on
0 0
0 12 24 36 48
time (h)
Figure 11.21 Blow–in of a blast furnace in November 2013: Coke rate charged, blast
volume, number of tuyeres open and hot metal silicon
224 Chapter XI
11.9 Casthouse challenges

11.9.1 Chills, loss of connection
A chilled hearth is a situation where the connection between tuyere and taphole
is lost. Liquids generated above the tuyeres can no longer drain to the hearth
and in most cases all tuyeres are filled with slag/melting materials. Sometimes a
catastrophic failure of blowpipe or tuyere stock was the reason to finally stop the
furnace. The furnace is filled with (semi) solid materials well above tuyere level,
see Figure 11.22.
Figure 11.22 Loss of connection of liquid flow to taphole
What are the possible causes?

– Poor burden descent/ slips, hanging:
– Gunnite or shotcrete falling off the walls
– Scabs falling off the walls
– Gas–reduction of burden deteriorated
– As a consequence of disturbed gas flow
– As a consequence of casting delays, especially if connection is lost after long
stops.
– High water input in lower part of furnace
– Burned tuyeres
– Leaking tuyeres, staves, cooling plates, hot blast valves
– Water input with burden materials
– Top sprays leaking through the shutoff valves
– Increased heat loss. Burden distribution, loss of refractory protection of
staves
– Casting problems
– Burden quality
– Improper charging. Burden materials or burden distribution.
– Loss of fuel injection.
How to recover?
Since the liquids in the furnace have been frozen, the following line of action
has to take place:
1. heat has to be brought into the furnace
2. the material in the furnace has to melt as easily as possible and
3. the molten material has to be taken from the furnace.
So, the recovery principle is:

– bring maximum heat into the furnace: note that the tuyeres are the motor of the
melting process.
– make slag as liquid as possible: very lean (typically a B2 basicity of 0.8–0.9).
– cast and clean to drain liquids form the furnace.
Step 1 is re–establishing the connection between tuyere and taphole. This can
be done with or without oxy–fuel lances (Figure 11.23). The connection is re–
established by starting at a few tuyeres above one taphole. We recommend at
least three tuyeres, since the tuyeres on the outside lose a lot of heat to laterally
to the adjacent tuyere areas that are not working,
Step 2 is to bring sufficient fuel to the tuyeres, so that the liquids draining to
the hearth are hot. This can be done by charging coke blanks. Note that a too–
high fuel rate only leads to burning coke but no hot metal reaching the hearth.
Step 3 is normalizing the process by placing more and more tuyeres in operation
and normalizing fuel rate and process settings.
Figure 11.23 Oxyfuel lances mounted in a taphole can establish taphole–tuyere

connection with oxygen and natural gas.
226 Chapter XI
11.9.2 Coke mess

Coke mess is a situation where a lot of coke comes out of the taphole. This can
happen when the taphole allows the coke through: it is normally rather short
and wide and can be caused by the fact, that a taphole was out of operation for
a long time, by poor–quality clay, or by water leakage. The coke quality also
plays a role: the poorer the coke quality, the easier coke mess can result. The
operational reaction is to lower the wind volume, plug the taphole as soon as
possible, and start carefully anew. If water leakage plays a role, the problem has
to addressed first before the next cast. The problem can also occur with a new
taphole or poor–quality clay which wears too rapidly.
11.10 Greenhouse gas emissions
The amount of carbon we put into the furnace for normal operation is the
carbon present in 300 kg coke and 200 kg coal (Table 11.1). If we assume that
this carbon is emitted as CO2, then the emission of CO2 is 1533 kg/tHM.
This is a simplification of reality, since carbon is used for sinter making, coke
making, blast generation and the electric power used for the blast furnace
operation, nor are credits for top gas taken into account. The European
Commission has made a study of CO2 emissions in the steel industry (Pardo et
al, 2012). This study comes to a total CO2 emission of 1279 kg/tHM. According
to Worldsteel, the energy intensity of steelmaking is 20 GJ/tonne, with an
average CO2 footprint per tonne of steel of 1.8 tonne, where for the integrated
steelmaking route, the hot metal from the blast furnace accounts for 70 % of
the CO2 output.
Since a blast furnace is operated close to the thermodynamic limitations,

greenhouse gas emission reductions coming from increased efficiency will be
relatively small. Table 11.1 shows the example of replacing 54 kg coal injection
with 50 kg gas injection. This reduces CO2 emissions by about 2 %. The “best”
way to reduce greenhouse gas emission from the BF–BOF route is to use high
coal injection rates, thus minimizing coke oven emissions, and to make efficient
use of top gas. When further CO2 reduction is required, the use of pellets will
further reduce carbon consumption in the sinter making process.
Scenario
Carbon Coal injection Co–injection
content (%) (kg/tHM) (kg/tHM)
coke 87.0 300 300
coal 80.0 200 146
natural gas 72.0 0 50
carbon input kg C/tHM 421 414
CO2 emission kg CO2 /tHM 1544 1517
Table 11.1 CO2 footprint assuming that all carbon input is converted to CO2
Charging metallic, like HBI and scrap, takes out carbon for reduction and
leaves only coke demand for melting these metallics into hot metal.
Hydrogen content can be increased through the co–injection of natural gas,
increased blast moisture or even hydrogen enrichment.
However, all the above methods will be limited once the minimum top gas
temperature is achieved, as shown in Chapter V, and further carbon reduction
is not feasible. A carbon–free ironmaking process needs to be separated into
a reduction step (i.e. direct reduction shaft or fines–based reduction) and a
melting step (i.e. Electric Arc Furnace).
It is the blast furnace operator’s responsibility to lower the carbon rate as much
as possible. In order to do this, the route from coal and ore to steel has to be
evaluated. In many companies the efficiency of the blast furnace is reported in
total fuel rate and the operator tries to reach as low total fuel rate as possible.
Total fuel rate is often defined as the sum of coke, nut coke and coal. This leads
to a penalty in total fuel rate at higher PCI, since coal replaces coke typically
by 0.8–0.9 kg coke per kg coal. The consequence is that in many plants, coke
rates are well above proven best practice levels. Especially in South–East Asia
there are many modern, well–operated blast furnaces with good quality raw
materials. A few of these reach coke rates well below 300 kg/tHM as an annual
average, while in other parts of the world these low coke rates have already been
standard operation for more than 10–20 years.
Management should judge total fuel rate of a blast furnace on the basis of
standardized coke equivalents, so that proper replacement ratios can be taken
into account. And for CO2 generated per tonne of hot metal the CO2 of coke
making has to considered as well.
Annex I Blast Furnace
Instrumentation
Instrumentation overview
Modern blast furnace operation can be controlled only by using a wide range of
instrumentation. A multitude of temperature, flow and pressure measurements
are installed in a modern blast furnace. A few of the instruments are discussed
in the present section. An overview of blast furnace instrumentation as
discussed in various parts of the text is given in Figure A1.1.
Recent developments are:

– Two–dimensional temperature measurement on burden surface as measured
with acoustic signals.
– Thermal image camera above the stockline.
– Tuyere cameras.
– Hot metal and slag flow.
Figure A1.1 Overview blast furnace instrumentation
Radial gas temperature above stockline

Traditionally, the gas temperature distribution in the blast furnace top is
measured at multiple points over one radius or several radii. The “SOMA”
system is based on a measuring principle that the speed of an acoustic signal
depends on gas temperature. By having 8 (or 10 for a large blast furnace)
transmitter/receiver units installed around the circumference above the
stockline, the SOMA system provides a gas temperature “map” over the entire
burden surface. It can also calculate radial temperature profiles.
230 Annexes
Figure A1.2 compares single day average temperatures as measured with cross
probes and the SOMA system. The SOMA data have been derived for the
same radius as the cross probes. SOMA gives lower central temperatures in the
example, but the major advantage is that the stockline is not disturbed by the
“shadow” of the cross probes when material is charged into the furnace.
500
Cross 1
Cross 2
400
SOMA 1
Temperature (°C)
SOMA 2
300 Av top
200
100
0
–1 –0.75 –0.5 –0.25 0 0.25 0.5 0.75 1
Radius (dimensionless)
Figure A1.2 Comparison of radial temperature measurements with cross probe and
SOMA, daily averages on a stable day
Pressure taps
Pressure taps indicate whether or not the gas flow is stable. Pressure taps
indicate whether or not “short circuiting” of gas flow along the wall takes place.
In stable periods the layers of coke and ore can be followed passing the taps.
Figure A1.3 Pressure taps indicating the stability of the process, 24 hour graphs. The
example shows stable (left) and unstable (right) operation.
Tuyere Cameras
Tuyere Cameras are becoming more standard. See Figure A1.4 for an example
display. They give real time continuous monitoring from the control room,
which enables immediate detection of tuyere blockage or lance damage,
which can prevent the burn down of a blowpipe. In some countries, using the
image analysis to automatically shut off the coal injection when a blockage is
detected is permitted. In others a simpler more robust system is required for
automatic shut-off, such as light meters or differential pressure on the downleg.
(Chapter V). Operators are now developing more sophisticated image analysis
231
techniques to generate more process condition indicators. For example, Stelco

