evolution of coke properties while descending through a ...€¦ · coke has ancient origins and...

LICENTIATE T H E S I S

Luleå University of TechnologyDepartment of Chemical Engineering and Geosciences, Division of Process Metallurgy

:|: -|: - -- ⁄ --

:

Evolution of Coke Properties while Descending through a Blast Furnace

Tobias Hilding

Evolution of Coke Properties while Descending Through a Blast Furnace

by

Tobias Hilding

Licentiate Thesis

Luleå University of Technology Department of Chemical Engineering and Geosciences

Division of Process Metallurgy SE-971 87 Luleå

Sweden

2005

Evolution of coke properties while descending through a blast furnace

Tobias Hilding, Div. of Process Metallurgy, LTU, 2005

1

ACKNOWLEDGEMENTS

I would like to express my gratitude to Professor Björkman and Professor Jan-Olov

Wikström for their supervision and for giving me the opportunity to perform my

research.

Special thanks to Professor Veena Sahajwalla for her supervision and support and Dr

Sushil Gupta for all help and discussion.

Also special thanks to Dr Lars Bentell for fruitful discussions and help.

Thanks to the members of committees JK21057, JK21060 and RFCS 7210-PR-324.

Further thanks to all the employees at Luleå University of Technology, in particular

my colleague Ryan Robinson for the good laughs and business lunches. Also, thanks

to my colleagues at MEFOS and employees at University of New South Wales who

have helped me throughout my studies.

A great amount of thanks to my parents, brother, relatives and my mates. Deep thanks

to my Luleå-love JK, you are the best! We did it!

I would like to especially acknowledge the Swedish Energy Agency, STEM, and

JERNKONTORET for financial support and LKAB, SSAB and Ruukki for supplying

research input.



2

ACKNOWLEDGEMENTS..............................................................................1

LIST OF PAPERS.............................................................................................4

SUMMARY .......................................................................................................6

1. INTRODUCTION .......................................................................................8

1.1 BACKGROUND............................................................................................................................................. 81.2 STATE OF THE ART ................................................................................................................................... 111.3 OBJECTIVES .............................................................................................................................................. 171.4 RESEARCH QUESTIONS............................................................................................................................. 18

2. METHODS.................................................................................................18

2.1 THE EXPERIMENTAL BLAST FURNACE................................................................................................... 182.2 THE STUDIED EBF CAMPAIGNS ............................................................................................................... 202.2.1 EBF campaigns followed by excavation ................................................................................................ 202.2.2 EBF trial with high CRI coke ................................................................................................................. 242.3 METHODS USED FOR CHARACTERIZATION OF COKE SAMPLES............................................................. 252.3.1 TGA/DTA–MS....................................................................................................................................... 262.3.2 CRI/CSR EQUIPMENT ............................................................................................................................. 272.3.3 SIEVING................................................................................................................................................... 292.3.4 X-RAY DIFFRACTION............................................................................................................................... 292.3.5 CHEMICAL ANALYSES............................................................................................................................. 292.3.6 SCANNING ELECTRON MICROSCOPE ...................................................................................................... 302.3.7 LIGHT OPTICAL MICROSCOPE................................................................................................................. 302.3.8 BET......................................................................................................................................................... 312.3.9 MICRO TEXTURE MEASUREMENT............................................................................................................ 31

3. RESULTS AND DISCUSSION ................................................................32

3.1 VARIATION IN PHYSICAL PROPERTIES ................................................................................................... 323.2 EVOLUTION OF CARBON STRUCTURE ..................................................................................................... 353.3 ALKALI UPTAKE AND DISPERSION IN COKE............................................................................................ 373.5 EVOLUTION OF COKE REACTION WITH CO2 .......................................................................................... 453.6 ISOTROPIC / ANISOTROPIC CHANGES IN THE COKE CARBON MICRO STRUCTURE ............................... 503.7 TRIAL WITH HIGH CRI COKE .................................................................................................................. 503.7.1 PROCESS ANALYSIS................................................................................................................................. 513.7.2 EVOLUTION OF CARBON STRUCTURE...................................................................................................... 523.7.3 EVOLUTION OF COKE ASH CHEMISTRY ................................................................................................... 533.7.4 EVOLUTION OF COKE REACTION WITH CO2 ............................................................................................ 553.7.5 ISOTROPIC / ANISOTROPIC CHANGES IN THE COKE CARBON MICRO STRUCTURE.................................... 573.7.6 POROSITY DIFFERENCES ......................................................................................................................... 58

4. CONCLUSIONS .........................................................................................59



3

4.1 EVOLUTION OF COKE CARBON STRUCTURE ........................................................................................... 594.2 ALKALI IMPLICATIONS ............................................................................................................................ 594.3 COKE REACTIVITY................................................................................................................................... 594.4 PHYSICAL PROPERTIES INCLUDING COKE STRENGTH AND ABRASION ............................................... 60

5. FUTURE RESEARCH..............................................................................62

5.1 COKE DEGRADATION.................................................................................................................... 625.2 OPTIMUM COKE PROPERTIES ...................................................................................................... 62

6. LIST OF ABBREVIATIONS ...................................................................63

7. REFERENCES ..........................................................................................63



4

LIST OF PAPERS

The outcome of project JK21057 “Coke Strength at High Temperatures”, that this

thesis is based on is a literature review, two conference proceedings and two journal

papers.

I Hilding, T., Sahajwalla, V., Gupta, S.K., Björkman, Bo, Sakurovs, R.,

Grigore, M., Saha-Chaudhury, N. Study of Gasification Reaction of

Cokes Excavated From Pilot Blast Furnace. Scanmet II, 2004, Luleå,

Sweden.

T. Hilding’s contribution to this publication were as a participant in

excavation of the EBF and investigation of changes of coke from the EBF

utilizing TGA, LECO, and XRD.

II Tobias Hilding, Nouredine Menad, Bo Björkman and Jan-Olov

Wikström. Thermal Analysis of Coke From Different Layers in an

Experimental Blast Furnace. Submitted to Thermochimica acta, 2005

T. Hilding’s contribution to this publication was as a participant in

excavation of the EBF and conduction of all experimental work.

III Tobias Hilding, Sushil Kumar Gupta, Veena Sahajwalla. Effect of Carbon

Structure and Coke-Alkali Reactions on the Coke Behaviour in an

Experimental Blast Furnace. Submitted to ISIJ, 2005

T. Hilding’s contribution to this publication was as a participant in

excavation of the EBF and conduction of all experimental work.

IV Tobias Hilding, Jan-Olov Wikström, Urban Janhsen, Olavi Kerkkonen.

Investigation of coke properties while descending through an

experimental blast furnace. Submitted to ECIC 2005



5

T. Hilding’s contribution to this publication was as a participant in probe

material and tuyere core sampling from the EBF and X-ray diffraction

and TGA measurements.

Apart from the supplements above, the following papers have been published during

the thesis work:

Veena Sahajwalla, Tobias Hilding, Anne von Oelreich, Sushil Kumar

Gupta, Bo Björkman, Jan-Olov Wikström, Patrick Fredriksson and

Seshadri Seetharaman. Structure and Alkali Content of Coke in an

Experimental Blast Furnace and Their Gasification Reaction. AIST 2004

T. Hilding’s contributions, see I.

Tobias Hilding, Kelli Kazuberns, Sushil Gupta, Veena Sahajwalla,

Richard Sakurovs, B. Björkman and Jan-Olov Wikström. Effect of

Temperature on Coke Properties and CO2 Reactivity under

Laboratory conditions and in an Experimental Blast Furnace. AIST 2005

T. Hilding was responsible for generating most of the data except

laboratory annealing measurements.



6

SUMMARY

Due to increasing price and economic pressures, there is a need to minimise coke

consumption. The lesser amount of coke used has indirectly set higher standards for

coke quality and lead to a wish for even more knowledge about its function in the

blast furnace.

Over the last 20 years, coke quality has been strongly dictated by the so-called CSR

value because it was believed that a higher CSR leads to improvement in productivity

and more stable operation. Due to lack of suitable coals, often cokes are made from

coals with relatively inferior quality leading to coke with lower values of the so-

called CRI indicia. Because of this, there was an indirect focus on cokes with lower

CRI values. Therefore, this thesis will address some of the important issues of coke

strength and focus on changes occurring with coke when it passes through a blast

furnace. The main aim of this study is to understand the degradation mechanisms and

reactivity changes of coke in order to investigate the factors that affect coke quality.

Cokes excavated from LKAB’s Experimental Blast Furnace (EBF) are used as a basis

for the research. Two campaigns with similar coke (low CRI/high CSR) but different

blast furnace injection material have been studied. The coke is supplied from SSAB

Tunnplåt Luleå AB. Physical and chemical properties of cokes samples from the EBF

were measured. Evolution of coke properties particularly carbon structure and alkali

uptake were related to CO2 reactivity as well as coke behaviour (e.g. CSR/abrasion).

In addition to this, a trial with very high CRI coke was studied. On the basis of this

study, following conclusions were made.

1. The order of carbon structure and concentration of alkali species were

increased and these were the most notable changes in the coke properties as it passed

through the shaft to the cohesive zone of the EBF.

2. The degree of graphitisation was increased while amorphous carbon

content was decreased in the hotter zones of the EBF. A linear correlation between



7

the height of the carbon crystallite (Lc) values and the coke bed temperature was

established to demonstrate the strong effect of temperature on the carbon crystallite

value (Lc) of coke in the EBF.

3. The alkali concentration of coke increased with increasing temperature of

the coke bed such that most of the alkali content was evenly distributed in the bulk of

the coke rather than in the periphery of the coke matrix.

4. The CO2 reactivity of coke was found to increase during progressive

movement of the coke from shaft to cohesive zone of the EBF, and was related to the

catalytic effect of increased alkali concentration in coke.

5. The deterioration of coke quality in the EBF, particularly coke strength

(CSR) and abrasion propensity (I drum test), was related to coke graphitisation,

alkalization and reactivity to demonstrate the strong effect of the coke graphitisation

on the propensity of coke degradation.

6. Differential Thermal Analysis indicated that reactions with CO2 are

enhanced as coke descends through the EBF.

In addition, a trial period with poor coke quality was studied by extensive sampling.

The results from this study gave the following additional conclusions:

7. Comparison between high and poor quality coke indicate structure to be

connected with alkali uptake, reaction with CO2 and degradation.

8. Isotropic coke carbon components are more resistant than anisotropic

components when passing through the EBF.

9. Both cokes develop a more ordered structure as they descend through the

EBF.



8

1. INTRODUCTION

1.1 Background

Coke has ancient origins and carbonisation of coal is mentioned in text as early as

371 BC. However, coke use as a sole source of fuel in a blast furnace began from

somewhere between the early to mid 1800th century. This coke was made in piles [1].

The knowledge of coke and its properties was lacking in the beginning of the coke

era. The higher demands incurred for better pig iron led to higher demands on the

coke.

The last decade, three consistent themes have appeared pertaining to coke properties

and blast furnace performance. They are related to the viability of the blast furnace,

improvement in blast furnace productivity and efficiency, and blast furnace

operations at lower coke rates.

The most consistent theme of recent literature is that the blast furnace will remain a

dominant method for production of hot metal worldwide [2-9]. Another theme shared

throughout the world relates to significant improvements in blast furnace productivity

[10].

A third common theme relates to coke replacement at the furnace with reductant

injection such as pulverized coal, natural gas and oil. However, coke is essential for

the blast furnace iron making process in order to support the burden and provide gas

permeability, thus a minimum coke burden limit exists.

Coke production has, since the last two decades, gone through some major changes.

The number of aging coke plants steadily increases while very few new plants are

being built, except in China. The coke export from China has however decreased due

to domestic usage. Prices of external coke have since the beginning of 2003 to mid

2004 increased by more than 400 %.



9

The most dominant hot metal making process in the world today is still the Blast

Furnace (BF) process, and the most important raw material fed into the BF, in terms

of operation efficiency and hot metal quality, is coke. Due to a decrease in the coke

supply and a desire to lower the energy consumption and to reduce CO2 emissions,

developments in the BF sector have long focused on replacing the coke by coal. One

of the major developments in the blast furnace operation is the introduction of

pulverized coal technology in which coke is substituted by Pulverized Coal Injection

(PCI) through the tuyeres. This technique was introduced in the early 1980’s.

Economic and environmental pressures are the primary driving force behind the

promotion of PCI technology. The old coking plants are gradually closing while few

new plants are being built to replace the coke supply, particularly in developed

countries, including Europe. New coke plants are extremely expensive due to

stringent environmental regulations. Therefore, in the future, blast furnace operations

will rely on less coke addition per unit hot metal production. During BF operation at

low coke rates, the coke experiences prolonged residence time. Regardless of

residence time, the coke must maintain satisfactory bed permeability for reducing

gases to flow upwards in the furnace and for liquids to flow downwards. Therefore,

high quality coke is essential for future blast furnace operations.

Coke is produced by heating a coal blend in the absence of oxygen. The most

common type of production technique is the so-called conventional or by-product

coke plant, see Figure 1. They are comprised of horizontal chamber ovens, measuring

12 to 18 m long, 3 to 8 m tall, and 0.4 to 0.6 m wide. Several chambers are grouped

to form one battery (Multi-Chamber-Systems). A single battery may consist of up to

85 ovens. The coal mix is charged through charging holes in the oven top. Following

15 to 25 hours coking time, the doors are opened and coke is pushed by the coke

pusher machine out of the oven into a coke quenching car. The coke is then cooled.

The oven chamber is again sealed, initiating a new carbonisation cycle. The gas

evolving on coal carbonisation enters gas treatment facilities and the by-product



10

recovery plant. The ovens are run with a slight over-pressure. The coke reaches a

temperature of approximately 1100°C to 1250°C.

Other types of coke

production techniques are

heat recovery coke plants and

non-recovery coke plants.

The heat recovery plants

utilize all the excess gas to

produce heat. The furnaces

are typically called Beehive

furnaces and work with

negative pressure and require

coking times of up to 48

hours.

Coke performs three functions in a blast furnace namely: a thermal function, as fuel

providing the energy required for endothermic chemical reactions and for melting of

iron and slag; a chemical function, as reductant by providing reducing gases for iron

oxide reduction; a mechanical function, as a permeable grid providing for passage of

liquids and gases in the furnace, particularly in the lower part of the furnace. When

coke passes through a blast furnace, the coke degrades and generates fines which

affect bed permeability and affects the process efficiency. The rate at which coke

degrades is mainly controlled by the solution loss reaction, thermal stress, mechanical

stress and alkali accumulation.

Coke quality is often characterized by measuring cold and hot strength, ash

composition and chemistry, which are largely dictated by coal properties. A range of

laboratory tests and procedures have been developed to characterize physical and

chemical properties of coke and their potential impacts in the blast furnaces. The

Figure 1. Illustration of a typical coke plant of the

conventional kind.



11

most often used and well-known tests are the Coke Reactivity Index (CRI) and the

Coke Strength after Reaction (CSR) developed by Nippon Steel Corporation (NSC)

in Japan in the early seventies, in order to assess the effect of CO2 reactions on coke.

There is no universally accepted standard procedure, however NSC/CRI test is

widely recognized around the world and was adopted by ASTM while being

considered for ISO standard [11, 12]. Generally high CSR coke is believed to prevent

the coke from breaking down, improve the permeability of gas and liquid and

increase the productivity as well as decrease the specific coke consumption of the BF

[13].

1.2 State of the art

No international agreement of an ideal way to determine the quality exists as each

industry relies on their empirical experience for the interpretation. These laboratory

tests are designed to test the coke properties under specific set of conditions which

might not be universally suitable. The reproducibility of CRI/CSR values among

different laboratories also varies considerably [14]. Whether the reactivity constitutes

an important factor in determining blast furnace performance has been a subject of

some controversy during the past decades. Some investigators suggest that most of

the reactions involving coke tend to take place in the high temperature zone of the

blast furnace, where diffusion or mass transfer are rate limiting and the mechanical

strength or integrity of the coke was thought to be the significant factor. Others say

coke reactivity is one of the most important factors which control the permeability

and that the lower the coke reactivity the higher is the permeability of the burden.