reported (Ward et al, 2018) a signal processing technique, based on lack of light
intensity, to record accretions appearing in front of the tuyeres. The frequency
of such events is related to stave leaks, poor ferrous material quality and too-
low cohesive zone position, which may be an early warning of tuyere failures.
Different companies have similar indicators, like “low–low flow” at a tuyere
stock ΔP below a minimum, or “dark” when using a light detector.
Figure A1.4. Typical tuyere camera and control room display.
Top cameras
Infrared camera technology now permits a small camera installed on the
furnace cone to measure the stockline surface temperature between charging
temperature up to 1200 °C. The camera creates digital images which can be
recorded many times per second when required. Image data post-processing
techniques have been reported (Huang et al, 2019) which generate virtual above
burden temperature probe profiles, central gas flow strength and location, wall
gas flow strength, and gas channeling events. Prior to processing the image, it
must be corrected for distortion and to remove the effect of the rotating chute.
Additional benefits reported are the ability to detect burden spraying, due to
excessive moisture, and the fine-tuning of the charging mathematical model,
particularly for coke/ore ratio in the center (Huang et al, 2017).
Figure A1.5. Example of wall gas flow, central gas flow and channel. (Huang et al, 2017)
232 Annexes
Burden Profile Measurement

When aiming for low coke rates, it is important to have an accurate
measurement of the layer shape in order to ensure that the coke is well
distributed across the radius. There must be sufficient coke in each radial
position to ensure that the ferrous layers do not stick together (Chapter VII).
The burden profile is to a blast furnace what the steering wheel is to a car.
Traditionally, measurement is done by a retractable profilometer. This device is
inserted above the burden between charges and continually scans the stockline
as it travels to the center and back. A sequence of measurements for each layer
of the charging sequence is typically made each shift. A radar instrument is
located at the back of the probe, with a 45 ° reflector at the front to divert the
signal onto the stockline. Figure A1.6 shows the reflector and a typical profile
result. This technique gives a high-resolution measurement of the layer shape,
but across only one radius.
Figure A1.6 Profilometer measuring head showing the radar reflector and example
output display from profilometer measurement
An alternative method is to use a 3D radar scanner located on the furnace cone.

This continuously scans the entire stockline and calculates a 3D image. This can
then be further processed to a number of 2D profiles along several diameters.
Some companies report that such instruments are installed on all of their blast
furnaces (Feilmayr, 2016).
233
Annex II Reference Furnace Data

Dimensions Chemical
Hearth diameter m 15 C g/mol 12
Throat diameter m 11 H g/mol 1
Working volume m³ 4250 O g/mol 16
Inner volume m³ 5000 N g/mol 14
Tuyeres 40 Fe g/mol 55.85
Top pressure bar 2.5 Si g/mol 28
Voidage in shaft % 30 1 kmol gas m³ STP 22.4
Voidage in hearth % 20
Charge
tHM/charge t 85
ferrous burden/charge t 136
coke/charge t 25.5
Operation Bulk density Chemical composition (wt%)
kg/m³ ash Fe SiO2 Al2O3 C H O N S VM moisture
Production tHM/d 12000 94.5 4.5
Coke kg/tHM 300 550 11 0.5 5.5 3.3 86 0.2 1.3 0.9 0.6 6
Coal blend kg/tHM 200 8.75 0.5 4.38 2.63 80 4.4 4.85 1.35 0.65 24.0 1
Coal HV kg/tHM 100 7.5 0.5 3.75 2.25 78 4.5 8 1.3 0.7 31.0
Coal LV kg/tHM 100 10 0.5 5 3 82 4.3 1.7 1.4 0.6 17.0
Natural gas kg/tHM pm 72 23 2 3 0
Fe SiO2 Al2O3 CaO MgO FeO O/Fe moisture
Sinter kg/tHM 1099 58.3 4.4 1.5 8.6 1.8 7 1.44
Pellets kg/tHM 314 65.0 4.0 1.4 0.5 0.3 1 1.492
Lump kg/tHM 157 62.0 5.0 1.8 0 1.50
Burden kg/tHM 1570 1800 60.0 4.4 1.5 6.1 1.3 1.46 1
Oxidation O/Fe at/at 1.456 5.1
Hot blast m³ STP/min 7500
Blast oxygen % 27.0
Moisture g/m³ STP 12
Hot blast temp °C 1200 CO CO2 H2 N2
Top gas 1.35 24.0 23.0 5.0 48.0

ηH % 40
2
Production Bulk density Chemical composition (wt%)
kg/m³ Fe C Si Mn P Ti S
Hot metal tHM/d 12000 7200 94.5 4.5 0.45 0.2 0.08 0.05 0.02
Slag kg/tHM 250 2300

234 Annexes
Annex III Rules of Thumb

coke rate Tflame Ttop ΔP
kg/tHM °C °C bar
PCI +10 kg/tHM –8.8 –26.6 9.4 0.014
Natural gas +10 kg/tHM –9.7 –75.8 20.1 0.054
Oil +10 kg/tHM –10.9 –45.7 13.5 0.027
O2 +1% 0.8 33.9 –14.0 –0.047
HBT +10 °C –0.9 5.2 –1.5 –0.007
Moisture +10 g/m³ STP 5.3 –42.5 8.6 0.040
Slag volume +10 kg/tHM 0.7 0.005
Heat loss +100 MJ/tHM 5.4 13.8 –0.4 0.039
Silicon +0.1% 3.5
235
Annex IV The Blast Furnace as a

Chemical Reactor
A 4.1 Carbon and hydrogen
Oxygen in the hot blast and injectant are shortly after the raceway transformed
to carbon monoxide (CO), while hydrogen (from blast moisture and injectants)
is converted to H2. CO and H2 are stable at high temperatures (above 1000 °C)
under blast furnace conditions. What happens with the gas when it ascends
through the furnace and cools down?
First consider what happens with the carbon monoxide.
Carbon can give two types of oxides:
C + ½ O2 CO + heat ΔH = –111 kJ/mole

C + O2 CO2 + heat ΔH = –394 kJ/mole
The first reaction takes place in the blast furnace, but CO2 is not the final
product in the raceway. The second reaction is more typical in a process like a
power plant.
Note that in the second step much more heat is generated than in the first step.
For this reason, CO should be converted to CO2 as much as possible in the
process. The ratio CO2/(CO+CO2) is called the gas utilization or gas efficiency
ηCO and is used extensively in blast furnace operation.
In Figure A4.1 on the next page, the equilibrium Boudouard reaction (2 CO

C + CO2) is presented for various temperatures. At temperatures above 1000
°C, CO2 is rapidly converted to CO, if in contact with coke. So, at the high
temperatures in the bosh and melting zone of the blast furnace, only carbon
monoxide is present. At temperatures below 500 °C, CO has a tendency to
decompose into C + CO2. The carbon formed in this way is very fine and is
called “Boudouard” carbon. The formation of carbon from CO slows down at
lower temperatures and is most pronounced at temperatures of 500–550 °C
(Biswas, p 70). In operational practice, the carbon monoxide decomposition can
be observed in refractory material, where there is a CO–containing atmosphere
in the correct temperature region. CO decomposition jeopardizes the heat
conductivity of refractories and can cause a brittle layer.
Water (H2O) from blast moisture and injectants behaves in a similar way, since
H2O is rapidly converted in the presence of coke to H2 and CO.
236 Annexes
%CO %CO2 CO2