Coke reactivity in itself might possibly not play a very important role, but the manner

in which the coke reacts could markedly influence its degradation characteristics and

hence the performance of the furnace as a whole [15].



12

Consumption of coal matter in coke has an impact on particle porosity. Therefore a

strong interaction exists between the chemical reactivity of coke and its remaining

mechanical strength. The test conditions for CRI and CSR do not truly simulate the

blast furnace and are too severe (time, temperatures and exposure of coke to CO2),

although actual field trials have indicated some correlation between the test and the

blast furnace process [16]. However, the CRI/CSR test has the limitations of a single

point test on coke, and includes poor reproducibility and also variable starting

material, varying porosity and particle surface area, and variability in shape and size

[10]. Also important to point out, is that coke is a very inhomogeneous material thus

making it difficult to characterise. Despite some results which counter a general

linear correlation between CSR and CRI, normally low CRI-values lead to high CSR

values. Coke reactivity is mainly influenced by the aging and the maceral

composition of the coal leading to isotropic or anisotropic coke structures (the

isotropic components are more reactive towards CO2), by the ash composition as well

as the carbonisation conditions. From the view of product quality and corresponding

behaviour in the blast furnace, an optimum has to be found between coke CSR, CRI

values and the carburization of the hot metal [17].

Coke reactivity is influenced by physical properties, including porosity as well as

chemical properties including coke minerals and carbon structure. Reactions with

oxidising gases affect the porous carbon matrix during combustion/gasification. As

coke descends in a BF, its chemical structure is expected to change. The evolution of

pore structure by growth and coalescence leads to increasing or decreasing available

surface areas, changes in pore structure/distribution, gas diffusion and reactivity.

Porous structure of coke is governed by the coking properties of coals, particularly by

maximum fluidity and swelling number [18].

Transformations of inorganic matter upon heat treatment include changes in chemical

bonding, sintering, melting and vaporization as well as mutual interactions with

organic matter. In addition to the catalytic affect on reactivity of carbonaceous



13

materials, high temperatures affect particle size of mineral matter and hence the

fragmentation and mechanical stability of the carbonaceous material. Hermann [17]

has evaluated the effect of chemical composition of coal ash on coke reactivity such

that CaO and SO3 are gasification stimulating components, Fe2O3 an Al2O3 have an

intermediate effect, and P2O5, TiO2, MgO are gasification-inhibiting. Feng et al [19]

have observed that iron is a major catalyst during gasification of bituminous coal as

well as resulting in organised crystalline structures of carbon in the vicinity of the

carbon/iron interface. With increasing burnout, mineral matter could have inhibiting

effect by forming a barrier for oxidizing gases that could influence carbon reactivity

[20].

During its descent through a blast furnace, coke is exposed to extreme reacting

conditions. The prevailing high temperatures in the cohesive zone areas lead to coke

graphitisation i.e. increased ordering of carbon structure. Synthetic graphite has a

highly ordered structure, high fixed carbon content with low levels of ash and volatile

matter. Graphite structure can be described by a regular, vertical stacking of

hexagonal aromatic layers with the degree of ordering characterised by the vertical

dimension of the crystallite Lc, see Figure 2. Each C atom within the aromatic layer

(basal plane) is linked through covalent bonds to three C atoms. However, bonding

between the layers is very weak and can easily be broken by external forces. Natural

graphite has highly ordered structure like synthetic graphite but contains high level of

impurities. The Lc for coal/char/coke can be measured by using X-ray diffraction

profiles [21]. The carbon structure is often believed to influence the carbon reactivity

[22].

As the coke descends through the

blast furnace it is initially dried by

the ascending hot gases. At

temperatures in the 800 – 850 °C

regions, alkali carbonate

compounds are deposited on the

Figure 2. A schematic of crystal structure of graphite.



14

coke surface, causing an increase in reactivity, but do not affect coke size or strength

[23].

Helleisen et al reported that potassium decreases the gasification threshold

temperature from the classical Figure of 950 °C down to 750 – 850 °C, depending on

the amount of potassium and the nature of coke.

When the temperature increases further to 900 – 950 °C, the carbon solution loss

reaction commences and any carbon dioxide produced by the gaseous reduction of

the iron oxides is immediately converted back to carbon monoxide.

The chemical reaction considered as most important is the solution loss reaction,

2CO(g)(g)2COC(s)+ , which normally starts at temperatures around 900 – 1000

°C. Alkalis, in particular potassium, enhance the solution loss reaction significantly

and the reaction starts at considerably lower temperatures by a catalytic effect of the

alkalis [24, 25].

Already in the early 1980’s, Japan raised interest for coke quality at high

temperatures. In order to clarify the degradation of coke in the blast furnace, a series

of fundamental studies on the degradation due to chemical, mechanical and thermal

effects were carried out. The work was based on probe samples and dissections. The

conclusions were as follows;

• When post-reaction strength decreased, the permeability became lower due to a

large amount of fine coke depositing in the lower part of the furnace.

• The tuyere flame temperature and blast velocity have a great influence on the

degradation of coke. Under the higher flame temperature, the cracking of coke

caused by the thermal stress occurs easier. An optimal blast velocity exists to

prevent an inactive dead man and the degradation of coke.



15

• The fines originating from coke in the lower part of blast furnace accumulate in

the dead man or travel upward in the furnace. The generation of coke fines is

dependent of the coke strength [26].

Below the cohesive zone the temperature of the coke increases to above 1500 °C.

Coke in the mobile bosh zone (between the cohesive zone and the stagnant deadman

coke) feeds the raceway. This coke is subjected to extremely rapid heating (up to

approx. 2200 °C), combustion and mechanical action in the hot blast. The decrease of

coke rate at high levels of coal injection would lead to higher degradation resulting

from thermal action. The catalytic graphitisation of the coke lump surface by iron and

slag derived from injected coal might also lead to reductions in coke abrasion

resistance [23].

Dissections and probing have indicated a rather complete vaporisation of potassium

in the raceway area, and a sharp rise of potassium towards the centre of the furnace.

The K2O content in coke ash may reach values as high as 30 % in the centre of the

dead man. Alkali distribution in coke is clearly a consequence of the thermal

conditions prevailing along the radius. In the raceway, temperature is the highest, and

alkalis are completely vaporized. In the centre of the furnace, at tuyere level, lower

temperatures exist, promoting the deposition of alkalis on the condensed phases, coke

and slag [25, 27], [28].

Although investigations of cokes from dissected furnaces have provided relationships

between alkali pick-up and coke properties, the actual mechanisms of alkali attack,

and in particular the effect of time of exposure to alkalis, are uncertain [23]. The zone

of maximum alkali pick-up and coke strength reduction is situated near the cohesive

zone.

The fact that coke reactivity in the blast furnace is strongly connected to the alkali

content of coke has been revealed by dissections. Besides a weakening of the pore



16

walls in the coke by the solution loss reaction, which is influenced catalytically by

alkalis, there are also observations indicating that alkalis by other mechanisms are

able to decrease the coke stability [24].

Studies of the effect of depositing potassium carbonate (K2CO3) and potassium

phosphate (K3PO4) up to 4 % K concentration on BF coke showed that the potassium

clearly increased the reactivity. Porosity measurements and microscopic studies

indicated the reaction to be progressively shifted towards the periphery [28].

The chemical composition of the coke strongly depends on the mineral matter. The

basic elements (Fe, Ca, Mg and alkalis) are included in minerals, which are active at

the CRI test temperature, destroying carbon textures. An increase of the ash basicity

catalyses the coke reactivity.

On the other hand, silicates (Si, Al and alkalis) in coke are inactive during the coal

coking and coke CRI test. Australian and Canadian coals give an increase in ash by

fine quartz or kaolinite dissemination. However, the amount of carbonates and

sulphides decreases. Non-reactive silicate dissemination reduces the micro pore

surface of the coke and delays gas penetration into the coke core. This favours a low

CRI and high CSR [29].

Van der Velden [16] wrote “both iron and alkali matter are good catalysts for coke

gasification. Deposition or condensation of these components on coke particles in the

shaft may therefore enhance coke gasification. However, carbon dioxide

concentrations are still very limited and deposition is mainly on the particle periphery

thus no extra pressure is developed on the bulk strength of the coke.”

Coke gasification in the BF preferentially occurs on the coke’s surface. This suggests

that the specifications for reactivity and post-reaction strength of BF feed coke are



17

somewhat questionable if no account is taken of the presence of alkalis in the furnace

[28].

However, Gudenau reported the contrary. A damage of coke structure by alkalis is

doubted by investigations that did not find a decrease of coke strength even at alkali

contents of 5 % in coke. Although the blast furnace coke consumption undoubtedly

depends on the alkali-input, this phenomenon cannot be explained with changing

CSR and CRI values and that these values are independent of the alkali content of

coke [30].

Helleisen et al wrote, ”potassium may induce dramatic effects on coke strength at

high temperature, even in the case of good quality coke” based on reference and K

enriched coke studies [25].

According to Beppler et al, alkali contents in coke were found to be lower during

PCI. This was explained by a longer residence time of coke in the BF during PCI and

the heavier stress incurred, thus leading to a higher degree of disintegration. At an

injection rate of 200 kg coal/THM, the coke has to perform about 75 % more direct

reduction work. Further, Beppler et al assumed that an alkali content gradient exists

in the coke lump, and that the alkali-rich layer is abraded to a greater extent as a

result of higher stress [31].

1.3 Objectives

The objectives of this thesis are to;

a) Develop understanding of coke properties and its behaviour in blast furnace,

b) Build-up knowledge regarding the changes of coke properties,

c) Attempt to understand the mechanism of changes,

d) Investigate and attempt to assess the significance of CRI&CSR tests to represent

how coke degrades in operating BF,



18

e) Investigate the mode of dispersion behaviour of alkalis, particularly if alkali exists

to a greater extent in the periphery of the coke.

1.4 Research questions

As the coke descends through a blast furnace it experiences fundamental changes in

temperature and atmosphere.

How does the coke change?

Sub-questions to be answered:

In which way is coke degraded?

What factors are affecting the solution loss reaction?

What role does the reactivity play?

What is important, high temperature strength or reactivity?

What affects the strength?

What is the influence of alkali and ash?

2. METHODS

In addition to bench-scale testing, a more comprehensive approach is the pilot-scale

testing of materials under a more realistic industrial environment. Even though these

tests are time consuming and very expensive, data generated in these tests are critical

to provide a comprehensive testing of raw materials such as coke. Coke excavated

from two campaigns was studied in LKAB’s Experimental Blast Furnace (EBF).

Both these campaigns utilised a relatively good quality coke i.e. low CRI (around 20)

and high CSR (around 70). A large number of samples and data were collected during

this campaign. In addition to this, a test with very high CRI and very low CSR was

conducted.

2.1 The Experimental Blast Furnace

A simplified layout of the EBF is shown in Figure 3.



19

The working volume is 8.2 m3,

the hearth diameter is 1.2 m,

and the working height is 5.9

m. It is equipped with three

tuyeres placed at 120-degree

intervals, and both oil and coal

injection can be used, as well

as other injection materials.

Insulating refractories are

installed to minimize heat

losses, and only the bosh area and the tuyeres are water-cooled. The blast is normally

preheated to 1200 °C in a new type of pebble heaters. The EBF can be equipped with

either a bell-type top with moveable armour, or a bell-less top, for burden distribution

control. Two mechanical stock rods monitor

the burden descent and control the charging

of the furnace. The EBF has one tap hole,

which is opened with a drill and closed with a

mud gun. The hot metal and slag are tapped

into a ladle. Probes for temperature

measurements, gas analysis and solid

sampling over the blast furnace diameter are

installed at three different positions, see

Figure 4. To facilitate excavation and repair,

the hearth is detachable and can be separated

from the furnace.

The EBF is run campaign wise and two 6-10 week campaigns normally take place

each year. It has a production rate of about 35-40 thm/day. The normal tap-to-tap

time is 60 minutes and normal tapping duration is 5-15 minutes. Process data are

logged continuously and stored in a database. The data are transferred at regular

Figure 3. The EBF plant and its design.

Figure 4. Illustration of the EBF and the included probe system.



20

intervals to another database from which reports and trend charts are generated and

process calculations are carried out. The coke used has been crushed and sieved to a

fraction of 15-30 mm. After each campaign an excavation is normally performed.

Prior to the excavation, the furnace is quenched with nitrogen with the objective to

stop chemical reactions. The EBF-operation together with the excavation gives an

opportunity to map and understand the changes to coke that occur at different levels

in the furnace.

2.2 The studied EBF campaigns

In the present thesis, high quality cokes from two excavations were studied. In

addition a trial with high CRI and low CSR coke was tested and compared with coke

from a reference period. The evaluation in this test is based on solid sampling of coke

through probing.

2.2.1 EBF campaigns followed by excavation The first campaign took place during

the fall of 2002. This campaign lasted

for almost two months and the furnace

was thereafter quenched with nitrogen

to stop prevailing reactions. A three

week long excavation occurred when

the furnace reached acceptable

temperature.

Two core-drilling events occurred with

success. This was done by removing a tuyere during furnace stoppage and thereafter

inserting a metal cylinder into the furnace to collect burden material. The metal

cylinder is then removed and quenched for later testing. The core was divided into

sections and photographed and then the coke was sampled.

Figure 5. Photo of an upside-down piece

from ferrous burden layer 08.



21

Material probes have been used frequently during the campaign and the material was

sampled.

Feed coke was sampled every week during the campaign. The feed coke has been

analysed for the following parameters;

• Moisture

• Volatile matter

• Ash

• Sulfur, Nitrogen, Carbon and Hydrogen

• CRI & CSR

• Sieve analysis

During the excavation, samples were taken from each coke layer. This was done at

three different locations for each layer i.e. close to the wall of the furnace, at the

centre, and in the intermediate part (between wall and centre). The volume for each

(a) (b)Figure 6 a) Photo of layer 3 and b) photo of layer 25 from inside the EBF,

campaign 11.



22

sample was around 4 litres. Each layer was photographed in four directions (west,

east, north, south), with digital and analogue camera, see Figure 5 and 6.

The depth was measured at five points (west, east, north, south, centre) for each layer.

A mapping of the locations of the coke layers of interest for campaign 10 has been

made, see Figure 7.

Figure 7. EBFC10. Illustration of how a few selected coke layers

were found when the EBF was excavated. The left hand side

displays cross-section from South to North, and the right hand

side, from West to East. Only the top of the layers are

displayed.

The second campaign with a followed excavation occurred during spring of 2003.

Process differences for the two campaigns can be seen in table I. This campaign



23

lasted for almost two months as well and was thereafter directly followed by

quenching with nitrogen and excavation.

Table I. Differences and similarities for EBFC 10 and 11.

Prior to quenching EBFC 10 EBFC 11

Injectant Oil. ~100kg/thm Coal. ~105kg/thm

Ferrous burden LKAB Pellets LKAB Pellets

Coke SSAB coke spring 2002 SSAB coke spring 2003

CRI & CSR 23.2 & 68.8 respectively 19.4 & 71.6 respectively.

Material probes have been used frequently during the campaign. Feed coke was

sampled every week during the campaign, and has been analysed in the same way as

coke was analysed during the EBFC10. During the excavation, samples were taken

from each coke layer.

Figure 8. EBFC11. Illustration of how a few selected coke layers

were found when the EBF was excavated. The left hand side

displays cross-section from South to North, and the right hand side,



24

from West to East. Only the top of the layers are displayed.

This was done at six different locations for each layer i.e. in the same radial positions

as used earlier but in two different directions. Apart from more extensive sampling in

some areas, the same procedure was used here as for campaign 10. A mapping of the

location of the coke layers of interest for campaign 11 has been made, see Figure 8.

2.2.2 EBF trial with high CRI coke The trial took place in the spring of 2004 and consisted of two parts i.e. a two day

reference period with a low CRI coke followed by two days of operation with high

CRI coke. During this trial solid sampling occurred at three positions, i.e. in the upper

shaft, lower shaft and through the cohesive zone, see Figure 1. In addition, a tuyere

core drilling was done.

The sampled coke material was separated from slag, fluxes and pellets. Prior to x-ray

diffraction, XRF and TGA reactivity measurements, small coke lumps

(approximately 6-8 cm3) were selected from each probe and crushed to powder (< 75

micron).

The tuyere drill core was divided in four equally large segments and labelled Centre,

Mid 1, Mid 2, and Wall. Thereafter the samples were sieved to fractions of -19 mm,

19-22.4 mm and +22.4 mm. The samples labelled Centre thus represent coke from

the centre of the furnace at tuyere level.

The process parameters were altered as little as possible as the coke type was

changed. The same amount of coal injection was used. The cokes types that were

tested are very different in quality, as can be seen in Table II.

Table II: Properties of the feed coke used in the current study

Parameters Low CRI coke High CRI coke CRI 19 48 CSR 72 35 Fe 0.35 1.05



25

SiO2 6.14 4.72 P2O5 0.022 0.053 Al2O3 2.82 2.26 MgO 0.04 0.2 Na2O 0.04 0.11 K2O 0.14 0.22 TiO2 0.18 0.1

2.3 Methods used for characterization of coke samples

To study the cokes, various instruments and methods have been used, i.e.

TGA (Thermal Gravimetric Analysis) and DTA (Differential Thermal

Analysis) with MS (Mass Spectrometry)

CRI (Coke Reactivity Index) and CSR (Coke Strength after Reaction)

Sieving

XRD (X-Ray Diffraction)

SEM (Scanning Electron Microscope) with EDS (Energy Dispersive

Spectroscopy)

LOM (Light Optical Microscope)

BET nitrogen adsorption

Each method is described below.



26

2.3.1 TGA/DTA–MS

Figure 9 shows the

schematic of Netzsch STA

409 instrument at Luleå

University of Technology,

which can be used for

simultaneous Thermal

Gravimetric and

Differential Thermal

Analysis. Non-isothermal

reactivity was measured by

using a small amount of

coke powder (60 ~ 80 mg)

in an Al2O3 crucible in TGA/DTA equipped with a Quadropole mass spectrometer

with the setting to detecting ions with mass of 1 to 65. The loss in sample weight is

recorded by a very accurate balance ±1 μg. All samples of interest have been reacted

under dynamic heating up to 1300 °C with a heating rate of 10K/minute. Various

gases can be used, but in this study 100 % CO2 or 100 % Ar gas was used.

N2

Gas outlet

Furnace

Sample carrier

protective tubevacuumreactive gasprotective gasinductive displacement transducerelectromagneticcompensation systemvacuum tight casing

DSC and TGcarrier

thermostaticcontrol

evacuationsystem

Computer

QMS

radiation shield

Figure 9. Schematic of TGA/DTA furnace used for

non-isothermal reactivity measurement of coke

samples.



27

A custom built TGA, see Figure

10, at the University of New

South Wales was used to

measure the weight loss in coke

samples during isothermal

heating at 900 °C for 2 hours

under 100% CO2 and at various

flow rates ranging from 1.5 to

2.0 l/min. The TGA furnace

consists of a recrystallised

vertical alumina (60 mm ID)

tube. Sample temperature is

controlled by an internal

thermocouple located close to

the sample holder.

Approximately 0.2 g sample was placed on a square alumina crucible (30X 30 mm)

holder at room temperature. Alumina sample assembly is suspended by a high

temperature stainless wire which is connected to a balance that can measure weight

changes of the order of 1 micro gram (Precisa® 1212 M SCS). The assembly was

kept at low temperature zone in the furnace followed by heating up to 900°C at the

rate of 2°C/minute while 5 l/min of N2 was continuously purged through the furnace

which was regulated by Brooks 5850E mass flow controller. As the furnace reaches

the required reaction temperature, the furnace chamber is raised to move the sample

in the reaction zone followed by reducing the N2 flow to 4 l/min and adding 1 l/min

of CO2. The weight loss of coke sample was continuously recorded by data logger

and used to calculate carbon conversion.

2.3.2 CRI/CSR equipment

In the present work a CRI and CSR equipment was constructed and installed at LTU

see Figure 11. It is based upon the ISO draft for CRI and CSR. It is used for

Figure 10. TGA reactor at UNSW used for

isothermal reactivity measurements.



28

determining lump coke reactivity in carbon dioxide gas at elevated temperatures and

its strength after reaction in carbon dioxide gas by tumbling in a cylindrical chamber,

called I-drum.

The coke tested should

consist of pre-dried coke

with sizes from 19.0 mm

to 22.4 mm. This sample

is then heated in a

reaction vessel to

1100°C in a nitrogen

atmosphere. For the test

the atmosphere is

changed to carbon

dioxide for exactly two

hours. After the test, the

reaction vessel is

allowed to cool down to

about 50°C in a nitrogen

atmosphere.

The comparison of the sample weight before and after the reaction determines the

coke reactivity index and is given as a percentage of the weight loss. The reacted

coke is rotated in the I-drum at 600 revolutions for 30 minutes. The CSR value is

determined by sieving and weighing the amount of the coke passing a 10.0 mm sieve.

The abrasion value is defined as the lack of resistance to abrasion of the coke after

reaction with carbon dioxide in the CRI test, measured as the percentage passing

through a 0.5 mm sieve after tumbling in an I-drum. CSRBF consist of the same step

as for the CSR part of the NSC test but with coke excavated from a blast furnace.

Figure 11. Image of the CRI and CSR equipment.

Inserted is a picture of the reactor when in uplifted

position.



29

The furnace consists of three Kanthal Fibrothal 200/200 heating elements and is

controlled in a PC-environment using the software LabVIEW. The gas system

consists of two digital BRONKHORST flow meters. The ISO-draft states that a CO2

flow of 5 dm3/min and a N2 flow of 10 dm3/min in STP must be used.

2.3.3 Sieving

Samples excavated from the EBF have been sieved by hand, using sieves with a mesh

of 22.4 mm, 19.0 mm, 14.0 mm and 10 mm. Samples for CSRBF were also sieved.

2.3.4 X-ray diffraction

Siemens 5000 X-ray diffractometer at the University of New South Wales (UNSW),

Australia was used to record scattering intensities of samples by using Copper K

radiation (30 kV, 30 mA) as the X-ray source. Samples were packed into an

aluminium holder and scanned over an angular range from 5-105° by using a step

size of 0.05° and collecting the scattering intensity for 5 seconds at each step. The

XRD data was processed to obtain crystallite dimension Lc in carbonaceous

materials. The average stacking height of 002 carbon peak can be calculated using

Scherrer’s equation by using K = 0.9 for Lc. A sharper 002 peak will indicate a larger

crystallite size and a greater degree of ordering in the carbon structure [32]. In most

cases Lc was calculated when Xa was determined by half-width criteria. When the

half-width criteria failed, Xa was determined using centre symmetry method.

2.3.5 Chemical analyses

Samples have been sent to laboratory for XRF chemical analysis. The laboratory at

SSAB Tunnplåt Luleå AB and the laboratory at UNSW have done the XRF analyses

while carbon and sulphur content was measured using LECO analyser at the UNSW.



30

2.3.6 Scanning Electron Microscope

Coke pieces were mounted in an epoxy slow-setting resin in plastic moulds (40 mm

diameter). Surfaces were ground on four different grades of silicon carbide paper

(120, 500, 800, and 1200 grit) with distilled water and polished with three different

grades of polishing paper with diamond paste of particle sizes of 15 μm, 9 μm, 3 μm,

and 1 μm. Lubrication fluid was used during the polishing.

Polished samples are fixed on aluminium mounts and coated with a thin layer of

gold-palladium alloy using a Bal-tec MCS 010 sputter coater. The coated specimens

were then examined with a Philips XL 30 scanning electron microscope equipped

with Energy Dispersive X-ray Analysis (EDS) for chemical mapping.

2.3.7 Light Optical Microscope

Coke pieces were mounted in an epoxy slow-setting resin in plastic moulds and

treated the same way as for preparation for SEM.

The coke porosity was measured;

1) Using a Leco 3001 image analysis program. Six polished samples are mounted in a

special holder and placed under the microscope. The microscope measures one coke

piece at the time and the motorized table shifts the samples. The image analysis

software calculates the number of pores seen on the polished surface and also report

pore size and pore size distribution. It measures at magnifications of 520x and 130x

which gives information of macro and micro pores respectively.

2) Using the analySIS 3.2 program and Olympus microscope with 520x

magnification. Coke porosity is calculated as the average value of the samples

measured for each of the four tuyere segments from the tuyere core drillings. Both

procedures were developed by Ruukki in Raahe, Finland.

Light optical microscope has also been used to manually study samples.



31

2.3.8 BET

The BET surface area was measured using a FlowSorb 2300 by determining the

quantity of N2 that adsorbs as a single layer of molecules, a so-called monomolecular

layer, on a sample. This adsorption is done at or near the boiling point of the

adsorbate gas. Under specific conditions, the area covered by each gas molecule is

known within relatively narrow limits. The area of the sample is thus directly

calculable from the number of adsorbed molecules, which is derived from the gas

quantity at the prescribed conditions, and the area occupied by each.

2.3.9 Micro texture measurement

The change in the coke microstructure passing the EBF was measured by an

automated microscopic measuring procedure developed at TKS to quantify the

ordering of the coke carbon microstructure. This measuring procedure is based on the

optical physics of the bi-reflectance. The dimension of the bi-reflectance is recorded

using a linear polarising filter in the reflected light of the sample at various

polarisation degrees.

The microscope employed is equipped with a scanning stage, an auto focus system

and a power-driven polarizer in the reflected microscopic light. An adapted image

analysing system enables quantification of the degree of anisotropic and isotropic

components calculated from the optical bi-reflectance.

Using this method at both coke operations (high and low level CRI) the feed cokes

were investigated in comparison to the tuyere coke material sampled by the tuyere

probe. The material of each tuyere core drilling was split radially into four segments.

The material of each segment was screened into three fractions (<19 mm, 19-22.4

mm and > 22.4 mm) and than separated into coke, metal and slag components. From



32

the crushed feed coke and each tuyere coke material a polished section was prepared

for determination by this microscopic measurement technique.

3. RESULTS AND DISCUSSION

Coke degradation and gasification is influenced by coke porosity, carbon structure

and its minerals. The variation in physical properties, the evolution of carbon

structure and the variation in chemical properties are discussed here.

3.1 Variation in physical Properties

Cokes from the

EBF were tested in

the I-drum to

determine a “Coke

Strength after BF

reactions” value.

Figure 12 illustrates

the result from the

CSR part of the ISO

draft for CRI &

CSR. The Y-axis

corresponds to the

“Coke Strength

after BF reactions”

values and the X-

axis displays the distance below the top of the furnace. As can be seen, a negative

trend line can be easily fitted. The sodium content in the abraded material of the coke

is also plotted as a function of distance.

Figure 13 illustrates the result from the abrasion index part of the ISO draft for CRI

& CSR. The Y-axis corresponds to the “Abrasion Index” values and the X-axis

R2 = 0,93

R2 = 0,91

80,0

81,0

82,0

83,0

84,0

85,0

86,0

87,0

88,0

3,5 4 4,5 5 5,5 6 6,5 7 7,5

Distance (m)

CSR

BF

0

0,2

0,4

0,6

0,8

1

1,2

% N

a 2O

in re

sidu

e fro

m a

bras

ionCSR BF, EBFC10

Na2O %

Figure 12. Coke treated according to the CSR part of the

CRI & CSR ISO-draft. Left Y-axis represents the mass of

coke larger or equal to 10.0 mm after treatment and right

Y-axis is sodium content in the coke.



33

displays the distance below the top of the furnace. As can be seen, a polynomial trend

line can be fitted. The potassium content in the abraded material of the coke is also

plotted as a function of distance.

The increase of

alkali in the

abraded material

corresponds well

with the chemical

analyses of the bulk

coke from

corresponding

layers. There is no

significant

difference, thus

indicating that

alkali not only

increases on the

surface of the coke

but that it actually penetrates the whole coke matrix.

Porosity measurements of the coke samples used in this study indicated no significant

variation in the porosity of coke samples from different locations in the EBF.

Examination of EBF coke samples under light optical microscopy indicate that open

pores could have increased marginally as coke descends towards the tuyeres. BET N2

surface area of the EBF coke samples suggest that surface area of cokes did not

change significantly in samples and hence might not have a significant influence on

possible differences of reactivity measurements. It may be noted that there could be

differences in the surface area of coke layers not included in this study. Further

discussion is mainly limited to changes occurring in carbon structure and coke

minerals.

R2 = 0,93

R2 = 0,84

0,0

1,0

2,0

3,0

4,0

5,0

6,0

3,5 4 4,5 5 5,5 6 6,5 7 7,5

Distance (m)

Abr

asio

n In

dex

0

0,5

1

1,5

2

2,5

3

3,5

% K

2O in

resi

due

from

abr

asio

n

Abrasion IndexK2O %

Figure 12. Coke treated according to the CSR part of the

CRI & CSR ISO-draft. Left Y-axis represents the mass of

coke larger or equal to 10.0 mm after treatment and right

Y-axis is sodium content in the coke.



34

The measurements of the BET surface measure open and accessible micro pores. The

coke was grinded in a ring mill for 15 seconds to produce a powder. This was done

for coke from both campaigns 10 and 11. In Figure 14 the BET surface is seen for the

specific layers.

(a) (b)Figure 14. To the left, BET measurements of cokes from EBFC 10 and to

the right from EBFC 11. The layers in campaign 10 and 11 do not have the

same positions.

The excavated coke was also hand sieved using sieves with mesh 22.4, 19.0, 14.0 and

10.0 mm and for each of the three horizontal positions. The peripheral coke becomes

smaller in size (the fraction equal to or above 22.4 mm is reduced) as it travels down

the furnace, in accordance to what one would expect. However, centre coke increases

in size, according to sieving results, as it travels down the furnace. This result is

contrary to what would be expected. This phenomenon can be due to small coke

being predominantly consumed in the centre of the furnace. However, variations were

small and the sampling is difficult for this purpose and could hence result in errors.



35

3.2 Evolution of Carbon structure

As the coke descends through the blast furnace, it reacts with upcoming CO2 gases

and loses carbon content. Figure

15 shows that carbon content of

coke samples is decreasing such

that around the bosh region in the

furnace (sample 35) coke

contained approximately 3% less

carbon content. Increased ash

content can be attributed to carbon

loss as well as increased alkali

uptake by coke in the EBF.

Figure 16 shows XRD patterns of

cokes samples from three different

locations. The chemical structure

of coke carbon is increased as

indicated by a sharpening of the

002 carbon peak in cokes taken

from locations 5 to 35. Further,

Figure 16 indicates less

background intensity in the XRD

patterns of coke samples from

lower levels in the EBF (location

35 is less than location 5). Lower

background intensity is often

indicative of decreasing

amorphous carbon content in

coke. Even though amorphous

Figure 15. Variation in carbon content of

EBF centreline coke samples plotted against

distance from top of EBF, tentative

associated temperatures in EBF are also

indicated. From campaign 10.

Figure 16. Variation in background intensity

of XRD patterns of coke samples from three

locations.



36

carbon content is not distinctively different in the coke samples shown in Figure 16,

the amorphous carbon content is believed to decrease as hearth coke samples

indicated significantly lower amorphous carbon content than coke samples from

upper parts of the EBF. The results suggest that amorphous carbon is increasingly

depleted as coke descends towards the hearth.

Carbon atoms become more ordered as coke passes from shaft to bosh region as

indicated by the increasingly higher Lc values as shown in Figure 17. This means that

coke structure becomes more ordered during its movement towards lower parts of

furnace. Samples from the hearth were strongly graphitised. Generally, highly

ordered carbons are expected to be

more reactive towards oxidising

gases, including CO2.

The linear correlation suggests

that increase in Lc value is

strongly influenced by

temperature in the EBF even

though other factors such as alkali

and iron species present in coke

could also influence the chemical

structure. The Lc values were

calculated from x-ray diffraction

spectrum after applying

corrections to raw XRD data.

15

30

45

60

75

90

105

120

4 5 6 7 8 9 10

Distance from top of furnace (meters)

Lc v

alue

s of

cok

e (A

ngst

rom

)

600

800

1000

1200

1400

1600

1800

2000

Temperature of coke bed layer ( oC

)

Central layer temperature

Lc values of central layercokes

Figure 17. Increase in Lc values of coke

during its journey towards cohesive zone in

the EBF and associated temperatures based

on assumptions generated from vertical,

horizontal and inclined temperature probe

measurements. Campaign 10.