in gas in gas CO+CO2
0 50 100
Equilibrium
at 2 bar
10 40 Equilibrium 80
at 0.6 bar
20 30 C+CO2 2 CO 60
Blast
30 20 Furnace 40
Gas
40 10 20
50 0 0
0 200 400 600 800 1000 1200 1400
Temperature (°C)
Figure A4.1 Boudouard reaction: the drawn lines indicates equilibrium at low top
pressure furnaces and high top pressure furnaces, after Biswas, 1981. The
dotted line shows the gas composition inside a blast furnace.
A4.2 Gas reduction of iron oxides
As soon as gas temperatures decrease below 1000 °C, the CO2 becomes stable
and reduction reactions can take place, such as (see Figure 4.2):
– For Hematite:
3 Fe2O3 + CO 2 Fe3O4 + CO2 ΔH = –53 kJ/mol
– For Magnetite:
Fe3O4 + CO 3 FeO + CO2 ΔH = 36kJ/mol
– For Wustite:
FeO + CO Fe + CO2 ΔH = –17 kJ/mol
The reduction is called “gas reduction” because the oxygen is removed from the
burden materials with CO gas. H2 reacts in a similar way. In the literature it is
also often called “indirect” reduction, since carbon is only indirectly involved in
this reaction. The reduction of the FeO0.5 takes place via direct reduction.
Following the burden descent from the stockline, the reduction from hematite
to magnetite starts at around 500 °C. The reduction from magnetite to wustite
takes place in the temperature zone from 600 to 900 °C, while the reduction
from wustite to iron takes place in the temperature region between 900 and
1200 °C. At the start of melting (1100–1200 °C) FeO0.5 is normally reached.
Here FeO is used as a symbol for wustite, however the most stable composition
is Fe0.95O or FeO1.05. The reactions are shown in Figure A4.2.
237
Figure A4.2 Overview of the reduction of iron oxides (black dots are carbon atoms,
blue dots are oxygen atoms and red dots are iron atoms)
%CO %CO2 CO2

in gas in gas CO+CO2
0 50 100
Fe 3O4 + CO
10 40 Magnetite 80
FeO + CO2
20 30 60
Wustite
30 20 FeO + CO 40
40 10 Iron Fe + CO2 20
50 0 0
400 600 800 1000 1200
Temperature (°C)
Figure A4.3 Schematic representation of the relation between temperatures, CO/CO2

gas composition and iron oxides, the drawn lines indicate equilibrium
The equilibrium between the various iron oxides and the gas is shown in Figure
A4.3. The figure shows the temperatures and gas compositions where further
gas reduction of the burden is no longer possible. The reduction of wustite to
iron requires gas with a relatively high percentage of CO. Gas utilization for
reduction of wustite should be below 30 %. If CO2 content is higher, wustite is
no longer converted to iron by gas reduction.
The progress of the reduction reactions in a blast furnace can be detected in two
different ways:
– Burden: from quenched furnaces an overview of the progress of the reduction
can be derived. An example is shown in Figure A4.2
– Gas: by sending gas sampling devices down into the furnace, the progress of
temperature/gas composition can be derived. Figure A4.4 shows typical results
238 Annexes
from a gas sampling exercise. The data can be depicted in the graph of the
equilibrium between gas and iron oxides. The gas normally shows a “thermal
reserve zone”, that is, a zone in which the temperature does not change rapidly,
as well as, a “chemical reserve zone”, a zone in which the chemical composition
of the gas does not change. The thermal reserve zone decreases and can
disappear when the furnace is pushed to high productivities.
1500
Temperature
Center
1000
Wall Thermal
reserve zone 100
500
80 Magnetite
0
60 60
ηCO Wustite
40 Chemical 40
ηCO reserve zone
20 20 Iron
0 0
0 100 200 300 400 600 800 1000 1200
Time Temperature (°C)
Figure A4.4 Gas composition in operating furnace. CO, CO2, H2 and temperature
were measured with descending probes (Chaigneau et al, 2001). Typical
measurements from various furnaces are shaded (after McMaster, 2002).
The coarse dotted line is the Boudouard equilibrium of Figure A4.1: CO2
is stable at the left side of the line.
A4.3 Direct reduction of iron oxides
Gas in the blast furnace reduces iron oxides from hematite (Fe2O3) to a
situation, where about half of the Fe atoms are metallic and half FeO. So,
the O/Fe ratio when the burden starts melting is about 0.5 O per Fe atom. In
efficient furnaces and furnaces using natural gas injection the O/Fe ratio can
be lower, the ratio of 0.5 is used for explaining the mechanisms. As soon as
burden materials start to soften, the layers become impermeable to gas and gas
reduction stops. When the burden starts to melt around 1300 °C, reduction
progresses, but any CO2 or H2O generated is rapidly converted to CO and H2.
The resulting chemical reaction is
FeO + C Fe + CO ΔH = 152 kJ/mol
The reaction is called direct reduction, since in total the melting iron oxide
consumes carbon directly. This is a confusing terminology, since “direct
reduction plants” convert iron oxides by gas reduction at temperatures of
1000–1100 °C. In addition to the direct reduction of iron oxides, in the lower
part of the blast furnace, a number of different reduction reactions take place:
reduction of SiO2 to Si, MnO to Mn, P2O5 to P, TiO2 to Ti.
239
A4.4 Gas reduction and direct reduction combined

The direct reduction and gas reduction reactions combine for a very efficient
process.
Suppose that all oxygen is removed by direct reduction. Then, the following
reaction takes place:
Fe2O3 + 3 C 2 Fe + 3 CO
Hot metal contains about 945 kg Fe per tonne. Coke contains about 86 %
carbon. Atomic weights of Fe and C are 55.85 and 12 respectively. A tonne of
iron contains 16.9 kmole Fe (945/55.85). For every atom of iron, we need 1.5
atoms of carbon, so the carbon requirement is 25.4 kmole (1.516,9), which is
305 kg carbon (25.512). In addition, about 45 kg carbon is dissolved in iron. In
total, 351 kg carbon is used per tonne of hot metal, which corresponds to only
408 kg of coke. This is a very low equivalent coke rate, and a blast furnace will
not work, because the heat generated in this reaction is too low.
Now consider that all reduction reactions are done via gas reduction, what coke
rate is required in this situation? It is assumed that coke combustion generates
the CO required. The reaction is:
3 FeO + 3 CO 3 Fe + 3 CO2
We only consider the reduction of wustite since the resulting gas is powerful
enough to reduce magnetite and hematite. We know from the above (Figure
A4.3) that for gas reduction the maximum gas utilization is 30 %. To get 30 %
gas utilization more CO is needed and the reaction is (since 3/(3+7) = 30 %):
3 FeO + 10 CO 3 Fe + 3 CO2 + 7 CO
So, the coke requirement is calculated as above: every tonne of iron contains
16,9 kmole. There is a need for 10 carbon atoms per 3 atoms of Fe. So, the
carbon requirement is 57 kmole (10/317), which corresponds to 684 kg carbon
(5712). Again, the extra 45 kg carbon in iron has to be added, yielding a carbon
rate of 729 kg/tHM and a coke rate of 848 kg/tHM (729/0.86). This reaction
has a poor coke rate and a high heat excess.
The conclusion of the considerations above is that the counter–current character

of the blast furnace works efficiently to reduce the reductant rate by combining
direct reduction in the lower furnace with gas reduction reaction in the upper
furnace. Approximately 60–70 % of the oxygen of iron oxides is removed by gas
and the remaining oxygen is removed by direct reduction.
240 Annexes
A4.5 Reduction by hydrogen

Hydrogen is formed from moisture (H2O) in the blast and injectants
(hydrocarbons) in the raceway. Hydrogen can act as a reducing agent to remove
oxygen and form water. The reaction is comparable with that for carbon
monoxide:
H2 + FeO Fe + H2O
The major differences with the reactions for hydrogen and carbon monoxide are
as follows:
– Figure A4.5 shows the equilibrium of the iron oxides and hydrogen (red lines)
superimposed on the CO/CO2 equilibrium lines. Hydrogen is more effective
at temperatures above 821 °C. From measurements in the blast furnace, it
has been shown that hydrogen reactions are already nearly complete at this
temperature.
– Hydrogen utilization as measured from the top gas is normally 40–45 % while
CO utilization is close to 50 %. At the FeO level (900 °C), hydrogen is utilized
for 35 %, which means that it is already close to its final utilization of 40 %.
– Hydrogen is less effective as a reductant at lower temperatures, because it
consumes heat when reducing iron oxides.
At high temperatures, the H2O that is formed in the furnace reacts with coke:
H2O (steam) + C H2 + CO ΔH = 124 kJ/mole
This reaction consumes a lot of heat. At higher temperatures (over 1000 °C) the
reaction proceeds rapidly to the right.
%CO %CO2 CO2 H2Oprocess

or
in gas in gas CO+CO2 (H2+H2Oprocess)
0 50 100
10 40 Fe3O4 80
20 30 60
FeO
30 20 40
40 10 Fe 20
50 0 0
400 600 800 1000 1200
Temperature (°C)
Figure A4.5 Equilibrium iron oxides with hydrogen (red) and carbon monoxide
At high temperatures (above 1000 °C) water vapor gasifies coke. At lower
temperatures (800–1000 °C), the water–gas shift reaction becomes important:
H2O + CO H2 + CO2 ΔH = –40,7 kJ/mole

241
In this temperature range, hydrogen is more effective as a reductant than

carbon monoxide. This is shown in Figure A4.6, where the H2 utilization/CO
utilization is larger than 1.
The water–gas shift reaction shifts to the right when the temperature decreases.
The reaction approaches equilibrium rather fast at temperatures above 800 °C,
but fails to reach equilibrium below approximately 730 °C.
The hydrogen utilisation of the top gas is defined as ηH = H2O/(H2+H2O), 2
where H2O refers to “process water”, water generated in the blast furnace
process without moisture from burden and coke. The utilization is estimated
from the H2 input through the tuyeres; the H2Oprocess is the difference between
the hydrogen input and the hydrogen leaving the furnace with top gas, as
measured with the gas analysis.
1.2
1
ηH and ηCO (%)
0.8
0.6
2
0.4
0.2
400 500 600 700 800 900 1000
Temperature (°C)
Figure A4.6 Water gas shift reaction showing the hydrogen utilization/CO utilization.
Above 821 °C hydrogen is a more efficient reducer than CO. The red
dotted line indicates the thermodynamic equilibrium that fails to set up
in operating blast furnaces below 730 °C. The line drawn in blue indicates
operational results. Dots are data from top gas in operating blast furnaces.
242 Annexes
Annex V Expert Systems

and Models
An expert system helps the operator to continuously analyze all process data
and to suggest or execute corrective actions in real time. The corrective actions
are based on the deviation of major process parameters from their target values
and predefined rules of how to correct for the deviations. Any company with
a book of “Standard Operating Procedures” can consider an expert system by
automating these operating procedures in the process computer. The major
advantages of expert systems are:
– Continuous monitoring of all relevant process parameters.
– Option of developing the system to improve decision–making rules.
– Apply operational experience developed over many years by various operators.
– Standardize decision making in process control and operations.
– Implement alarms or take actions if a parameter exceeds the control range.
However, an expert system has its limitations for some parts of the process
and is not suitable for controlling special situations (blow–in, sudden chill).
An expert system is no replacement for experienced blast furnace operations
experts.
A5.1 Visualization of process conditions in real time
The process control system shows screens where the operator can monitor all the
variables for a certain aspect of the process. An example is shown below. There
are three important areas:
– Diagnostics: process parameters are scaled between –1 and +1, or between 0 and
1, and “red” and “green” areas are presented, red showing parameters outside the
normal control area.
– Control actions or suggested control actions. The suggested actions serve
to establish a dialogue within the shift, which allows the operator to decide
whether or not the suggested action is to be implemented. The comment of
the operator on the suggested actions even when not implemented, allows
optimization of the process rules within the system. In this case, the expert
system helps for consistency of measures, since actions executed manually are
operator–dependent and may come into conflict with the expert system. It is
preferable to adjust the system and improve the recommended actions.
– “Open” or “closed” decisions on suggested actions: in the open situation, the
expert system presents a suggested action, but does not react if no decision
is made. Closed decisions are executed automatically after 20 minutes if no
rejection has been received by the system. These actions must be taken based on
the speed of the process, typically between 5 and 7 hours. 20 minutes may be
(too) long, since 20 minutes plus system adjustment of weighing and burdening
program leads to a delay of 45 to 50 minutes until a change enters physically
into the blast furnace.
243
A5.2 What process parameters

and diagnosis can be made?
The expert system follows a huge number of measurements in the furnace and
makes a diagnosis whether or not the parameter is in a desired range. Among
others, the following diagnoses are made:
– Burden descent: compares the burden descent of the last 24 hours with the
last hour and analyzes whether or not the burden descent is slow, normal or
accelerated. (See section on direct reduction)
– Slag and hot metal quantity produced and drained from the furnace. The
production balance is made, based on the progress of the last 25 casts as a
reference. Subsequently, the last cast and the last three casts are analyzed to
monitor high, normal or low slag drainage. The same is done for hot metal.
– Thermal state of the furnace: based on temperature, hot metal carbon, silicon,
sulfur and titanium as well as slag FeO level. The thermal state is considered
low, normal or high.
– Stave temperatures and stave heat fluxes: continuous analysis of stave
temperatures and heat fluxes on the various levels, defining 2 types of
conditions: the current level and current tendency. The diagnosis sounds the
alarm when 25 % of the temperatures are below lower level or above upper level
limits with the corresponding decreasing or increasing tendency.
– Skull/scaffold formation: follows the temperature and heat fluxes through
the staves and defines scaffold formation as decreasing tendencies of both
parameters for the last 24 hours. To this end, the individual staves are taken
in “areas” representing several staves. Occasionally it is found that the loss of
thermal activity in the staves is also manifest in neighboring areas. The opposite
of the skull/scaffold formation is called “peeling”, which is the loss of the skull.
This is analyzed based on the same data.
– Charged burden: compares setpoint weights of materials charged with actual
weights and monitors the deviation from setpoint.
– Hot blast stoves: analysis of the moments that stoves are changed for every
stove, allowing the operator to operate the stoves within ±10 °C of the target
value.
A5.3 Process control actions
Process control in the blast furnace has various areas and targets.
– Thermal control: the liquid products have to be drained from the hearth at the
desired temperatures.
– Chemical control: the hot metal and slag need to have the desired chemical
composition.
– Gas flow control: the gas flow in the furnace can be monitored and has to be
optimized for efficiency of contact between gas and burden.
– Casthouse control: the hot metal and slag have to be removed from the furnace
hearth at the desired rates and intervals of time.
– Control of equipment like tuyeres and coolers.
244 Annexes
– Incident control: how to manage unwanted and unexpected process conditions

for which no cause is known: what to do if a furnace is out of control: slips,
hangs, chills suddenly, etc.
In Table A5.1 below the various subjects for process control are summarized: the
indicators used and the corrective actions. It shows a typical example of what
can be done. Every company develops its own expert system, based on its own
experiences and operating practices.
Field indicator/diagnostics corrective action

Thermal Control Hot Metal Silicon Step 1: adjust injection rate 2–5 kg/tHM
Hot Metal Temperature Step 2: adjust coke rate
Hot Metal Carbon (Ti, Mn) in severe chills:
Slag FeO Extra coke
Chemical Control New chemical analysis Change input according to guidelines
burden materials
Slag basicity Change input
Burden input Adjust input
Hot metal Manganese Adjust input
Gas Flow Control Coke % in center Adjust central flow by increase/
decrease coke % in center
Coke % at wall Correct burden distribution
Drying capacity/water Increase gas volume/tHM (lower
eimination oxygen enrichment, higher fuel rate)
Too high ΔP, total column Lower blast volume
Too high ΔP, upper or lower Check fines coming into furnace, check
part hearth drainage
Wall gas flow as manifest Adjust blast volume
from spikes in stave (longer term: burden distribution)
temperatures (peeling),
high local heat loss or
decreasing peaks
Casthouse Control Slag and hot metal
drainage
Incidents Burden level too low Caused by charging system: extra coke,
pull wind
Caused by process: decrease wind
volume, extra coke. control ΔP
Sudden chills (temperature Extra coke, lower wind volume
and silicon drop)
Water leakage/top gas From cooling system, burnt tuyeres:
hydrogen repair, a stop may be required
Table A5.1 Indicators and corrective actions in an expert system
An expert system is able to check in real time a huge number of measurements

and is able to suggest process control actions to the operators for the various
area. In the present section a few examples are provided of how an expert system
works.
245
A5.4 Process control with an expert system

Thermal control
Thermal control can be based on the hot metal temperature and hot metal
silicon. Both factors have their own weight. An example is, that the thermal
stage of the furnace is judged from the average hot metal temperature and
silicon of the last three casts. Since three casts represents a production, which
is larger than the production coming from the complete working volume of the
furnace, the hot metal temperature and silicon are always delayed by a number
of hours. This means that the system checks whether or not there are indications
of increased production rate, for example from the charging rate. If so, this is
taken into account. The correction can be executed by changing fuel injection
or by adjusting the coke rate. This type of choice is built into the system
according to local experience and preferences.
Different blast furnaces can have different methods for thermal control. For
instance, it can be focused on hot metal temperature. Hot metal silicon often
gives an earlier indication of a cooling trend than hot metal temperature.
Control of injection (coal, gas)

The operator targets a constant injection rate per tonne of hot metal produced.
Since the actual flow rate of the injectant is measured in kg/hr (coal) or m³
STP/hr (natural gas), the set point for the injection has to be corrected for the
actual production. The main factor determining the actual production is the
amount of oxygen per hour blown into the blast furnace. So, injectant control
should take into account hot blast volume changes, as well as changes in oxygen
enrichment and steam injection. An additional correction can be made for
deviations in the direct reduction rate: if direct reduction increases (for example
as a consequence of the input of more fines), then the production rate will
increase, even if all set points at tuyeres are maintained. This can be observed
from the charging rate. A faster response can be generated from continuous
mass balance of the gas, as shown in Chapter VI. The replacement ratio of
injectants has to be considered in order to obtain a constant energy balance.
Tracking the charge