37

3.3 Alkali uptake and dispersion in coke

The XRF analyses of coke

samples from the centreline

position in the furnace are

indicated in Table III and

IV. Figure 18 plots the

alkali content in coke ash

against the furnace depth

and shows that alkali

content (K2O and Na2O) in

coke increases as the coke

moves through the shaft to

the cohesive zone. It is

obvious that the alkali

present in recirculation

gases inside the blast

furnace have condensed on

coke surface or penetrated inside the coke matrix followed by reactions with other

minerals. In order to understand the alkali distribution, each coke sample was

analysed for three regions namely outer, middle and core region of sample.

Table III. Chemical composition of EBFC10 coke samples.

XRF(SSAB) Sum ox. SiO2 Al2O3 Fe2O3 CaO MgO K2O Na2O TiO2 P2O5 SO3

Coke samples from EBF Campaign 10 KL01C 11.4 6.50 3.15 0.58 0.15 0.10 0.37 0.15 0.17 0.04 0.18 KL05C 11.59 5.76 2.63 1.20 0.01 0.06 0.17 0.10 0.16 0.03 1.47 KL10C 12.53 6.39 2.77 1.32 0.04 0.06 0.35 0.16 0.17 0.03 1.25 KL15C 13.26 6.30 2.65 1.06 0.00 0.06 1.24 0.43 0.15 0.02 1.35 KL20C 12.98 5.58 2.57 0.92 0.02 0.07 1.78 0.61 0.14 0.02 1.27 KL25C 13.77 5.83 2.61 0.92 0.02 0.08 2.31 0.67 0.14 0.02 1.17 KL30C 13.89 5.97 2.66 0.97 0.04 0.08 2.07 0.68 0.14 0.02 1.25 KL35C 14.80 5.81 2.64 0.77 0.00 0.08 3.21 0.85 0.12 0.22 1.10

Figure 18. Alkali concentration in EBF coke ash

plotted against distance from top of furnace.

Approximate temperature profile of EBF is also

indicated.



38

Table IV. Chemical composition of EBFC11 coke samples.

XRF(SSAB) Sum ox. SiO2 Al2O3 Fe2O3 CaO MgO K2O Na2O TiO2 P2O5 SO3

Coke samples from EBF Campaign 11 KL01C 10.10 6.13 2.63 1.12 0 0.06 0.18 0.05 0.17 0.026 1.45 KL05C 16.59 6.27 4.07 5.63 0.39 0.11 2.35 0.22 0.11 0.025 1.10 KL10NC 10.28 6.23 2.74 1.09 0 0.04 0.16 0.05 0.16 0.032 1.52 KL15NC 10.59 6.31 2.74 1.32 0 0.05 0.26 0.06 0.17 0.03 1.50 KL20NC 11.44 6.66 2.76 1.54 0.14 0.11 0.43 0.07 0.17 0.029 1.45 KL25NC 10.04 5.61 2.44 2.06 0.07 0.05 0.29 0.07 0.16 0.029 1.42 KL30NC 13.93 6.21 2.74 0.83 0 0.06 3.59 0.42 0.12 0.026 1.20 KL35C 18.98 4.63 1.86 17.84 1.94 0.19 1.01 0.19 0.08 0.025 0.65 KL40C 12.44 6.03 2.53 0.97 0 0.05 2.58 0.34 0.13 0.023 1.27

Figure 19-24 provides the

SEM analysis of coke

samples from two widely

different locations in the

EBF (from coke layer 10

and 35 from campaign 10)

at various magnifications.

Figure 19 & 20 illustrates

the inhomogeneity of

mineral distribution in

coke sample 10C. The

EDS analysis (see Table V

to IX) suggested the alkali

content of the

aluminosilicate phases of

coke sample 35C (35th

layer) was higher than in

sample 10C. Alkali also

appear to be incresaingly associated with the carbon matrix when comparing sample

35C to sample 10C. EDS analysis of mineral grains in Figures 19 and 20 indicated

that alkali contents of minerals was in the normal range of often observed

a) Outer coke matrix 10C

b) Central coke matrix 10C

Figure 19. a) SEM images illustrating mineral

distribution in upper/outer coke layer in sample

(10C) from upper part of the EBF, b) central core

region of the same coke at various magnifications.



39

aluminosilicate phases throughout the coke. The middle sample position can be seen

in Figure 20.

Table V. EDS analysis of coke layer 10. See Figure 19.Values reported in Wt %.

Image Point Na K C Al Si O a 1 6.1 1.8 21.7 12.2 25.9 24.9 a 2 17.5 0.8 30.2 6.2 11.4 15.1 a 3 0.3 0.4 39.7 0.9 29.5 26.6 a 4 0.7 0.3 19.6 3.8 40.3 33.0 a 5 3.7 2.3 33.8 14.3 18.0 21.3 a 6 1.9 0.8 26.3 5.9 33.8 27.9 a 7 1.1 0.3 83.4 1.5 3.0 2.3 a 8 0.4 0.2 37.2 0.8 30.2 27.8 a 9 1.5 0.6 16.5 6.9 38.2 32.7 b 1 0.2 0.2 32.8 0.9 65.9 -b 2 0.1 0.3 41.5 0.8 57.3 -b 3 0.4 0.1 91.5 0.5 7.5 -

a) b)

c) d)



40

Figure 20. Middle coke matrix 10C

Table VI. EDS analysis of coke layer 10. See Figure 20.Values reported in Wt %.

Image Point Na K C Al Si O b 1 1.1 0.6 97.7 0.3 0.3 -b 2 1.5 0.5 97.2 0.4 0.5 -b 3 1.2 1.5 24.4 14.8 58.1 -b 4 1.8 0.3 94.8 1.5 1.6 -b 5 1.5 0.3 90.5 2.8 5.0 -b 6 1.4 0.4 97.6 0.4 0.3 -c 1 1.3 0.8 23.4 3.8 70.7 -c 2 1.3 3.5 13.1 34.6 47.6 -c 3 1.1 0.5 32.8 1.9 63.7 -c 4 1.0 1.2 18.6 2.8 76.4 -c 5 1.4 2.1 55.4 15.6 25.5 -c 6 1.2 1.4 63.9 11.6 22.0 -c 7 0.9 0.5 96.6 0.9 1.2 -c 8 1.5 1.9 44.3 20.5 31.8 -c 9 0.6 0.4 98.0 0.2 0.9 -d 1 2.4 0.9 - 7.5 11.2 8.5 d 2 3.8 0.7 - 14.7 19.7 31.1 d 3 1.0 0.9 - 5.8 8.1 13.4 d 4 1.1 0.4 - 4.3 6.3 6.9 d 5 1.3 1.8 - 22.6 23.5 29.3 d 6 1.2 2.3 - 29.3 31.4 32.2 d 7 1.0 2.2 - 26.3 32.6 32.8

EDS has been used to create mappings of coke samples as well as for point chemical

analysis. Alkali is found together with Aluminium, Silica and Oxygen, e.g. see Figure

21. The probable compounds are the more common (K,Na)AlSi2O4 and the less

common (K,Na)AlSi2O6. Alkali is found all over the coke matrix. The average alkali

content in coke from layers just above the cohesive zone reached levels to above 4

wt.%.

Table VII. EDS analysis of coke layer 35. See Figure 22. Values reported in Wt %

Point Na K C O Al Si1 2.9 10.1 2.6 31.4 24.7 25.12 0.1 0.1 3.3 39.6 0.6 55.23 3.5 10.8 2.5 30.1 24.8 25.04 1.0 5.6 86.8 3.0 0.5 0.45 3.1 10.4 2.1 30.0 24.5 26.66 0.9 5.8 84.8 3.3 0.6 0.6



41

Figure 21. EDS mapping of coke from the periphery of layer 35, EBFC10.

Figure 22-24 compare the physical appearance of three regions within the coke

matrix of sample 35C from a lower part of the EBF. In general, the alkali content of

the aluminisilicates analysed in sample 35C were found to be higher when compared

to alkali content of similar phases from the coke sample 10C, see Table V to IX. No

apparent cracks or significant changes in macro pores were visible in coke sample

35C. Visual examination of SEM images of coke 10C and 35C did not indicate any

significant changes in their physical structure. Alkali could influence the surface area,

chemical structure and could also display catalytic effect, see paper IV.



42

Figure 22. Periphery of sample from coke layer 35.

a)

b)

d)

c)



43

In Figure 23 d), the upper half of the coke matrix is shown. This coke is from the

Figure 23. SEM images of centre of sample 35.

a)

b)

d)

c)



44

periphery of layer 35. Images 23 a), b) and c) display an area from the peripheral part

of the coke. Displayed in table VIII are a few points and their estimated content of

some elements. In Figure 24 d), the lower half of the coke matrix is shown. Images

24 a), b) and c) display an area from the central part of the coke. Displayed in table

IX are a few points and their estimated content of some elements.

Table VIII. EDS analysis of coke layer 35. See Figure 23. Values reported in Wt %

Point Na K C O Al Si1 4.3 8.5 4.5 29.9 22.8 26.72 3.6 6.1 9.7 29.0 23.2 24.63 1.2 2.8 85.7 2.1 1.3 1.74 1.0 2.6 86.8 3.1 0.2 0.55 0.9 2.4 89.1 0 0.8 0.96 3.9 5.2 4.9 29.9 30.7 21.37 1.5 2.1 87.4 1.7 0.3 0.68 3.6 6.3 4.9 30.0 28.2 23.99 3.7 8.0 5.2 29.8 23.9 25.6

Middle coke matrix C35

Figure 24. SEM images illustrating mineral distribution in middle part of coke

layer in sample (35C) from from lower part of EBF at various magnifications.

Table IX. EDS analysis of coke layer 35. See Figure 24. Values reported in Wt %

Point Na K C O Al Si1 3.7 8.6 12.4 27.7 20.3 25.12 3.6 8.8 10.5 29.5 22.1 23.83 1.3 14.9 45.6 16.8 5.0 4.74 3.4 9.7 24.2 20.1 17.4 20.55 2.9 9.5 27.5 25.5 14.7 16.26 3.5 8.4 10.2 28.5 22.6 24.9



45

3.5 Evolution of Coke reaction with CO2

Figure 25

compares the non-

isothermal

reactivity of coke

samples from

different locations

in the EBF. This

figure

demonstrates that

as coke descends

in the blast

furnace, its

reactivity

increases. A similar trend was also observed during isothermal reactivity

measurements as shown in Figure 26.

Comparison of isothermal reactivity

of coke samples from upper (5C) and

lower zones (35C) of EBF suggests

that CO2 reaction of cokes from

lower parts of the EBF was faster, see

paper I, II and III. Porosity did not

significantly change for these

samples; therefore the increase in

coke reactivity is most likely related

to the presence of enhanced alkali

concentration. This means that in the

EBF, the coke reactivity can increase

due to catalytic influence of alkali

Figure 25. Non-isothermal reactivity illustrated as loss in wt.

of EBFC10 coke samples with increasing temperature in a

TGA/DTA furnace.

Figure 26. Isothermal reactivity of

EBFC10 coke samples illustrated as loss in

weight of two coke samples (5C & 35C) at

900°C in a TGA furnace.



46

even when the carbon structure becomes more ordered. Relatively strong influence of

coke minerals on coke reactivity was also observed in a recent laboratory study of

reactivity based on Australian cokes [33].

Non-isothermal TGA/DTA analyses were made on cokes from EBFC11 as well, see

Figure 27. The results are similar to the results from tests on coke samples from

EBFC10.

In order to further study the differences, derivate TGA plots were generated for both

EBFC10 and EBFC11 cokes. As can be seen in Figure 28 a), samples from 15C and

20C experience a more severe weight loss rate than samples from 01C, 05C, and 10C.

It is even greater for samples from 25C, 30C, and 35C. In Figure 28 b), sample 30C,

and definitely sample 40C, experience a greater weight loss rate. The gasification

threshold temperature was lowered from approximately 1000°C down to 800°C when

comparing top and bottom layers.

This is the same

for cokes

excavated from

campaigns 10

and 11. The

increase of

alkali content

from top layer to

bottom layer is

about 10-fold.

This alkali

uptake

phenomenon

corresponds well with some previous studies.

Figure 27. Non-isothermal reactivity illustrated as loss in wt.

of EBFC11 coke samples with increasing temperature in a

TGA/DTA furnace.



47

(a) (b)

Figure 28 a). Derivative of TGA results for selected cokes from EBFC10 and b)

Derivative of TGA results for selected cokes from EBFC11.

Figures 29 a) and b)

give the DTA curves

of coke layers taken

from EBFC10 and

EBFC11 respectively.

To understand

different reaction

transitions of these

coke layers, the

derivative of the DTA

curves are calculated

and shown in Figures

30 a) and b). Both types of figures show the exothermicity or the endothermicity of

reactions involved. Two categories of coke layers appear in the DTA curves.

The first category includes the coke layers KL30C, KL35C and KL43C containing a

high amount of alkaline (K, Na), see Tables III & IV, which are distributed in

Alumina silicate matrices as observed by SEM. The second category of coke layers

includes KL01C and KL05C that contain a low amount of alkaline. DTA results from

900 950 1000 1050 1100 1150 1200 1250 1300Temperature /°C

-1.4-1.2-1.0-0.8-0.6-0.4-0.2

00.2

DTA /(uV/mg)

30C

35C

05C

01C

1066.6 °C

1218.7 °C1059.6 °C

1240.1 °C exo

Figure 29 a). DTA results for selected cokes from

EBFC10.



48

the first category coke layers from both campaigns show endothermic peaks at

approximately 1050°C

and exothermic peaks

occurring between

approximately 1200 and

1300°C. This behaviour

can be attributed to

possible alkali uptake, as

samples with a higher

alkali content (KL30C,

KL35C and KL43C)

show separate reactions

occurring at high

temperature that do not occur in samples with lower alkali content (KL01C and

KL05C).

DDTA results

demonstrate even more

clearly that additional

reactions are occurring

in coke layer samples

from further down in the

EBF. Coke samples

from EBFC10 show

greater thermal effects

above 1000°C than

samples from EBF11C.

This can also be due to

average alkali content being greater in samples from EBF10C (14.5wt.% in coke ash)

than from EBF11C (10.8wt.% in coke ash). That alkali content can catalyse coke

800 900 1000 1100 1200 1300Temperature /°C

-1.5

-1.0

-0.5

0

DTA /(uV/mg)

01C

05C30C

43C

1033.6 °C

1050.1 °C

exo

Figure 29 b) DTA results for selected cokes from

EBFC11.

800 900 1000 1100 1200 1300Temperature /°C

-0.20

-0.10

0

0.10

0.20

0.30

DDTA /(uV/mg/min)

35C

30C

01C

05C

Figure 30 a). Derivative of DTA results for selected

cokes from EBFC10.



49

reactivity is also supported by Figure 28 a) and b) were coke layers with increasing

alkali content show increased reaction at lower temperatures. This is true regarding

the reactivity of coke

layers in each campaign

as well as between the

overall reactivity of coke

in the separate

campaigns 10 and 11.

Analysis of Figure 31 supports the idea that alkaline substances contained in coke

layers act as catalyst. Figure 31 shows the evolution of carbon monoxide from the

reaction of carbon

with CO2 as a

function

of time for the

different coke layer

samples in EBFC10.

From this Figure, it

can be seen that the

coke layers KL35C,

KL25C and KL30C,

which contain more

alkali generate CO

earlier than upper coke layers such as KL05C containing lower alkaline.

800 900 1000 1100 1200 1300Temperature /°C

-0.20

-0.10

0

0.10

0.20

0.30

DDTA /(uV/mg/min)

01C

05C30C43C

Figure 30 b). Derivative of DTA results for selected

cokes from EBFC11.

Figure 31. CO gas evolution of selected cokes from EBFC10.



50

3.6 Isotropic / anisotropic changes in the coke carbon micro structure Coke from the EBF campaign number 10 was analysed by TKS micro texture

measurement apparatus. The results of anisotropic and isotropic changes can be

seen in Figure 32. The relative amount of isotropic components increases as the

coke goes through

the EBF. However,

it is a known fact

that isotropic

components are

more reactive with

CO2, thus the

results are in

contradiction to

what was expected

when considering

previous studies.

This could confirm that other reactions occur and suggest that alkali catalyses

reactions with anisotropic components in the coke.

3.7 Trial with high CRI coke In the EBF, coke undergoes many modifications in carbon structure and constituent

minerals including alkali phases and porosity as discussed below. Each of these

properties could influence the coke behaviour in an operating blast furnace,

particularly its strength. The trial with 100% high CRI / low CSR coke lasted for two

days and was very unique. With such poor quality coke, regarding strength, it would

have been very difficult, if not impossible, to try use this in a full scale blast furnace.

But it was possible to test this coke in the EBF and the interesting aspect was the high

CRI value.

69

71

73

75

77

79

4,00 4,50 5,00 5,50 6,00 6,50 7,00 7,50Distance from top of EBF (m)

Ani

sotr

opic

tot i

n vo

l. %

19

21

23

25

27

29

Isot

ropi

c to

t in

vol.

%

AnisotropicIsotropic

Figure 32. Anisotropic and isotropic coke carbon components.



51

3.7.1 Process analysis

The most important results are summarised as follows:

- The coke rate was increased by more than 30 kg/thm.

- Even with a much higher reductant rate it was not possible to keep an

acceptable hot metal heat level. Silicon content and hot metal temperature

was significantly lower and sulphur content much higher. Surprisingly, the

carbon content was not so heavily influenced.

- The gas utilisation was significantly increased, although with very high

fluctuations, as shown in Figure 33. The same phenomena occur in the

EBF when operating with a ferrous burden with a high degree of swelling

and disintegration.

- Greater slag volume and lower basicity, because of lower silicon content

in the hot metal.

- The temperature distribution in the EBF was changed in such a way that

the temperature in the lower shaft became much higher. The cohesive

zone moved upwards which was also verified from shaft pressure

measurements.

- The amount of flue dust was almost tripled, but the carbon content only

increased by 30 %, indicating a more irregular gas flow, also causing

more iron units to leave the furnace with the gas.

Figure 33 Illustrate gas utilization value, EtaCO, CO/(CO2+CO). The arrow indicates the shift from low CRI to high CRI coke.



52

3.7.2 Evolution of carbon structure

Figures 34 and 35 compare the XRD patterns of cokes from four representative

locations of the EBF. It is seen that the width of the 002 carbon peak becomes sharper

as the coke descends. The background intensity of lower zone coke samples (e.g.

tuyere level) are less than those of coke samples from the upper part of the EBF

(upper shaft samples). Low background intensity is indicative of less proportion of

amorphous carbon. The amorphous carbon of coke was found to decline sluggishly

for both cokes, but more so for the low CRI coke, up to the cohesive zone (Figures 34

a) and b) and 35 a)) and then changed rapidly as the coke descended in the hearth

regions (Figure 35 b)) of the EBF. This means that coke carbon becomes increasingly

ordered as the coke passes from shaft to bosh region while amorphous carbon is

increasingly depleted.

The comparison between the high and low CRI coke reveals that the temperature

profiles have been different during the reference and trial periods. It verifies the fact

that the cohesive zone was shifted upwards during the trial period with high CRI coke

resulting in a higher thermal state in the furnace increasing coke reactivity higher up

in the furnace shaft. As a consequence of this, the thermal state in the furnace hearth

during the trial with high CRI coke was lower than the reference period with low CRI

coke.

Upper probe

10 20 30 40Diffraction angle, 2 Theta

Low CRI cokeHigh CRI coke

Lower probe



(a) (b) Figure 34 a) Comparison of x-ray diffraction patterns between low and high CRI coke from the upper probe b) coke from lower probe



53

Inclined probe



Tuyere samples, centre



(a) (b) Figure 35 a) Comparison of x-ray diffraction patterns between low and high CRI coke from the inclined probe b) coke from the tuyere core drilling, from the centre of the furnace

The Lc value increases as the coke

descends which can be seen in Figure

36. The lower Lc value generated from

the trial coke, compared to the Lc

value for the reference coke, from the

tuyere core centre can be explained by

the lower hot metal temperature i.e.

lower heat level in the hearth during

the trial period.

3.7.3 Evolution of coke ash chemistry

In addition to carbon structure, coke ash chemistry is also continuously changing in

the EBF for both of the different sample coke qualities. The alkali uptake is higher

for the high CRI coke as can be seen in Figure 37 a). It is clear that the increase of

potassium and sodium is proportional. However, the amount of potassium is much

higher for all samples. Figure 37 b) reveals the alkali distribution in the tuyere core

probe.

Lc

0

10

20

30

40

50

60

70

80

Upper Lower Inclined Tuyerewall

Tuyerecentre

Ång

strö

m

Lc Ref cokeLc Trial coke

Figure 36 Calculated Lc values from x-ray diffraction spectrum after processing of raw XRD data.



54

Alkali as function of probes

0

0,5

1

1,5

2

2,5

3

3,5

Feed Upper Lower Inclined TuyereW

TuyereC

% P

otas

sium

oxi

de

0

0,1

0,2

0,3

0,4

0,5

0,6

0,7

% S

odiu

m o

xide

Low CRI K2OHigh CRI K2OLow CRI Na2OHigh CRI Na2O

Tuyere core samples

0,0

0,5

1,0

1,5

2,0

2,5

3,0

Na2O Na2O K2O K2O

Low CRIcoke

High CRIcoke

Low CRIcoke

High CRIcoke

% in

cok

e

wallmid 2mid 1centre

(a) (b) Figure 37 a) Alkali in coke as a function of samples b) Alkali distribution in

tuyere core.

It can be seen that the highest amount of alkali is found at the wall of the EBF.

Furthermore, the low CRI coke contained very little sodium and potassium when

compared with the high CRI coke. The reason to this could also point to the different

temperature profiles in the hearth between the reference and trial periods which is

supported by XRD and Lc value results.

Alkalis are recognized as a low temperature indicator when connected with silicates.

Considering the high CRI coke, the increased alkali seems to be interconnected with

the carbon textures. As isotropic content in carbon phases of coke increases, the

potassium uptake increases and possibly enhances the abrasion of coke. An influence

of alkalis on carbon disappearance is not clear in any way. Even gasified areas inside

fused isotropic textures are detected. At the tuyere level there seems to be a more

dominant relationship between silica vaporization from silicates and increased coke

porosity. However, there seems to be a positive relationship between increased

gasification spots of isotropic texture and coke porosity. A small decrease of the high

CRI coke porosity with an increase in alkali content can be seen.



55

3.7.4 Evolution of coke reaction with CO2

A TGA comparison between feed, upper, lower and inclined probes, as well as a

tuyere core probe for both high and low CRI coke was done. Figure 38 to 40

compares the non-isothermal reactivity of the EBF coke. The weight losses of lower

zone coke samples are consistently higher compared to upper zone coke samples.

This implies that the reactivity of coke increases as it descends through the EBF. The

coke reactivity is often related to the carbon structure, surface area and coke minerals.

In the previous section it was seen that carbon structure of lower zone EBF cokes was

increasingly ordered. Therefore, the carbon structure alone could not be responsible

for increased reactivity of lower zone coke samples as increased crystalline order of

carbon is believed to retard the reactivity.

Figure 38 a) compares the feed of the high and low CRI cokes. The solution loss

reaction starts earlier for the high CRI coke and at 1300°C the weight loss is 53 % for

the trial coke and 46 % for the reference coke. However, the trial coke had

approximately 5 % higher moisture content. In Figure 38 b) the upper probe samples

of the reference and trial coke are compared. It is interesting to notice that the low

CRI coke actually reacts faster than the high CRI coke. The mass change is 64 % and

37 % respectively. The same pattern is seen in Figure 39 a) and b). At tuyere level the

high CRI coke does react faster and has a greater mass loss, see Figure 40a) and b).

However, the difference between the two different coke samples is small even though

one would expect a larger difference.



56

400 500 600 700 800 900 1000 1100 1200 1300Temperature /°C

50

60

70

80

90

100

TG /%

High CRI coke

Low CRI coke

Mass Change: -52.62 %

-46.00 %

400 500 600 700 800 900 1000 1100 1200 1300Temperature /°C

40

50

60

70

80

90

100

TG /%

High CRI coke

Low CRI coke

-36.76 %


(a) (b)

Figure 38 a) Reference and trial coke from feed b) Reference and trial coke from upper probe

400 500 600 700 800 900 1000 1100 1200 1300Temperature /°C

40

50

60

70

80

90

100

TG /%

High CRI coke

Low CRI coke


-49.92 %

400 500 600 700 800 900 1000 1100 1200 1300Temperature /°C

20

30

40

50

60

70

80

90

100

TG /%

High CRI coke

Low CRI coke


-81.74 %

(a) (b)

Figure 39 a) Reference and trial coke from lower probe b) Reference and trial coke from inclined probe

400 500 600 700 800 900 1000 1100 1200 1300Temperature /°C

20.0

30.0

40.0

50.0

60.0

70.0

80.0

90.0

100.0

TG /%

High CRI coke

Low CRI coke


-76.24 %

400 500 600 700 800 900 1000 1100 1200 1300Temperature /°C

20

30

40

50

60

70

80

90

100

TG /%

High CRI coke

Low CRI coke


-72.70 %

(a) centre (b) wall

Figure 40 a) Reference and trial coke from centre of the EBF at tuyere level b) Reference and trial coke from wall side of the EBF at tuyere level



57

The TGA study implies that the CRI value has little influence when the cokes have

been crushed, thus suggesting that both structure and specific surface area for

reaction play an important role.

3.7.5 Isotropic / anisotropic changes in the coke carbon micro structure

Figures 41 a) and b) show results of the mean values of Anisotropictotal and

Isotropictotal components of the high and low CRI coke samples.

The measurements for feed cokes result in very different levels of detected

anisotropy. The high quality reference coke contains 79 vol.% anisotropes

corresponding to 21 vol.% isotropes, the low quality high CRI coke contains 62

vol.% anisotropes and 38 vol.% isotropes. This result reflects a very different origin

of the coals (coal blends) used for production of the different coke types.

With respect to the coke texture of the tuyere coke for both coke types the isotropic

fraction increases towards the centre of the EBF. In the case of the high CRI coke, an

increase of 10% isotropic content is corresponding to a 10% decrease of the

anisotropic content towards the centre. The changes in low CRI coke are less, 5%

isotropic content increase corresponding with a 5% decrease in anisotropic content.

45

50

55

60

65

70

75

80

85

centre mid 1 mid 2 wall

Ani

sotr

opic

tota

l in

vol.%

Tuyere coke (High CRI)

Tuyere coke (Reference)

High CRI feed coke

Low CRI feed coke (Reference)

15

20

25

30

35

40

45

50

55


Isot

ropi

cto

tal i

n vo

l. %



High CRI feed coke


(a) (b) Figure 41 a) Anisotropic coke carbon components b) Isotropic coke carbon components

With respect to the behaviour of these cokes while descending in the EBF, it can be

concluded that the isotropic carbon components are more resistant to metallurgical



58

treatment under load compared to the anisotropic components. Apparently the

gasification attacks the anisotropic carbon components more than the isotropic

components.

3.7.6 Porosity differences

Variation in porous structure was

measured for two sets of tuyere

cokes, low and high CRI, see

Figures 42 and 43. Coke porosity

is calculated as the average value

of the sample porosity measured

for the tuyere segment.

Image analysis shows a greater

porosity for low CRI coke which

progressively increases towards

the EBF centre, see Figure 44. A

similar increasing porosity is

true for the high CRI coke but

with the difference of an

opposite downward trend

between segment 2 and the EBF

centre.

Low CRI coke porosity versus pore area

1000

3000

5000

40 50 60 70 80

Porosity (%)

Por

e (u

m2) Wall

0.25-0.5 m0.5-75 mCentre

Figure 42 Pore area vs. porosity for the reference coke

High CRI coke porosity versus pore area

1000

3000

5000

40 50 60 70 80

Porosity (%)

Por

e (u

m2) Wall

0.25-0.5 m0.5-75 mCentre

Figure 43 Pore area vs. porosity for the trial coke

Tuyere coke porosity versus pore area

40

50

60

70

Wall Seg. 3 Seg. 2 Centre

Distance from tuyere

Poro

sity

(%)

2000

2500

3000

3500

4000

Cok

e si

ze (u

m2)

Low CRI/PorosityHigh CRI/PorosityLow CRI/PoreHigh CRI/Pore

Figure 44 The tuyere coke porosity vs. pore area for both of the coke types



59

4. CONCLUSIONS

4.1 Evolution of coke carbon structure

In the experimental blast furnace, coke properties continuously change as the coke

moves down from the shaft towards cohesive zone such that carbon structure

becomes increasingly ordered and a linear relationship between stack height of

carbon crystallite in coke and EBF temperature was established. The chemical

structure of carbon crystallites of cokes was found to increase while amorphous

carbon was found to decrease during downward movement such that cokes excavated

from the hearth contained significantly lower amorphous carbon compared to coke

samples from other parts of the EBF.

4.2 Alkali implications

The total alkali concentration of coke was found to increase during its descent

towards lower parts of blast furnace from less than 1% to more than 4%. However,

coke excavated from the hearth was nearly depleted of all the accumulated alkali,

which can confirm alkali circulation occurring above the hearth. EDS mapping

confirmed that alkali accumulate throughout the whole coke matrix. It also suggested

that alkali reacts with aluminium-silica-oxides to form stable compounds.

4.3 Coke Reactivity

Both isothermal and non-isothermal reactivity based on TGA measurements clearly

established that coke reactivity with CO2 is improved as the coke moves from shaft to

cohesive zone and downward in the EBF despite increasing order in their carbon

structures. The catalytic effect of alkalis in the EBF samples appears to have a strong

effect on gasification reactivity. The study suggests that the potential adverse effect

of coke graphitisation due to thermal annealing could be compensated by catalytic

influence of alkali components of coke.



60

4.4 Physical Properties including Coke Strength and Abrasion

LOM porosity measurements indicated that the number of pores decrease, both at the

magnitude of 130x and 520x. One could expect the opposite, because carbon is being

consumed as coke travels through the EBF. However, depositions of alkali and iron at

lower parts of the furnace could fill the open and easily accessible pores thereby

retarding the carbon gasification.

The fraction of isotropic component increases and anisotropic component decreases

as the coke moves from shaft to cohesive zone and downward in the EBF. According

to literature, it is suggested that CO2 predominantly reacts with isotropic component

of a coke on the basis of laboratory experiments. Therefore, the laboratory

measurements of isotropic component conducted under CRI test conditions do not

reflect the actual behaviour of a coke in the EBF. This could be explained on the

basis that the BF atmosphere is different to that in the CRI test and furthermore this

could also illustrate that isotropic measurements may not necessarily be relevant to

coke behaviour in a blast furnace.

The coke strength is reduced which is confirmed by CSREBF and abrasion test results.

The results from CSREBF testing support the idea that alkali accumulation is not

exclusively occurring at the coke surface (+0.5-10 mm fraction in CSREBF) but

penetrates further into the coke structure.

4.6 Comparison of low and high CRI Cokes and their Implications:

A comparison was made with a reference period using high quality coke (low

CRI/high CSR). During high CRI coke trial, the cohesive zone was shifted upwards.

Gas utilization was high but more fluctuating. Despite much high reductant rates, the

high CRI coke resulted in too low hot metal temperature and also provided

significantly high flue dust in off gases when compared to low CRI coke trial.



61

• In both cases, graphitisation of cokes increased as coke descended through the

EBF. However, high CRI coke indicated lower degrees of graphitisation compared

to low CRI coke. The differences in the graphitisation behaviour of two cokes at

tuyere could be attributed to differences in the temperature profiles of the EBF or

due to differences in coke properties particularly their mineral matter.

• Alkali uptake by high CRI coke was greater compared to low CRI coke, which

could be related to differences in coke structure including surface area and

temperature. In both cases, CO2 reactivity of coke increased as they moved into

lower parts of the EBF and could be related to catalytic effect of absorbed alkalis.