When the burden is changed, the revised burden descends to the tuyeres in 5–7
hours. The tracking system visualizes where in the furnace the change is present
in real time. This is important for extra coke and/or basicity changes when
preparing for a furnace stop or shutdown. An example is shown in Figure A5.1
on the next page.
246 Annexes
Figure A5.1 Tracking the charge: to monitor when changes are being processed with
extra coke descending in the charge.
A5.5 Examples of models
A5.5.1 Mass and heat balance and minimum fuel rate model
The mass and heat balance model checks the input and output of the furnace on
a daily basis. It does so by closing the material balance as well as the enthalpy
balance. Most companies use a two–stage mass and heat balance, where
the upper furnace and lower furnace are separated by the plane where it is
considered that the gas temperature is 900 °C and the O/Fe ratio in burden is 1.
The mass and heat balance when used continuously can serve as an early
warning signal for sudden chills, because the increase of direct reduction can be
calculated in real time from the top gas composition. Moreover, whether there
are deviations in the chemical composition and weights of raw materials can be
ascertained.
A5.5.2 Hearth wear model

The hearth of a blast furnace is equipped with numerous thermocouples. The
temperature readings combined with the thermal properties of the refractories
can be used to calculate the isotherms in the furnace hearth. The isotherm of
1150 °C represents the solid–liquid interface, so is indicative for the hearth
refractory wear. An example is shown in Figure A5.2
247
Figure A5.2 Hearth isotherms
The actual condition of the hearth is shown by the actual thermocouples. But to
estimate the refractory condition of the hearth, the campaign highest readings
have to be analyzed: the highest value in the campaign indicates the limit of
good hearth refractories, since once refractories have been lost, they are not
restored. Instead, scabs are formed in the hearth. Some blast furnace hearth
models even calculate the thickness of the skull on the hearth sidewall and
bottom.
Use of the model helps to maintain the integrity of the hearth wear monitoring
system. Many thermocouples fail during a campaign, so a program for repair
and/or elimination from the model has to be in place in order to secure accurate
results.
While the model is very useful for repair/reline planning, daily operations are
more reactive to high temperature readings in a certain area. If temperatures at
the hearth wall rise too high, there is the risk of a break–out. Most operators
know the weak points of their hearth refractory and alarm temperatures are put
into the hearth monitoring system. When temperatures rise too high, corrective
actions are possible, like plugging tuyeres above the affected area.
248 Annexes
Annex VI Coke quality tests

Coke quality is distinguished by cold strength and hot strength. Coke cold
strength is measured with drum tests. Drum tests are simply cold simulations of
the load on the coke during its descent through the blast furnace. The standard
is ISO 556–2020: “Coke larger than 20 mm determination of mechanical
strength”. If starting with a different coke size, such as larger than 60 mm, the
coke starting size used must be stated. The standard is larger than 20 mm. The
standard sample weight is 50 kg, although most plants now use 25 kg. If so, this
should be stated. The micum slope is generally done on a sample of coke larger
than 20 mm. It was developed under a European research project to assess
bosh coke compared to the feed coke, being more sensitive and enabling better
differentiation
To have a better understanding of the coke degradation mechanism under

mechanical stress we look at Figure A6.1. Here the percentage of the coke larger
than 40 mm and smaller than 10 mm of the sample is presented as a function of
the number of rotations of the drum.
Dff
Coke breakage
M40 Pure abrasion

%>4 of coke lumps
0 mm
Weight %
I40
Stabilization Point
I10
M10 % < 10 mm
100 150 500

Micum Irsid
Number of rotations of drum
Figure A6.1 Comparison of different mechanical tumble tests and results.
From this figure it can be seen that the lumps larger than 40 mm start to
degrade by breakage only, until the point of stabilization is reached, when no
further breakage occurs. This point is reached at between 100 and 150 rotations.
From this point on, further degradation is mainly by abrasion, although some
breakage still occurs, to a lesser and lesser extent as the number of rotations
increases. The weight percentage of coke larger than 40 mm after 100 rotations
is called M40 and the percentage after 500 rotations is called the I40. The weight
percentage of coke smaller than 10 mm is called M10 and I10 respectively.
249
Besides these values, the Fissure Free Size, the Stabilization Index and the
Micum slope have been introduced as coke quality parameters. Although in
this test the parameter used is not the percentage larger than 40 mm of the coke
but the average mean size (AMS) as a function of rotations. We will explain
these concepts with Figure A6.1 as well. First a line (shown in green) is fitted
to the curve of abrasion–only. Then the green line of abrasion is extrapolated
only to the y–intercept (zero rotations) and the AMS of the coke at this point is
calculated. This yields the Fissure Free Size (FFS), also known as Dff. This then
represents the size at which there would be no degradation due to breakage, but
only abrasion.
The slope of the green line of abrasion–only is called the Micum Slope. Some
mills consider this to be a better way to evaluate abradability than traditional
M10 or I10. The FFS was developed to simulate a maximum obtainable
(theoretical) size for stabilized coke. Some believe the FFS approximately
represents the size of stabilized industrial coke at the blast furnace stock line,
which is then considered a more suitable controlling parameter. A stabilization
index can also be defined as FFS/AMS, for which the maximum will be 1 for
fully stabilized coke.
Tests for chemical reactivity and hot strength are elaborated in Section 4.5.3.
250 Annexes
Annex VII Rist diagram

The Rist diagram is a graphical representation of the mass and heat balance of
the blast furnace process. Visually, it shows how certain blast furnace process
parameters are linked and what the consequences are of changes in these
parameters.
This appendix explains the construction of the Rist diagram, based on

simplifications of the combined mass and heat balances.
If the graphical representation is used in a predictive manner, a third balance

is required by dividing the furnace into an upper and a lower part, with the
balance over the lower part giving the model its predictive character. Further
details are in the references: Rist and Bonnivard (1963), Rist and Meysson
(1964), Rist (1977) and Peacey and Davenport (1979), which are at the base of
the analyses below.
A7.1 Mass balance
For a simplified mass balance over the blast furnace, only three elements are of
importance, Fe, C and O.
niFe = noFe n iC = n oC n iO = n oO
n indicates mole of input (i) and output (o) of each of the three elements,
expressed per mole Fe in the hot metal. For simplification, it is assumed that the
elements enter and leave the furnace as follows:
Element Input Output

Fe Iron oxide Hot metal
C Coke CO, CO2, C in hot metal
O Iron oxide, hot blast CO and CO2
So, nitrogen in the air and slag components from burden, coke and or coal are
ignored. To develop the mass balances further, definitions of both ratios and
fractions are introduced.
g
Top gas composition can then be described as the
following ratio: Q C V . For pure CO2 this ratio is 2 and for pure CO this ratio
O
is 1. Consequently, the molar fractions X of CO2 and CO in the top gas are
respectively:
g g
X CO = Q O C V - 1
2
g g
X CO = 2 - Q O C V
This obeys the rule that the sum of these two fractions equals one.
251
The amount of carbon in the top gas (per mole Fe in hot metal) is defined asx
ng C. In a similar way the molar ratio of O in the oxides can be defined: Q Fe V .
O
For hematite (Fe2O3), this ratio is ⁄ and for magnetite (Fe3O4) this is ⁄
m
Finally, for carbon in the hot metal the ratio is given as Q V which equals the
C
Fe
moles of carbon in hot metal per mole Fe in hot metal.
Now it is possible to define the mass balances.
Hot metal
The iron enters the furnace with the ferrous burden and leaves the furnace as
hot metal. Per mole Fe in the hot metal the balance is:
niFe = noFe = 1
Carbon
Carbon enters the furnace as coke and leaves the furnace with the top gas and
partly with the hot metal:
m
niC = noC = ng C + Q Fe V
C
Oxygen
Oxygen enters the x
furnace with the hot blast and with the iron oxide:
niO = nBO + Q Fe V
O
and oxygen leaves g

the furnace with the top gas:
noO = ng C  Q C V
O
If these two x
equations gare combined with of course niO = noO:
Q V = ng C  Q C V
B O O
n O+ Fe (A7.1)
Equation (A7.1) can graphically be represented as a straight line with as x–axis

the O⁄C and y–axis the O⁄ Fe: Figure A7.1. The intercept with the y–axis is nBO
and the slope is ng C.
O
C
CO CO2
0 0.5 1.0 1.5 2.0
1.5
top gas
1
O oxygen from
Fe iron oxide blast furnace
operating line
0.5
0
ngC
–0.5
oxygen from
n BO hot blast
–1
–1.5
Figure A7.1 Rist diagram based on the mass balance of a blast furnace
This very simplified mass balance model can be used to check a measured top
gas composition against the inputs of the furnace based on the operating line of
that furnace.
252 Annexes
However, if for example a prediction of the coke consumption or hot blast is

required, the mass balance model needs to be expanded with a heat balance over
the furnace.
A7.2 Heat balance
With respect to the heat balance over the furnace: heat in = heat out. This is
equal to the balance between heat demand and heat supply:
Heat Demand D = reduction of iron oxides and melting of hot metal.