Despite large differences in CRI reactivity, the TGA reactivity of coke powders at

lower temperatures was of the similar order. However, low CRI coke did display

less reactivity in the EBF at high temperatures.

Near tuyeres, the porosity of centreline cokes was higher compared to cokes samples

close to furnace walls particularly the proportion of large pores.

In both cokes, anisotropic coke carbon was decreased as the coke descends through

the EBF which is contrary to trends often seen under laboratory conditions. Different

behaviour of isotropic carbon in the EBF and laboratory furnace could be related to

different atmosphere and the more complex environment found in a BF.

The laboratory samples of coke obtained after CRI test did not reflect similar results

as that compared to coke samples obtained from a blast furnace. An increase in coke

alkali content as well as the existence of gases other than CO2 could alter the coke

reactivity to CO2. The study demonstrates the strong influence of carbon structure

and alkalis which are not simulated in conventional CRI tests.



62

5. FUTURE RESEARCH

Several questions about coke degradation, the exact reaction mechanisms that occur

while coke is descending in a BF, still need to be answered. A large amount of coke

excavated from the EBF still remains unstudied, and therefore is useful material for

further investigations.

5.1 Coke degradation

Coke degradation mechanisms should be investigated to a larger extent. This could be

achieved by;

• More SEM studies

• Isolate effects and study them individually.

• Comparison of coke type with extreme variations in quality.

• Further studies can clarify the impact of alkali mineral distribution on coke

reactivity as a consequence of mode and extent of alkali penetration in the coke

matrix, which could influence the coke reactivity mechanisms.

• Further analysis is recommended in order to understand the complex role of

alkali on chemical structure and the mechanisms of alkali assimilation in coke

matrix via condensation or intercalation.

• According to previous studies, coke carbon is believed to be gasified

preferentially along basal planes resulting in reduction of lateral crystallite size

(La) [28]. Further work should be carried out to determine other structural

parameters and their variation correlated with coke descent in EBF.

• Thermodynamic evaluation with regards to stabilisation, absorption, alkali

capacity, experimentation and modelling.

5.2 Optimum coke properties

When a better understanding exists one should try to answer the question how to

make optimum coke and what optimum coke properties consist of. One approach is

to find coals with known resistance to alkalis and use these coals in various mixes

and study the results.



63

6. LIST OF ABBREVIATIONS

EBF Experimental Blast Furnace

CRI Coke Reactivity Index

CSR Coke Strength after Reaction

CSRBF Coke Strength after Blast Furnace Reaction

SEM Scanning Electron Microscope

EDS Energy Dispersive X-ray Analysis

XRD X-Ray Diffraction

XRF X-Ray Fluorescence

LECO Measurement of carbon and sulphur

TGA Thermal Gravimetric Analysis

DTA Differential Thermal Analysis

MS Mass Spectrometer

LOM Light Optical Microscope

Lc The length in the vertical dimension of a crystal

La The length in the horizontal dimension of a crystal

KL Koks Lager (Coke Layer)

CL Coke Layer

C Centre

R Periphery

7. REFERENCES

1. Kobus, K., The History Of Cokemaking, McMaster Cokemaking Course. 2003: Hamilton. p. 1.1-1.13.

2. Carter, W.L., P.C. Chaubal, and M.G. Ranade. Experience with the use of Coke from IHCC at Ispat Inland's No.7 Blast Furnace. IISI Seminar on Coke. 2001. Brussel.

3. Poveromo, J.J. Outlook For Blast Furnace Ironmaking. Coke outlook 2000. 2000. 4. Fruehan, R.J. Future Ironmaking In North America. ICSTI/Ironmaking Conference. 1998.

Toronto.5. Okuno, Y. and M. Nose. Current Status And Future Prospects Of Hot Metal Production In

Japan, South Korea And Taiwan. ICSTI/Ironmaking Conference. 1998. Toronto. 6. Kolb, G. and H.B. Lungen. Current Status And Future Aspects Of Hot Metal Production In

Western EuropE. ICSTI/Ironmaking Conference. 1998. Toronto.



64

7. Song, Y. Development Of Ironmaking Technologies In China. ICSTI/IronmakingConference. 1998. Toronto.

8. Baer, H. Cokemaking With SCS Module Technology Ready For Industrial Implementation.ICSTI/Ironmaking Conference. 1998. Toronto.

9. Gudenau, H.W. Coke Quality Requirements for Modern Blast Furnace Operation. ISS,ICSTI/Ironmaking Conference. 1998. Toronto.

10. Best, M.H., J.A. Burgo, and H.S. Valia. Effect of coke strength after reaction (CSR) on blast furnace performance. 61st Ironmaking Conference Proceedings, Mar 10-13 2002. 2002. Nashville, TN, United States: Iron and Steel Society.

11. Standard Method for measuring Coke Reactivity Index (CRI) and Coke Strength after Reaction (CSR). 1996.

12. Coke - Determination Of Coke Reactivity Index (CRI) And Coke Strength After Reaction (CSR), ISO TC 27/SC 3. 2001.

13. Grosspietsch, K.H., et al. Coke Quality Requirements by European Blast Furnace Operators in the Turn of the Millenium. 4th ECIC. 2000. Paris.

14. Arendt, P., F. Huhn, and H. Kühl, CRI and CSR - A survey of International Round Robins,Cokemaking International. 2001. p. 50-53.

15. Valia, H.S. Prediction of Coke Strength After Reaction With CO2 From Coal Analyses At Inland Steel Company. 1990. Warrendale: Iron and Steel Society.

16. van der Velden, B., International Cokemaking Issues, McMaster Cokemaking Course. 2003: Hamilton. p. 22.1-22.112.

17. Hermann, W., Coke Reactivity and Strength, Coke reactivity - Summary and Outlook Part 1., Cokemaking International. 2002. p. 18-31.

18. Sato, H., J.W. Patrick, and A. Walker, Effect of coal properties and porous structure on tensile strength of metallurgical coke. Fuel, 1998. 77(11): p. 1203-1208.

19. Feng, B., S.K. Bhatia, and J.C. Barry, Structural ordering of coal char during heat treatment and its impact on reactivity. Carbon, 2002. 40(4): p. 481-496.

20. Smoot, L.D. and P.J. Smith, Hetergenous char reaction processes, In Coal Combustion and Gasification. 1985.

21. Sahajwalla, V., M. Dubikova, and R. Khanna. Reductant Characterisation and Selection - Implications for Ferroalloys Processing. Tenth International Ferroalloys Congress. 2004. Cape Town, South Africa.

22. Lu, L., V. Sahajwalla, and D. Harris, Coal char reactivity and structural evolution during combustion-factors influencing blast furnace pulverized coal injection operation.Metallurgical and Materials Transactions B: Process Metallurgy and Materials Processing Science, 2001. 32(5): p. 811-820.

23. Willmers, R.R. and C.R. Bennington. Coke Quality Requirements for Efficient, High Productivity Blast Furnace Operation at High Coal Injection Rates. 2nd International Cokemaking Congress. 1992. London.

24. Forsberg, S. Chemical and physical effects of alkali on blast furnace coke. 1st International Cokemaking Congress. 1987. Essen.

25. Helleisen, M., et al. Characterization of the Behaviour of Coke in the Blast Furnace By Dead Man Coke Samples. 1st International Cokemaking Congress. 1987. Essen.

26. Iwanaga, Y., K. Takatani, and Y. Aminaga. Behaviour of Coke under High Temperature Conditions. European Ironmaking Congress. 1986. Aachen.

27. Steiler, J.M., et al. Tuyere Probing Into the Dead Man of the Blast Furnace - A way to Assess Hearth Phenomena and Coke Behaviour. IIronmaking Conference Proceedings.1991. Washington DC.

28. Tucker, J. and J. Goleczka. Blast Furnace Coke and Its Resistance To Alkalis. 1stInternational Cokemaking Congress. 1987. Essen.

29. Kerkkonen, O. Coal and Coke Requirements and Outlook for Europe. ISSTech 2003. 2003.



65

30. Gudenau, H.W. Fundamentals of New Cokemaking. The first International Congress of Science and Technology of Ironmaking. 1994. Sendai: ISIJ.

31. Beppler, E., et al., Requirements on the Coke Properties Especially When Injecting High Coal Rates, Cokemaking International. 1994. p. 15-24.

32. Lu, L., et al., Quantitative X-ray diffraction analysis and its application to various coals.Carbon, 2001. 39(12): p. 1821-1833.

33. Sahajwalla, V., et al. Reaction Rates and Properties of Cokes during Reaction with Carbon Dioxide and Liquid Iron. Workshop on Science and Technology of Innovative Ironmaking for aiming at Energy Half Consumption. 2003. Tokyo, Japan.

Paper I

STUDY OF GASIFICATION REACTION OF COKES EXCAVATED FROM PILOT BLAST FURNACE

V. Sahajwalla*, T. Hilding**, *S. Gupta, B. Björkman**, R. Sakurovs#, * M. Grigore and N. Saha-Chaudhury*

*University of New South Wales, Sydney, NSW 2052, Australia ** Luleå University of Technology, S-971 87 Luleå, Sweden

# CSIRO, Newcastle, Australia

ABSTRACTFundamental understanding of coke reactions with gas, metal and slag phases is essential for ensuring smooth operation and optimisation of coke performance in existing and advanced blast furnace process, and is dictated by coke properties and blast furnace process conditions. In this study, coke samples excavated from LKAB’s Experimental Blast Furnace (EBF) at MEFOS in Luleå, Swedenwere collected. The centreline quenched coke samples from different zones of this EBF were used to observe the influence of in-furnace reactions on the evolution of coke properties and their associationswith CO2 reactivity. Carbon structure of coke was found to be increasingly ordered, silicon and iron concentration in the coke samples decreased, while alkali concentration particularly potassium and sodium were found to increase as the coke descended towards lower part of the EBF. Both isothermaland non-isothermal reactivity based on TGA measurements showed that coke reactivity towards CO2is increased as coke descends towards cohesive zone despite increasing order of carbon structures.Increased reactivity of cokes at lower parts of EBF was related to alkali enrichments of cokes. The study further shows that increased alkali components in cokes have a strong impact on CO2gasification in EBF such that influence of coke graphitisation could be compensated by catalytic influence of alkalis. To further assist with development of understanding of reactivity of coke,gasification studies were also conducted in a fixed bed reactor at 900ºC using a series of cokes made from Australian coals (varying in rank, maceral and mineral matter). The CO2 reactivity of cokes in a fixed bed reactor was also found to be strongly influenced by the coke minerals compared to carbonstructure. Further studies are required to provide a critical insight into the influence of key parameters such as coke graphitisation and mineral reactions on coke gasification particularly at higher temperatures.

1 INTRODUCTIONCoke has been used as a sole source of fuel in a blast furnace from early 18th century. Sincethen, lots of studies have been conducted on coke and its function in a blast furnace. However, still there are many questions to be answered when it comes to coke and its behaviour and degradation in a blast furnace. One of the major developments in the blast furnace operation is the introduction of pulverized coal technology in which coke is substituted by pulverized coal injection through tuyeres. Economic and environmental pressures are the primary driving force behind the promotion of PCI technology. The old coking plants are gradually closing while few new plants are being built to replace the coke supply particularly in developed countries including Europe. New coke plants are extremely expensive due to stringent environmental regulations. Therefore, in any future, blast furnace operations have to rely onless coke per unit metal production. At low coke rate operation, less amounts of coke in the burden is available in the blast furnace shaft, but it still has to maintain the bed permeabilityfor reducing gases upwards, and liquids passing downwards. Therefore high quality coke is essential for future blast furnace operations.

Coke is produced by heating coal blend in the absence of oxygen to about 1100°C. Cokeperforms three functions in a blast furnace namely thermal, chemical and mechanical: a fuel providing the energy required for endothermic chemical reactions and for melting of iron and slag; a reductant by providing reducing gases for iron oxide reduction; a permeable grid providing for passage of liquids and gases in the furnace, particularly in the lower part of the furnace. When coke passes through a blast furnace, the coke degrades and generates fines which affect bed permeability and affects the process efficiency. The rate at which coke degrades is mainly controlled by solution loss reaction, thermal stress, mechanical stress and alkali accumulation.

Coke quality is often characterized by measuring cold and hot strength, ash composition and chemistry which are largely dictated by coal properties. A range of laboratory tests and procedure have been developed to characterize physical and chemical properties of coke and their potential impacts in blast furnaces. The unspeakably most used and well-known tests are the so-called Coke Reactivity Index and the Coke Strength after Reaction developed by Nippon Steel Corporation (NSC) in Japan in early seventies to assess the effect of CO2reactions on coke. There is no universally accepted standard procedure however NSC/CRI test is widely recognized around the world and was adopted by ASTM as a standard procedure. The test is currently under draft stage for becoming an ISO standard [1, 2].

CRI and CSR have quite different judgements in the interpretation of coke performance in a blast furnace. Generally high CSR coke is believed to prevent the coke from breaking down, improve the permeability of gas and liquid in the wet part of the blast furnace and increase theproductivity as well as decrease the specific coke consumption [3]. However, no international agreement of an ideal way to determine the quality exists as each industry relies on their empirical experience for the interpretation. These laboratory tests are designed to test the coke properties under specific set of conditions which might not be universally suitable. The reproducibility of CRI/CSR values among different laboratories also varies a lot [4]. Therefore, in addition to bench-scale testing, a more comprehensive approach is the pilot-scale testing of materials under more realistic industrial environment. Even though these tests are time consuming and very expensive, data generated in these tests is critical to furtherimprove the interpretation of bench-scale tests.

In the present, a campaign was conducted in LKAB’s EBF at MEFOS in Luleå, Sweden to test the performance of raw materials including coke. The test was conducted using relatively good quality coke i.e. low CRI and high CSR. A large number of samples and data were collected during this campaign. However, a limited part of study is presented here to illustratethe variation in coke properties and its reactivity. Coke samples excavated from the centreline of EBF are used to investigate some of the changes that coke undergoes as it descends into lower part of furnace. The primary focus of this study is to investigate the effect of in-furnacereaction on coke and their consequences on CO2 reactions using weight loss in a TGA reactor.In addition, CO2 reactivity measurements of custom made coke using Australian coals arealso presented to compliment the EBF findings in order to develop an understanding of critical parameters influencing coke reactivity under both lab-scale and industrial environments.

2 EFFECT OF COKE PROPERTIES ON REACTIVITY WITH GASESReactivity towards oxidizing gases including CO2, oxygen, air and steam play an importantrole in many metal-smelting processes. As the primary focus of this study is the gasification,it becomes imperative to discuss some of the factors that influence the coke reactivity withCO2. Combustion/gasification of carbonaceous materials can be divided into three different regimes depending on the steps limiting the reaction rate. Combustion is controlled bychemical reaction at low temperatures (regime I), by pore diffusion at moderate temperatures (regime II) and by gas phase mass transfer at high temperatures (regime III) [5]. Coke properties such as porosity, chemical structure and minerals could influence the coke reactivity in different regions to different extent. For example, atomic structure plays an important role under regime I and II. However, reactivity under combustion regime II, where external or internal diffusion becomes rate limiting is also affected by particle size, porosity.

Porosity of Coke: Reactions with oxidising gases change porous carbon matrix during combustion/gasification. The evolution of pore structure by growth and coalescence leads toincreasing or decreasing available surface areas, changes in pore structure/distribution, gas diffusion and reactivity. Porous structure of coke is governed by the coking properties ofcoals, particularly by maximum fluidity and swelling number [6].

Chemical Composition of Minerals: Transformations of inorganic matter upon heat treatment include changes in chemical bonding, sintering, melting and vaporization as well as mutual interactions with organic matter. In addition to catalytic affect on reactivity ofcarbonaceous materials, mineral matter affects the thermal behaviour of char, and aggregation and particle size of mineral matter affect the fragmentation and mechanical stability of thecarbonaceous material. Hermann [7] has evaluated the effect of chemical composition of coal ash on coke reactivity such that CaO and SO3 are gasification stimulating, Fe2O3 an Al2O3have an intermediate effect, and P2O5, TiO2, MgO are gasification-inhibiting. Feng et al [8] have observed that iron is a major catalyst during gasification of bituminous coal as well organised crystalline structures of carbon were found predominantly in the vicinity of the carbon/iron interface. Moreover, an increasing burnout mineral matter could have inhibitingeffect by forming a barrier for oxidizing gases [9].

Carbon Structure: During its descent through a blast furnace, coke is exposed to extremeconditions. The prevailing high temperatures in the cohesive zone areas lead to a moregraphitized coke i.e. a more ordered structure. Synthetic graphite has a highly ordered structure, high fixed carbon content with low levels of ash and volatile matter. Graphite structure can be described by a regular, vertical stacking of hexagonal aromatic layers with

the degree of ordering characterised by the vertical dimension of the crystallite Lc (Figure 1). Each C atom within the aromatic layer (basal plane) is linked through covalent bonds to three C atoms. However, bonding between the layers is very weak and can easily be broken by external forces. Natural graphite has highly ordered structure like synthetic graphite but contains high level of impurities. Lc for coal/char/coke has been often calculated from X-raydiffraction profiles using Scherrer equation [10].

Figure 1: A schematic of crystal structure of graphite.

3 EXPERIMENTAL

3.1 Pilot-Scale Furnace (LKAB) and Coke SamplesThe current study is based on coke samples from an Experimental Blast Furnace (EBF) situated in Luleå, Sweden which was designed in designed in 1995. Further details and its unique features are described elsewhere [11].The EBF is run campaign wise and a typical campaign last for 6-10 weeks. At the end of each campaign, the furnace is quenched with nitrogen for about ten days in order to stop the prevailing reactions and to cool down the furnace. Within about a minute, the reducing gases are removed and the chemical reactions stop. As the quenching gas is added from the top, an upward moving heat wave, is avoided, thus limiting reactions caused by heat such as further smelting and changes of slag composition. The dissection takes about two weeks to complete. Each burden layer is carefully removed and sampled according to a fixed sampling routine, after observing and documenting the shape, position and any anomalies in the layers that might occur.

The coke samples used in current study are based on cokes excavated from the tenth campaign which lasted for 8 weeks. All coke layers were sampled at three positions; centreline, middle (in-between the centre and the wall) and wall. Figure 2a illustrates the locationsof coke bed layers and codes of excavated coke samples. The drawing of the furnace and the layers are according to scale. The reactivity measurements are carried out only on the centre line coke samples. A simple attempt to estimate the temperature profile was done. Temperatures were estimated from trend lines from a graph based on plot using measuredtemperatures received from probe measurements. Figure 2b illustrates the coke layers ofinterest as a function of tentative temperature estimations. Samples 5 and 10, 15 & 20 are considered to represent coke in stock line and thermal reserve zone while samples 25 & 30 represent coke from the cohesive zone. Sample 35 represents bosh coke.

>700 (C5)

900 – 1200(C10, C15, C20)

1200 –1500 (C30)

> 1500 (C35)

3035

25

2015

10

54.0

5.0

6.0

7.0

8.0600 800 1000 1200 1400 1600 1800

Temperature of furnace zone (oC)

Dis

tanc

e fr

om to

p (m

eter

s)

(a) (b)Figure 2 a) Schematic of EBF illustrating the locations of coke samples; b) Temperatureprofiles of coke samples from EBF.

3.2 Coke Analysis and TGA MeasurementsFigure 3 shows the schematic of Netzsch STA 409 instrument at Luleå University of Technology, which can be used for simultaneous Thermal Gravimetric and DifferentialThermal Analysis. Non-isothermal reactivity was measured by using a small amount of coke powder (60 ~ 80 mg) in an Al2O3 crucible in TGA/DTA equipped with a mass-spectrometerwith the setting to detecting ions with mass of 1 to 65. All samples of interest have been reacted under dynamic heating up to 1300 °C with a heating rate of 10K/minute.

Gas outlet

Furnace

Sample carrier

Radiation shieldProtective tube

Vacuum

Reactive gasProtective gas

Inductive displacementtransducerElectromagneticcompensation system

Vacuum tight casing

TG/DTA carrier

Thermostatic control

Evacuation system

Figure 3. Schematic of TGA/DTA furnace used for non-isothermal reactivity measurement of coke samples.

A TGA at UNSW was used to measure the weight loss in coke sample during isothermalheating at 900 °C for 2 hours under CO2 and at various flow rates ranging from 1.5 to 2.0 l/min. The TGA furnace consists of recrystallised vertical alumina (60 mm ID) tube. Sampletemperature is controlled by an internal thermocouple located close to the sample holder. Approximately 0.2 g sample was placed on a square alumina crucible (30X 30 mm) holder at room temperature. Alumina sample assembly is suspended by a high temperature stainless wire which is connected to a balance that can measure weight changes of the order of 1 microgram (Precisa® 1212 M SCS). The assembly was kept at low temperature zone in the furnace followed by heating up to 900oC at the rate of 2oC/minute while N2 (4-6 l/m) was continuously purged through the furnace which were regulated by Brooks 5850E mass flowcontroller. As the furnace reaches the required reaction temperature, the furnace chamber is raised to move the sample in the reaction zone followed by switching on CO2. Dilution withN2, and various total flow rates were also tested (up to 10 l/min).

Coke samples were analysed by XRF both at SSAB’s laboratories and the University of NewSouth Wales (UNSW), Australia while carbon content was measured using LECO analyser. XRD was used to measure carbon structural parameters. Siemens 5000 X-ray diffractometerat UNSW was used to record scattering intensities of samples by using Copper K radiation(30 kV, 30 mA) as the X-ray source. Samples were packed into an aluminum holder and scanned over an angular range from 5-105° by using a step size of 0.05° and collecting the scattering intensity for 5 seconds at each step. The XRD data has been processed to obtaincrystallite dimensions, e.g. Lc values, in carbonaceous materials. The average stacking height of 002 carbon peak is calculated using Scherrer’s equation. A sharper 002 peak will indicate a larger crystallite size and a greater degree of ordering in the carbon structure.

4 RESULTS & DISCUSSION

4.1 Coke Gasification in Experimental Blast Furnace (EBF) Coke gasification is influenced by coke porosity, carbon structure and its minerals.Preliminary porosity measurements of the coke samples used in this study indicated no significant variation in the porosity of coke samples from different locations in the furnace. Therefore, most of the discussion is limited to variation in carbon structure and coke ash chemistry.

4.1.1 Evolution of Carbon Structure of Coke in EBF

As the coke descends into blast furnace, it reacts with upcoming CO2 gases and loses its carbon. Figure 4 shows that carbon content of coke samples is decreasing such that around bosh region (sample 35) coke loses approximately 5% carbon content resulting in increasedconcentration of ash content in a similar proportion. On the other hand, the nature of carbon phases also gets modified such that less ordered carbon material is lost. Figure 5a shows that carbon atoms become more ordered as coke passes from thermal reserve zone to bosh region as indicated by higher Lc values. Higher Lc value indicates increased number of orderedcarbon layers in the coke. Figure 5a plots the calculated Lc values, generated from XRD measurements on the cokes of interest against the distance. Figure 5b plots the change in Lcvalues against the estimated temperatures of coke layers. The linear correlation suggests thatincrease in Lc value is primarily determined by temperature experienced by coke at a given location. Generally, higher ordered carbons are believed to be less reactive towards oxidising gases including CO2.

82

83

84

85

86

87

4 5 6 7 8


Cok

e ca

rbon

con

tent

(wt.%

)

600

800

1000

1200

1400

1600

1800

2000


)

Central Layer TemperatureCoke carbon content (wt%)P l (C l L T )

Figure 4 Variation in carbon content of EBF centreline coke samples plotted against distance from top of EBF, tentative associated temperatures in EBF are also indicated.

15

30

45

60

4 5 6 7 8


Lcva

lues

of c

oke

(Ang

stro

m)

600

800

1000

1200

1400

1600

1800

2000Tem

perature of cokebed layer ( oC

)Central Layer Temperature

Lc values of central layer cokes

10

20

30

40

600 800 1000 1200 1400 1600 1800

Temperature of coke bed layer (oC)

Lcva

lues

of c

oke

( Ang

stro

m)

(a) (b)Figure 5 a) Increase in Lc values of coke during its journey towards cohesive zone in the EBF and associated temperatures; b) Effect of coke layer temperature on Lc value of cokes.

4.1.2 Evolution of Coke Ash Chemistry in EBF

The XRF analyses of coke samples from centreline position are indicated in Table 1. Figure 6a plots the alkali in coke ash against the furnace depth and shows that alkali content (K2Oand Na2O) in coke increases as the coke moves from shaft to cohesive zone. It appears thatalkali present in recirculation gases inside the blast furnace might be condensed on coke surface or penetrated on the outer surface. Micro structural examination of coke surface will clarify the impact of alkali on coke structure and will be reported later. Figure 6b illustratesthe variation in silica and iron in coke ash. The results are consistent in the sense that silicaand iron reduction is believed to increase at higher temperatures.

Table 1. Chemical composition of EBF coke samples and coke used in fixed bed reactor.XRF(SSAB) Ash SiO2 Al2O3 Fe2O3 CaO MgO K2O Na2O TiO2 P2O5 SO3Coke samples from Pilot-scale Furnace (LKAB)KL05C 9.83 5.76 2.63 0.42 0.01 0.06 0.17 0.10 0.16 0.03 0.59KL10NC 10.89 6.39 2.77 0.46 0.04 0.06 0.35 0.16 0.17 0.03 0.50KL15NC 11.66 6.30 2.65 0.37 0.00 0.06 1.24 0.43 0.15 0.02 0.54KL20NC 11.52 5.58 2.57 0.32 0.02 0.07 1.78 0.61 0.14 0.02 0.51KL25NC 12.39 5.83 2.61 0.32 0.02 0.08 2.31 0.67 0.14 0.02 0.47KL30NC 12.39 5.97 2.66 0.34 0.04 0.08 2.07 0.68 0.14 0.02 0.50KL35C 13.34 5.81 2.64 0.27 0.00 0.08 3.21 0.85 0.12 0.22 0.44Cokes used in fixed reactor (UNSW) Coke-A1 9.0 5.12 1.65 1.15 0.33 0.14 0.08 0.04 0.10 0.12 0.18Coke-A2 11.8 5.70 4.47 0.63 0.30 0.07 0.06 0.08 0.17 0.22 0.04

*CO2 surface areas of coke A-1 and coke A-2 were 142 and 74 m2/g respectively at 15% burnout. KL05C represents the feed coke sample in EBF.

0

10

20

30

40

4 5 6 7 8


K2O

and

Na 2

O in

cok

e as

h (w

t.%)

600

800

1000

1200

1400

1600

1800

2000


)

Central Layer TemperatureK2O %Na2O %

SiO2 %

Iron

Central LayerTemperature

0

10

20

30

40

50

60

70

4 5 6 7 8


SiO

2 in

coke

ash

(wt.%

)Iro

n ox

ides

in c

oke

ash

(10x

wt%

)

0

200

400

600

800

1000

1200

1400

1600

1800

Tem

pera

ture

of c

oke

bed

laye

r (°C

)

(a) (b)

Figure 6 a) Variation in sodium and potassium concentration in coke ash samples in EBF, b)Loss of silica and iron content in coke ash in EBF coke samples.

4.1.3 Evolution of Coke Reactivity with CO2

Figure 7a compares the non-isothermal reactivity of coke samples from different locations inthe EBF. This demonstrates that as the coke descends in the blast furnace, its reactivity increases. Similar trend was also observed during isothermal reactivity measurements. Figure7b compares the isothermal reactivity of coke samples collected from top (5C) and bottom zone (35C) indicating that CO2 reaction is faster for cokes from lower parts of the furnace.Interestingly, this reactivity trend could not be explained on the basis of carbon structure which is found to increase as the coke moves down the furnace. Most likely, the increase incoke reactivity is related to alkali enrichments. This means that in the EBF, the coke reactivity can increase due to catalytic influence of alkali even the carbon structure becomes ordered.

0.0

0.2

0.4

0.6

0.8

0 1200 2400 3600 4800 6000 7200Time (seconds)

Car

bon

conv

ersio

n, (w

o-w

)/ w o

KL35C

KL05C

(a) (b)

TGA results of centre line cokes

15

25

35

45

55

65

75

85

95

105

800 850 900 950 1000 1050 1100 1150 1200 1250 1300

Temperature °C

Wei

ght l

oss

%

KL01C

KL10C

KL05C

KL15C

KL20C

KL25C

KL30C

KL35C

Figure 7 a) Loss in wt. of EBF coke samples with increasing temperature in a TGA/DTAfurnace; b) Loss in weight of two coke samples (5C & 35C) with time at 900oC in a TGA furnace at UNSW.

4.2 Coke Gasification in a Fixed Bed Reactor

In a pilot-scale facility, it is difficult to isolate the effect of various factors of coke properties on coke reactivity. Fundamental aspects of coke reactivity can be investigated under laboratory conditions by selecting suitable coke properties. For this purpose, CooperativeResearch Centre for Coal in Sustainable Development (CCSD), is conducting a comprehensive research program under which a range of coke samples were prepared using Australian bituminous coals in the 8 kg capacity ovens at 1000 C and heating rate of 3 C min-

1 in nitrogen environment. Coke samples were tested for their reactivity towards metal and CO2 and H2O. Gasification results of two cokes (similar rank, ash content and carbon structure) are presented to demonstrate the effect of coke minerals on their reactivity differences under laboratory conditions. Coke analysis and chemical composition of its ash isprovided in Table 1. Coke reactivity was measured in a fixed bed reactor using 1-2 g of sample at atmospheric pressure and flow rate of 0.700 L min-1 of CO2. Two infrared analyserswere used to determine the concentration of CO and CO2 in exhausted gases, which were used to calculate the reaction rate. Further details can be found elsewhere [21].

Figure 9a compares the apparent reaction rate at 900 ºC with increasing burn off. At 15% conversion, the CO2 surface areas of coke A-1 and coke A-2 increased to 142 and 74. The apparent reaction rates were divided by surface are to calculate the specific reaction rates inorder to exclude the effect of surface area. Figure 9b shows that specific reaction rate of coke A-1 is higher than that of coke A-2. In this study, it was shown that higher CO2 reactivity of coke A-1 could not be attributed to carbon structure [21]. Authors proposed that differences in coke reactivity could be attributed to the presence of sulfide and oxide mineral in cokes as illustrated in Figure 10. Coke A-1 which indicated higher frequency of sulfide minerals suchas pyrrhotite (Fe1-xS)/troilite (FeS) was found to be more reactive when compared to coke A-2 which contained higher amounts of iron oxides. The observations were consistent with previous researchers who also noticed the catalytic effect of iron on coke gasification with CO2 [22-24]. Comparison of coke gasification in both laboratory and EBF cases highlights the significance of coke mineral reactions on coke gasification. Further research is required tounderstand the exact mechanisms of coke mineral reactions during different conditions of gasification and their potential influence on CO2 reactivity.

0

10

20

30

40

50

0.00 0.04 0.08 0.12 0.16Fractional Weight Loss

App

aren

t Rat

e x

106 (

g/g/

s)Coke A-1

Coke A-2

0.0

0.2

0.4

Coke A-1 Coke A-2

Spec

ific

Rat

e (m

g/g/

s)

(a) (b)

Figure 9 a) Weight loss in cokes with increasing burnout during CO2 gasification in a fixed bed reactor at 900oC; b) Specific reactivity of the same cokes at 15 % burn off.

(a) (b)

Figure 10 a) SEM images of typical sulfides and b) oxides mineral phases of iron in both cokes. Coke A appears to have higher iron sulphide minerals.