Heat Supply S = CO and CO2 generation from carbon at 298 K (25 °C)
With D = S
S = ng CO  H0298(CO) + ng CO  H0298(CO2)
2
H0298(CO) = –111000 kJ⁄mole CO
H0298(CO2) = –394000 kJ⁄mole CO2
In front of the tuyeres, only CO is generated, all originating from C. In the

upper part of the furnace, a certain fraction is converted into CO2
ng CO= ng C  X g CO = ng C  {2 – (O/C)g}
ng CO = ng C  X g CO = ng C  {(O/C)g – 1}
2 2
When these two equations are combined:
D = S = ng C  {283000  (O/C)g – 172000} (A7.2)
Other sources of heat demand which have been neglected so far can be added,
like heat losses and slag reactions.
If the heat balance is combined with the mass balance, the model becomes
predictive with respect to coke demand and top gas analysis.
A7.3 Combination of mass and heat balance

By combining (A7.1) and (A7.2) into:
x
n O + Q O Fe V - 283000 = n C $ 172000
B D g
283000 (A7.3)
x
an equation with 4 variables is obtained. If burden composition Q Fe V and
O
heat demand D are specified, two variables remain. If either ng C (moles carbon
associated with gas ≈ coke consumption) or nBO (moles oxygen in hot blast ≈
productivity) is chosen, the other variable is defined and the whole blast furnace
operation is defined.
Equation A7.3 can be rewritten in a form {y2 – y1} = M {x 2 – x1}:
x
I – Q –n O V = n C F 172000
B g
FQ O Fe V – D
283000 283000
– 0I (A7.4)
253
The equation in this format equals in Figure A7.1 the straight line with slope ng C
and at the intercept of the y–axis, where (O⁄C) = 0, (O⁄ Fe) = –nBO.
However, this straight line also has to go through the point:

x
Q O C V = 283000
172000
= 0.61 O
Fe = Q O Fe V – D
283000
x
At a given burden composition Q O Fe V and heat demand D, the actual O⁄ Fe
associated with (O⁄C) = 0.61 can be determined and with that, the enthalpy
point H in the Rist diagram is fixed.
Furthermore, top gas composition will be between 100% CO and 100% CO2
and with that, the whole operating area of a blast furnace is fixed, based on heat
and mass balances. In Figure A7.2, hematite is chosen as burden and the specific
heat demand is set at D = 486000 kJ/mole Fe.
O
C
CO CO2
0 0.5 1.0 1.5 2.0
1.5
1
O
Fe
0.5
blast furnace
operating area
0
H = 0.61, –0.22
n BO
–1
–2
–3
Figure A7.2 Rist diagram with the working area of the blast furnace. Boundaries set
by heat and mass balance
One parameter still has to be chosen in this model (either nBO or ng C).
In order to make the optimal operating condition completely independent,
another fixed point apart from H is required in equation A7.4. This requires
an additional equation, derived from dividing the furnace into an upper and a
lower part, see paragraph A7.4.
A7.4 Mass balance over lower part of the furnace
The blast furnace can be split into two sections which can be seen
independently of each other, as long as for both parts a correct heat and mass
balance can be made. This split is best done at the thermal and chemical reserve
zone, where equilibrium between gas and burden can be assumed.
254 Annexes
If we assume that all direct reduction occurs in the bottom part, and that
wustite is the only oxide phase in equilibrium with CO/CO2, than it is possible
to create the following mass balance, analogue to the mass balance equation of
paragraph A7.1.
n O $ Q O Fe Vwustite = n C $ Q O C Vwustite equilibrium

B g
(A7.5)
From the iron oxide equilibrium diagram, it is known that wustite (O⁄ Fe =
1.06) at 950 °C is in equilibrium at ηCO = 30% (O⁄C = 1.3).
The optimum operating line should intercept with this point, and a second
point, the wustite point W, is created. This fixes the operating line, combined
with point H and both nBO and ng C are now set: A7.3.
O
C
CO CO2
0 0.5 1.0 1.5 2.0
1.5
1 W = 1.3, 1.06
O
Fe
0.5 defined blast furnace
operating line
0
H = 0.61, –0.22
n BO
–1 ngC
–2
–3
Figure A7.3 Rist diagram with an operating line fixed by its enthalpy point and the
wustite equilibrium line
The model has now become predictive with respect to coke consumption (ng C),
hot blast (nBO) and top gas composition (O⁄Ctopgas). Any change in burden, hot
blast temperature, PCI, oxygen enrichment, other reducing
x
reactions, slag and
so forth can be incorporated in the model through Q Fe V and heat demand D.
O
A measured operating line of an existing operation can also be offset against the
optimal operating line and the efficiency of the process can be judged.
The effects of introducing hydrogen from PCI, natural gas injection or

simple moisture in the hot blast can also be calculated through the mass and
heat balances, but this will complicate the simple graphical representation
considerably.
255
Annex VIII References

Agarwal, J.C.; Brown, F.C.; Chin, D.L.; Frydenlund, A.R.; Jessiman, N.S.:
The Use of Total Thermal–plus–Chemical Energy above 2700 °F to Describe Blast
Furnace Operations. Charles River Associates Incorporated, 1991.
Bar–Ziv, E.; Kantorovich, I.I.: Mutual Effects of Porosity and Reactivity in Char
Oxidation. Progress in Energy & Combustion Science, Vol. 27, Issue 6, 2001,
pages 667–697.
Biswas, A.K.: Principles of Blast Furnace Ironmaking. Cootha Publishing House,

Brisbane, Australia, 1981.
Bonnekamp, H.; Engel, K.; Fix, W.; Grebe, K.; Winzer, G.: The Freezing with
Nitrogen and Dissection of Mannesmann’s No. 5 Blast Furnace. Proceedings of the
43rd Ironmaking Conference, Chicago, USA, 1984, pages 139–150.
Cameron, I.; Sukhram, M.; Lefebvre, K.; Davenport, W.: Blast Furnace
Ironmaking, Analysis, Control, and Optimization. Elsevier, Amsterdam, The
Netherlands, 2020.
Carpenter, A.: Use of PCI in Blast Furnaces. IEA Clean Coal Center, London,
2006.
Chaigneau, R.; Bakker, T.; Steeghs, A.; Bergstrand, R.: Quality Assessment
of Ferrous Burden: Utopian dream? Proceedings of the 60t Ironmaking
Conference, Baltimore, USA, 2000, pages 689–703.
Chaigneau, R.: Complex Calcium Ferrites in the Blast Furnace Process. PhD
Thesis, Delft University Press, Delft, 1994.
Committee for Fundamental Metallurgy of the Verein Deutscher

Eisenhüttenleute: Slag Atlas. Verlag Stahleisen, Düsseldorf, 1981.
Danshin, V.; Chernousov, P.: Handbook of Blast Furnace Worker. Metallurgiya,

Chelyabinsk, Russia, 1989, page 190.
Gavel, D.J.: Shifting the limits of nut coke use in the ironmaking blast furnace.
PhD Thesis, Delft University of Technology, Delft, The Netherlands, 2020.
Feilmayr, C.; Schuster, S.; Ganglberger, V.; Lackner, B.; Zankl, D.; Stegfellner,
L.: Innovative Tools for Process Optimization – Burden Surface Scanning via
Beamforming Radar and Measurement of the Top Gas Temperature Distribution.
Proceedings of the 7th European Coke and Ironmaking Congress, Linz, Austria,
2016, pages 151–158.
256 Annexes
Etchevarne, P.; Margot, S.; Musante, R.; Oreggioni, S.; Baglivo, O.: Influence
of Chemical Composition of Sinter on Metallurgical Properties of the Burden of
Ternium Argentina Blast Furnace No. 2. Proceedings of the 22nd IAS Steel
Conference, Rosario, Argentina, 2018.
Fernández–González, D.; Ruiz–Bustinza, I,; Mochón, J.; González–Gasca,

C.; Verdeja González, L.F.: Iron Ore Sintering: Process. Mineral Processing and
Extractive Metallurgy Review, Volume 38, Issue 4, 2017, pages 215–227.
Geerdes, M.; Van der Vliet, C.; Driessen, J.; Toxopeus, H.: Control of a High
Productivity Blast Furnace by Material Distribution. Proceedings of the 50t
Ironmaking Conference, Washington, USA, 1991, pages 367–378.
Geerdes, M.: Co–Injection of Coal and Gas in Blast Furnaces: Are there Hidden
Benefits? Iron and Steel Technology, Vol. 13, Issue 4, 2016, pages 82–88.
Geerdes, M.; Warren, P.: How Iron Ore Melts: Role of Coke and Nut Coke.
Proceedings of the 8th Internation Congress on Science and Technology of
Ironmaking, Vienna, Austria, 2018, pages 380–385.
Grebe, K.; Keddeinis, H.; Stricker, K.P.: Untersuchungen über den

Niedrigtemperaturzerfall von Sinter. Stahl und Eisen, No. 100, 1980, pages
973–982.
Grosspietsch, K.–H.; Lüngen, H.–B.: Coke Quality Requirements by European

Blast Furnace Operators. Stahl und Eisen, Vol. 120, Issue 9, 2000, pages 35–42.
Gupta, S.; Al–Omari, Y.; Sahajwalla, V.; French, D.: Influence of Carbon
Structure and Mineral Association of Coals on Their Combustion Characteristics
for Pulverized Coal Injection (PCI) Application. Metallurgical and Materials
Transactions B, Vol. 37B, 2006, pages 457–473.
Hartig, W.; Langner, K.; Lüngen, H.B.; Stricker, K.P.: Measures for Increasing
the Productivity of Blast Furnace. Proceedings of the 59t Ironmaking
Conference, Pittsburgh, USA, 2000, pages 3–16.
Kolijn, C.: International Cokemaking Issues. 3rd McMaster Cokemaking Course,

McMaster University, Hamilton, Canada, 2001.
Hippo, E.; Walker, P.L.: Reactivity of Heat–Treated Coals in Carbon Dioxide at

900 °C. Fuel, Vol. 54, Issue 4, 1975, pages 246–248.
Huang, D.; Andrade, M.; Walpole, C.; Caplick, C.: Data Post–Processing for
Above Burden Infrared Camera (ABirC) System. Proceedings of the AISTech
2019 Conference & Exposition, Pittsburgh, USA, 2019, pages 547–560.
257
Huang, D.; Andrade, M.; White, D.: Blast Furnace Above–Burden Infrared
Camera. Iron & Steel Technology, No. 3, 2017, pages 44–56.
Raipala, K.: On Hearth Phenomena and Hot Metal Carbon Content in Blast
Furnace. PhD Thesis, Helsinki University of Technology Publications in
Material Science and Metalurgy, Espoo, 2003.
Lüngen, H.B.; Schmöle, P.: Comparison of Blast Furnace Operation Modes

in the World. Proceedings of the 8th International Congress on Science and
Technology of Ironmaking, Vienna, Austria, 2018, pages 1054–1063.
Nightingale, R.J.: The Development of Hearth Voidage Estimation and Deadman

Cleanliness Index. PhD Thesis, University of Wollongong, 2000.
Opbergen, R. van; Toxopeus, H.: Modelling the Taphole and Casting Practice.
Proceedings of the 27th McMaster Symposium on Iron and Steelmaking,
Hamilton, Canada, 1999.
Pagter, J. de; Molenaar, R.: Taphole Experience at BF6 and BF7 of Corus Strip
Products IJmuiden. Proceedings of the 29th McMaster Ironmaking Conference,
Hamilton, Canada, 2001, pages 107–125.
Pardo, N.; Moya, J.A.; Vatopoulos, K.: Prospective Scenarios on Energy Efficiency
and CO₂ Emissions in the EU Iron & Steel Industry. EUR Scientific and
Technical Research Series, Publications Office of the European Union, 2012.
Peacy, J.G.; Davenport, W.G.: The Iron Blast Furnace – Theory and Practice.
Pergamon Press, Oxford, United Kingdom, 1979.
Rist, A.; Bonnivard, G. – Revue de Métallurgie, Vol. 60, 1963, pages 23–38.
Rist, A.; Meysson, N. – Revue de Métallurgie, Vol. 61, 1964, pages 121–46.
Rist, A.; Meysson, N.: A Dual Graphic Representation of the Blast Furnace
Mass and Heat Balances. Proceedings of the 25t Ironmaking Conference,
Philadelphia, USA, 1996, pages 88–98.
Rist, A.: Solving Simple Blast Furnace Problems by Means of the Operating
Diagram. Blast Furnace Ironmaking, 1977, pages 4/1–4/37.
Santos Assis, P.; Batista Vieira, C.; Nolasco Sobrinho, P.J.: New Developments
for Powder Coal Injection into the Blast Furnace. Steel Research International,
Vol. 75, Issue 4, 2004, pages 235–239.
Schoone, E.E.; Toxopeus, H.L.; Vos, D.: Trials with a 100% Pellet Burden.
Proceedings of the 54t Ironmaking Conference, Nashville, USA, 1995, pages
465–470.
258 Annexes
Sergeant, R.; Bonte, L.; Huysse, K.; Daelman, A.: Hearth Management at
Sidmar for an Optimal Hot Metal and Slag Evacuation. Proceedings of the 5t
European Coke and Ironmaking Congress, Stockholm, Sweden, Vol. 2, 2005.
Singh, B.N.; De, A.; Rawat, Y.F.; Das, R.C.; Chatterjee, A. – Iron and Steel
International, 1984, pages 135ff.
Song, Q.: Effect of Nut Coke on the Performance of the Ironmaking Blast Furnace.
PhD Thesis, Delft University of Technology, Delft, The Netherlands, 2013.
Steer, J.M.; Marsh, T.R.; Sexton, D.; Greenslade, M.: A Comparison of Partially
Burnt Coal Chars and the Implications of their Properties on the Blast Furnace
Process. Fuel Processing Technology, Vol. 176, 2018, pages 230–239.
Toxopeus, H.L.; Steeghs, A.G.S.; Van den Boer, J.F.C.: PCI at the Start of the
21st Century. Proceedings of the 60t Ironmaking Conference, Baltimore, USA,
2001, pages 736–742.
Yaniga, S.E.; Francis, W.P,: Tuyere Life Improvements and Operating

Enhancements on Great Lakes B–2 Blast Furnace. Proceedings of the AISTech
2010 Conference & Exposition, Pittsburgh, USA, 2010, pages 499–507.
Yu, J.; Lucas, J.A.; Wall, T.: Formation of the Structure of Chars during
Devolatilisation of Pulverized Coal and its Thermoproperties: A Review. Progress
in Energy and Combustion Science, Vol. 33, Issue 2, 2006, pages 135–170.
Vander, T.; Alvarez, R.; Ferraro, M.; Fohl, J.; Hofherr, K.; Huart, J.; Mattila,
E.; Propson, R.; Willmers, R.; Van der Velden, B.: Coke Quality Improvement
Possibilities and Limitations. Proceedings of the 3rd International Cokemaking
Congress, Gent, Belgium, 1996, pages 28–37.
Ward, N.; Klaas, M.; D’Alessio, J.; Badglet, P.: Blast Furnace Process Monitoring
and Control Through the Use of Tuyere Camera Technology. Iron & Steel
Technology, March 2018.
259
Index
Active coke zone 13
Adherences to furnace wall 211
Alkali 208
Asymmetry, circumferential 114
Asymmetry, radial 113
Asymmetry, tuyere without PCI 115
Banded iron formation 24

Basicity 155
Blast furnace contents, quantitative 20
Blast furnace cooling systems 9
Blast furnace efficiency 109
Blast furnace process, countercurrent 13
Blast furnace process, gas from coke and coal 11
Blast furnace process, general 3
Blast furnace process, internal layer structure 14
Blast furnace profile 15
Blast furnace temperature profile 5
Blast furnace, defintions of volume 7
Blow down 219
Blow–in, empty furnace 221
Blowpipe failure 216
Boudouard reaction 235
Briquettes 41
Burden calculation 88
Burden composition, worldwide 190
Burden descent, erratic 120
Burden descent, vertical forces 122
Burden distribution, cohesive zones 125
Burden distribution, fines 129
Burden distribution, high pellet operation 196
Burden distribution, ideal 128
Burden distribution, nut coke 129
Burden distribution, two types sinter 130
Burden fines 197
Burden movement 118
Burden optimization 195
Burden profile measurement 232
260 Index
Burden selection 190

Burden, cold strength 29
Burden, fines 26
Burden, interaction components 42,43
Burden, processing the 17
Burden, quality demands 26
Burden, quality tests 28
Burden, reducibility 30
Burden, reduction–disintegration 30
Burden, softening and melting 28, 193
Camera, top 231

Camera, tuyere 231
Carbon, reactions 235
Casthouse operation 161
Casthouse operation monitoring 167
Casthouse simulation 169
Casting delays 178
Central working process 126
Channelling 207
Charges, number in blast furnace 21
Chemical reserve zone 238
Chilled furnace 224
Chute rotation direction 116
Coal blending for injection 74
Coal gasification 74
Coal gasification after injection 75
Coal gasification, incomplete 79
Coal injection, circumferential symmetry 80
Coal injection, coal backing up 81
Coal injection, equipment 70
Coal injection, process control with 76
Coal injection, replacement ratio 71
Coal injection, thermal control 79
Coal–oxygen working area 77
Coal quality for injection 72
Coal volatile matter 72
Coal–gas coinjection 85
Cohesive zone 13
Coke deadman 13
Coke deadman 58
Coke deadman cleanliness 60
Coke degradation 51
Coke layer 133
Coke mess 226
Coke quality 50
Coke quality parameters 63
261
Coke quality tests 248

Coke rate, minimum 134
Coke size 54
Coke strength 55
Coke use in furnace 101
Coke, Cfix 49
Coke, function 47
Coke, hot strength 57
Coke, movement towards tuyeres 164
Coke, nut 65
Coke, reactivity 57
Coke, residence time 48
Coke, screening 65
Coke, stockyard 64
Coke, voidage 65
Control , gas flow 189
Control hot blast 188
Control, chemical ~ burden 44
Control, slag basicity 187
Control, thermal 186
CRI 57
CSR 57
Direct reduction 92
Direct reduction 238
Direct reduction estimation 94
Double bell furnaces 136
Dry furnace condition 166
Efficiency, furnace: see utilization, gas

Enthalpy balance, see Heat balance
Expert systems 242
Flame temperature 96
Flame temperature estimation 97
Flooding limits 147
Gas cleaning 9
Gas composition in blast furnace 98
Gas composition in blast furnace, measured 191
Gas flow through furnace 16
Gas reduction 236
Greenhouse gas emissions 226
Hanging and slipping 205

Harmonic and average mean size 51
Hearth wear model 247
262 Index
Heat balance 99
Hematite 17
Hot metal formation 147
Hot metal production, global 2
Hot metal temperature 187
Hot metal, properties of high silicon ~ 151
Hydrogen from injectants 85
Hydrogen utilization 241
Hydrogen, reducing iron oxide 240
Injection, coal, natural gas, oil 67

Instrumentation 229
Iron ore 24
K–factor 103
Kinetic energy blast 105
Liquid level and process 162

Lump ore 40
Magnetite 17
Melting of iron ore 145
Metallic charge 41
Mixed layer 132
Moisture input with burden and coke 200
Moisture recirculation 203
Momentum, blast see Kinetic energy
Monitoring drying capacity 204
Natural gas injection 82

Natural gas injection, thermal control 83
Natural gas oxygen working area 83
Ore free center 128

Ore layer 133
Oxygen lancing 180
Pellet chemical composition 36

Pellet quality 35, 39
Pellet size 37
Pellet, cold compression 37
Pellet, swelling 38
Pellets, acid 36
Pellets, fluxed 36
Pellets, olivine 36
Permeability and max ΔP 107
Permeability blast furnace 102
263
Permeability, cohesive zone 137

Permeability,raw materials 102
Permeabilty and PCI rate 107
Pressure difference 105
Pressure difference and top temperature 106
Pressure taps 230
Primary slag 101, 144
Processing time burden 21
Processing time gas 22
Production rate 183
Productivity 111
Productivity, conditions for high 112
Qcrack 71
Raceway structure 12
RAFT, see Flame temperature
Reduction disintegration 142
Reduction ferrous burden, progress 193
Reduction iron oxides 236
Reduction reactions 142
Reference furnace data 233
Replacement Ratio 71
RIST diagram 250
Rubble ore: see lump ore
Rules of thumb 234
Scaffolds to furnace wall 212

Scrap, charged in blast furnace 41
Separation hot metal and slag 180
SFCA 33
Silicon 149
Sinter, Al2O3 33
Sinter, basicity and productivity 34
Sinter, fluxes and quality 33
Sinter, quality 31
Slag design for stops and starts 158
Slag formation 101
Slag liquidus temperatures 157
Slag liquidus temperatures, basicity, FeO 144
Slag viscosity 156
Slip, mechanism 206
Softening and melting burden 194
Softening and melting burden components 193
Softening ferrous burden 143
SOMA 229
Steel production, global 1
264 Annexes
Stockhouse management 200

Stockyard materials 199
Stopping a blast furnace 218
Stops and starts 217
Stove changes and pressure 120
Sulfur 152
Sulfur distribution 154
Symmetry, process 112
Taphole 172
Taphole clay 174
Taphole drilling 176
Taphole face 176
Taphole plugging 175
Temperature profile and reactions 91
Thermal reserve zone 238
Top gas composition, calculation 89
Total energy approach for gas injection 84
Tracking, charge 246
Tuyere blockage by scabs 216
Tuyere failures 213
Tuyere, body burnt 214
Tuyere, nose burnt 214
Tuyere, tipping 215
Utilization, gas 18
Variability charging rate 203

Voidage, creation of 118
Wall working process 127

Water discipline 213
Watergas shift reaction 240
Wind velocity in tuyeres 104
Wustite 17
Zinc 210

M. Geerdes, R. Chaigneau, O. Lingiardi, R. Molenaar, R. Van Opbergen, Y. Sha, J. Warren - Modern Blast Furnace Ironmaking - An Introduction-IOS Press (2020)

Uploaded by

Copyright:

Available Formats

M. Geerdes, R. Chaigneau, O. Lingiardi, R. Molenaar, R. Van Opbergen, Y. Sha, J. Warren - Modern Blast Furnace Ironmaking - An Introduction-IOS Press (2020)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

M. Geerdes, R. Chaigneau, O. Lingiardi, R. Molenaar, R. Van Opbergen, Y. Sha, J. Warren - Modern Blast Furnace Ironmaking - An Introduction-IOS Press (2020)

Uploaded by

Copyright:

Available Formats

Modern

Fourth Edition, 2020

ISBN 978–1–64368–122–1 (print)

Published by IOS Press under the imprint Delft University Press

Compared to the third edition, we have added operational examples of many

We learn by sharing our knowledge. We wish the same to our readers.

Maarten Geerdes, Rénard Chaigneau, Oscar Lingiardi, Ron Molenaar,

Chapter I Introduction to the Blast Furnace Process 1

Chapter II Blast Furnace Process and Internal Structure 11

Chapter III The Ferrous Burden 23

Chapter V Injection of Coal, Natural Gas and Oil 67

Chapter VI Blast Furnace Gas and Processing the Ferrous Burden 87

Chapter VII Burden Distribution 117

Chapter VIII Hot Metal and Slag 141

Chapter IX Casting Hot Metal and Slag 161

Chapter X Control of Blast Furnace Operationand Burden Selection 183

Chapter XI Operational Challenges 197

Annex I Blast Furnace Instrumentation 229

List of Symbols and Abbreviations

Figure 1.1 World crude steel production (World Steel Association)

The blast furnace–oxygen steelmaking route mainly produces ﬂat products,

Crude Steel Production and

Metallic Charge (Mtpa)

Figure 1.2 World crude steel production and metallic charge

1.2 Hot metal production

Steel production based on hot metal from blast furnaces contributes

1.3 Blast furnace process

Figure 1.4 Input and output of a blast furnace

As a consequence of the gas movement, there are large temperature diﬀerences

Figure 1.6 Temperature proﬁle and various zones in a blast furnace

1.4.1 Equipment overview

Figure 1.7 Blast furnace general arrangement

Figure 1.8 Blast furnace top charging by bell–less top

1.4.2 Blast furnace and its development

Bottom Bottom Bottom of

Figure 1.9 Deﬁnitions of working volume and inner volume

1.4.3 Cooling systems

Figure 1.10 Cooling systems: plate coolers, staves and hybrid

1.4.4 Gas cleaning

temperature can be a concern, since high productivity in a blast furnace can

1.5 Book overview

Blast furnace ironmaking can be discussed from three diﬀerent perspectives:

Throughout the book, examples are based on an operating “reference furnace”: a

Figure 2.1a The raceway, horizontal and vertical sections

Figure 2.1b The raceway after a blowdown and nitrogen quench

2.2 Counter–current reactor

Figure 2.2 The blast furnace as a counter–current reactor

2.3 Internal layer structure

Figure 2.3 Blast furnace zones

As indicated in Figure 2.3, an operating blast furnace contains:

Figure 2.4a Dissections of quenched blast furnaces Kakogawa 1 and Tsurumi

Figure 2.4b Cohesive zone left after blow–down

Example 2.1 Furnace proﬁle and burden and coke layers

This cylinder represents about 50 % of the total working volume of the

2.4 How gas ﬂows through the furnace

2.5 Processing the ferrous burden

Figure 2.7 Schematic presentation of reduction of iron oxides and temperature

The second step in reduction is magnetite (Fe3O4) to wustite (FeO). This

2.6 Furnace efﬁciency