5 CONCLUSIONS

The coke samples from an experimental blast furnace were reacted with CO2 to monitorweight loss at 900oC in TGA furnace while at the same temperature, custom made coke from Australian coals were reacted in a fixed reactor. The study showed that coke properties in experimental blast furnace continuously change as coke moves down from thermal reservezone towards cohesive zone which influences the CO2 reactivity. In experimental blast furnace, carbon structure of coke becomes increasingly ordered leading to increasedgraphitisation, which is attributed to increasing temperatures experienced by coke as it descends into the blast furnace. In the same coke samples, silicon and iron concentration was decreased while alkali concentration, particularly potassium, was found to increase. Both isothermal and non-isothermal reactivity based on TGA measurements clearly established that coke reactivity towards CO2 is improved as the coke moves from thermal reserve zone tocohesive zone and onward in the EBF despite increasing order in their carbon structures. Increased reactivity of cokes at lower parts of furnace could be related to increasedconcentration of alkali components in cokes. The study suggests that the potential adverse effect of coke graphitisation due to thermal annealing could be compensated by catalytic influence of alkali components of coke in the present test campaign. Most likely, the increased alkali components in coke might be related to condensed alkali from recirculating

gases, however microstructural examination of coke will confirm the mechanisms of alkali enrichment in the EBF samples.

Australian research shows that coke gasification in a fixed bed reactor at 900ºC is also strongly influenced by coke mineral rather than degree of ordering of carbon layers in cokes. In both laboratory and industrial and environments of coke gasification, coke minerals were found to have a significant influence on CO2 gasification. Further studies are required toprovide a critical insight into the key factors that affect the CO2 gasification of coke particularly the mechanisms of coke mineral reactions and coke graphitisation during different environments of gasification.

6 REFERENCES

[1] ASTM Standard D 5341-93. Standard Method for Measuring Coke Reactivity Index (CRI) and Coke Strength After Reaction (CSR).

[2] ISO / CD 18894. Coke – Determination of Coke Reactivity Index (CRI) and Coke Strength After Reaction (CSR)

[3] Grossspietsch, K.H., Lyngen, H.B., Dauwels, G., Ferstl, Karjalahti, T., van der Velden, B.,Willmers, R. “Coke quality requirements by European blast furnace operators in the turn of the millennium”, 4th European Coke and Ironmaking Congress Proceedings, vol. 1,2000, pp 1-11.

[4] Arendt, P., Huhn, F., and Kühl, H. ”CRI and CSR – A survey of International Round Robins”,Cokemaking International, 2/2001, pp 50-53.

[5] Smith, K.L., Smoot, L.D., Fletcher, H.T. and Pugmire, R.J. “The structure and reactionprocesses of coals”, The Plenum Chemical Engineering Series, Plenum Press, New York, USA,1994.

[6] Sato, H., Patrick, J.W. and Walker, A. “Effect of coal properties and porous structure on tensile strength of metallurgical coke”, Fuel, 1998, 77, pp. 1203-1208.

[7] Hermann, W. “Coke reactivity and strength, Part 1: Coke reactivity-summary and outlook”,Cokemaking International, 1, 2002, pp. 1, 18, 20, 22, 24,26, 28-31.

[8] Feng, B., Bhatia, S.K. and Barry, J.C. “Structural ordering of coal char during heat treatment and its impact on reactivity”, Carbon, 40, 2002, 481-496.

[9] Smoot, L.D. and Smith, P.J. “Heterogeneous char reaction processes”, In Coal Combustion andGasification. Plenum, New York, 1985, pp. 77-110.

[10] Sahajwalla, V., Dubikova, M. and Khanna, R. ”Reductant Characterisation and Selection –Implications for Ferroalloys Processing”, Proceedings of the Tenth International FerroalloysCongress, Cape Town, South Africa, February 1-4. 2004.

[11] Dahlstedt, A., Hallin, M. and Wikström, J.-O. (2000), “Effect of Raw Materials on Blast Furnace Operation: The Use of an Experimental Blast Furnace”, 4th European Coke andIronmaking Congress Proceedings, 1, pp. 138-145.

[12] Hooey, L., Sterneland, J. and Hallin, M. (2001), Evaluation of Operational Data from the LKAB Experimental Blast Furnace”, 60th Ironmaking Conference Proceedings, 60, pp. 197-208.

[13] Raanes, O., Kolbeinsen, L. and Byberg, J.A. “Statistical analysis of properties for coals used in the production of silicon rich alloys”, 8th International Ferroalloys Congress ProceedingsINFACON, Beijing, China, China Science & Technology Press, 1998, pp. 116-120.

[14] Todd, S.J. and Hansen, H. “Nature and origin of coal and its petrographic characteristics”, Electric Furnace Conference Proceedings, vol. 53, Orlando, 1995, pp. 3-13.

[15] Smith, K.L., Smoot, L.D., Fletcher, H.T. and Pugmire, R.J. “The structure and reactionprocesses of coals”, The Plenum Chemical Engineering Series, Plenum Press, New York, USA,1994.

[16] Pistorius, P.C. “Reductant selection in ferro-alloy production: The case for the importance of dissolution in metal”, J. S. Afr. Inst. Min. Metall., 200, 33-36.

[17] Spratt, D.M. and Brosnan, J.G. ”Successful production of silicon metal in western Australia”,Electric Furnace Conference Proceedings, vol. 48, Iron and Steel Society/AIME, New York,1990, pp. 217-223.

[18] FAO.” Simple technologies for charcoal making” FAO Forestry Paper-41, 1987.[19] Emmerich, F.G. and Luengo, C.A. “Babassu charcoal: a sulfurless renewable thermo-reducing

feedstock for steelmaking”, Biomass and Bioenergy, 10, 1, 1996, pp. 41-44. [20] Areklett, I. and Nygaard, L.P. “The ferroalloys industry: Back to charcoal?” Greenhouse Issues,

No 60, 2002.[21] Sahajwalla, V., Sakurovs, R., Grigore, M., Cham, S.T., and Dubikova, M. ( 2003), “Reaction

rates and properties of cokes during reaction with carbon dioxide and liquid iron, Proceedings ofthe Workshop on Science and Technology of Innovative Ironmaking for aiming at Energy HalfConsumption, Tokyo, November 27-28, 2003.

[22] Price, T.,. Iliffe, M.J., Khan, M.A. and Gransden, J.F.: Ironmaking Proc. ISS-AIME Conference, (1994), 79.

[23] Van der Velden, B., J. Trouw, R. Chaigneau, J. Van den Berg: Proc. ICSTI/IronmakingConference, (1999), 275.

[24] Leonard, D.C., Bonte, L., Dufour, A., Ferstl, A., Raipala, K., Schmole, P. Schoone, E.E.,Verduras, J.L. and Willmers, R.R.: Proc. 3rd European Ironmaking Congress, (1996), 1.

Paper II

Thermochimica Acta

THERMAL ANALYSIS OF COKE SAMPLED FROM DIFFERENT LEVELS IN AN EXPERIMENTAL BLAST FURNACE

Tobias Hilding1, Bo Björkman1, Jan-Olov Wikström2 and Menad Nourreddine1

1 Department of Chemical Engineering and Geosciences. Division of Process Metallurgy. Luleå University of Technology, S-971 87 Luleå, Sweden

Tel: +46 920 49 10 00; Fax: +920 49 11 99 E mail: [email protected]

2MEFOS – Metallurgical Research Institute AB BOX 812, 971 25 Luleå, Sweden

Tel: +46 49 20 19 00 E-mail: [email protected]

Corresponding author: Tobias Hilding

AbstractModern blast furnace technology is characterized by a minimum of total fuel consumption. Today coke is replaced to a high extent by injection of hydrocarbons. The less available coke needs to have high mechanical strength and be resistant toward chemical attacks, particular CO2 and alkali. In this study, coke samples excavated from LKAB’s Experimental Blast Furnace (EBF) at MEFOS in Luleå, Sweden were collected. The centreline and peripheral nitrogen-quenched coke samples from different zones of this EBF were used to observe the influence of in-furnace reactions on the evolution of coke properties and their associations with CO2 reactivity. It was found that reactivity with CO2increases while going from shaft to tuyere level. Further, the reaction pattern is different for the layers studied. Increased reactivity of cokes at lower parts of EBF was related to alkali enrichments of cokes. The study further shows that increased alkali components in cokes have a strong impact on CO2gasification in EBF such that influence of coke graphitisation could be compensated by catalytic influence of alkalis.

Key words: coke, blast furnace, reactivity, ThermoGravimetric Analysis (TGA)

1. Introduction Coke has been used as a sole source of reductant in a blast furnace from early 18th century. A

lot of studies have been conducted on coke and its function in a blast furnace. However, still

there are many questions to be answered when it comes to coke and its behaviour and

degradation in a blast furnace. One of the major developments in the blast furnace operation

is the introduction of pulverized coal technology in which coke is substituted by pulverized

coal injection through tuyeres. Economic and environmental pressures are the primary driving

force behind the promotion of PCI technology. The old coking plants are gradually closing

while few new plants are being built to replace the coke supply particularly in developed

countries including Europe. New coke plants are extremely expensive due to stringent

environmental regulations. Therefore, in any future, blast furnace operations have to rely on

less coke per unit metal production. At low coke rate operation, less amounts of coke in the

burden is available in the blast furnace shaft, but it still has to maintain the bed permeability

for reducing gases upwards, and liquids passing downwards. Therefore high quality coke is

essential for future blast furnace operations [1-3].

Coke is produced by heating coal blend in the absence of oxygen to about 1100 °C. Coke

performs three functions in a blast furnace namely thermal, chemical and mechanical: a fuel

providing the energy required for endothermic chemical reactions and for melting of iron and

slag; a reductant by providing reducing gases for iron oxide reduction; a permeable grid

providing for passage of liquids and gases in the furnace, particularly in the lower part of the

furnace. When coke passes through a blast furnace, it degrades and generates fines which

affect bed permeability and affects the process efficiency. The rate at which coke degrades is

mainly controlled by solution loss reaction (C + CO2 2 CO), thermal stress, mechanical

stress and alkali accumulation [4-6].

Coke quality is often characterized by measuring cold and hot strength, ash composition and

chemistry which are largely dictated by coal properties. A range of laboratory tests and

procedure have been developed to characterize physical and chemical properties of coke and

their potential impacts in blast furnaces. The unspeakably most used and well-known tests are

the so-called Coke Reactivity Index (CRI) and the Coke Strength after Reaction (CSR)

developed by Nippon Steel Corporation (NSC) in Japan in early seventies to assess the effect

of CO2 reactions on coke. There is no universally accepted standard procedure, however

NSC/CRI test is widely recognized around the world and was adopted by ASTM as a standard

procedure. The test is currently in the draft stage for becoming an ISO standard [1, 7, 8].

CRI and CSR have quite different judgements in the interpretation of coke performance in a

blast furnace. Generally high CSR coke is believed to prevent the coke from breaking down,

improve the permeability of gas and liquid in the wet part of the blast furnace and increase the

productivity as well as decrease the specific coke consumption [3]. However, no international

agreement of an ideal way to determine the quality exists as each industry relies on their

empirical experience for the interpretation. These laboratory tests are designed to test the coke

properties under specific set of conditions which might not be universally suitable. The

reproducibility of CRI/CSR values among different laboratories also varies a lot [9].

Therefore, in addition to bench-scale testing, a more comprehensive approach is the pilot-

scale testing of materials under more realistic industrial environment. Even though these tests

are time consuming and very expensive, data generated in these tests is critical to further

improve the interpretation of bench-scale tests.

In the present study, a campaign was conducted in LKAB’s EBF at MEFOS in Luleå, Sweden

to test the performance of raw materials including coke. The test was conducted using

relatively good quality coke i.e. low CRI and high CSR. A large number of samples and data

were collected during this campaign. However, a limited part of results are presented here to

illustrate the variation in coke properties and its reactivity towards CO2. Coke samples

excavated from the centreline of EBF are used to investigate some of the changes that coke

undergoes as it descends into lower part of furnace. The primary focus of this study is to

investigate the effect of in-furnace reaction on coke and their consequences on CO2 reactions

using weight loss in a Thermogravimetric analysis equipment (TGA).

2. Material and experimental procedure

The current study is based on coke samples from an Experimental Blast Furnace (EBF),

owned by LKAB and situated in Luleå, Sweden. Further details and its unique features are

described elsewhere [10-12].

The EBF is run campaign wise and a typical campaign last for 6-10 weeks. At the end of each

campaign, the furnace is quenched with nitrogen for about ten days in order to stop the

prevailing reactions and to cool down the furnace. Within about a minute, the reducing gases

are removed and the chemical reactions stop. As the quenching gas is added from the top, an

upward moving heat wave, is avoided, thus limiting reactions caused by heat such as further

smelting and changes of slag composition. In the dissection each burden layer is carefully

removed and sampled according to a fixed sampling routine, after observing and documenting

the shape, position and any anomalies in the layers that might occur.

2.1 Coke Samples used in this study

The coke samples used in current study are based on cokes excavated from the tenth and

eleventh campaigns which both lasted for 8 weeks.

For experimental blast furnace campaign number 10, all coke layers were sampled at three

positions; centre line, intermediate (in-between the centre and the wall) and wall. Figure (1a)

illustrates the locations of coke bed layers and codes of excavated coke samples. The drawing

of the furnace and the layers are according to scale. The reactivity measurements are carried

out only on the centre line coke samples. A simple attempt to estimate the temperature profile

was done. Temperatures were estimated from trend lines from a graph based on plot using

measured temperatures received from probe measurements. Figure (1b) illustrates the coke

layers of interest as a function of tentative temperature estimations. Samples 5, 10, 15 & 20

are considered to represent coke in stock line and thermal reserve zone while samples 25 &

30 represent coke from the cohesive zone. Sample 35 represents bosh coke.

For campaign number 11, sampling procedure was the same, except for that six samples were

collected from each layer instead of three. Further, it was possible to sample 43 coke layers.

The difference and similarity in material and operation are reported in table 1.

2.2 Experimental procedure

Figure 2 shows the schematic of Netzsch STA 409 instrument at Luleå University of

Technology, which can be used for simultaneous ThermoGravimetric and Differential

Thermal Analysis. Non-isothermal reactivity was measured by using a small amount of coke

powder (60 ~ 80 mg) in an Al2O3 crucible in TGA/DTA equipped with a quadro pole mass-

spectrometer with the setting to detecting ions with mass of 1 to 65. All samples of interest

have been reacted under dynamic heating up to 1300 °C with a heating rate of 10 K/minute

and in 100 % CO2 atmosphere.

3. Results and discussion

Measurements of carbon and sulfur content in the LECO equipment indicate that carbon

content decreases as coke descends through the EBF (see figure 3 and ref. [13]), thus

confirming what to be expected.

The X-ray fluorescence (XRF) analyses of coke samples from centreline position are

indicated in Tables 2 and 3. Previous study [13] showed that alkali content (K2O and Na2O) in

coke increases as the coke moves from shaft to cohesive zone. It appears that alkali present in

recirculation gases inside the blast furnace condenses on coke surface or penetrate into the

coke matrix. It is generally believed that the presence of high alkali could enhance coke

reactivity towards oxidizing gases including CO2.

The results obtained from TG measurements of different coke layers taken from EBFC10 and

EBFC11 are shown in Figure 4 and 5 respectively. These figures illustrate the evolution of the

mass % change of the coke layers between 800 and 1 300 °C under CO2 atmosphere. From

these curves, it can be observed that the reactivity of the coke is increasing as it descends in

the blast furnace as shown in figure (1a). Similar trend was also observed during isothermal

reactivity measurements [13]. It can be noted that the trend is the same for both of the

campaigns covered in this study.

Figures 6 and 7 give the DTG curves of coke layers from EBFC10 and EBFC11 respectively.

It can be noted that solution loss reaction starts earlier for coke layers found in the lower parts

of the EBF. The alkaline substances contained in the coke play a role of catalyst. In such a

way as to increase the reaction rate of the reaction with coke and CO2. Comparing layer 01

and 35 from EBFC10, the solution loss reaction starts at below 800°C and at 1000°C,

respectively. The increase in alkali when comparing these layers is about tenfold.

The gas analyser was primarily used to detect the evolution of carbon monoxide. In figure 8

the ion current of CO is plotted as a function of time. It can be observed that the evolution

starts earlier for increasing depth in the furnace. The results support the conclusions drawn

from TGA and DTG results.

Interestingly, this reactivity trend could not be explained on the basis of carbon structure

which is found to increase as the coke moves down the furnace, see figure 9 and ref. [13]. An

increased order of the carbon structure leads to a lower reactivit. Thus, most likely, the

increase in coke reactivity is related to alkali enrichments which catalyse the solution loss

reaction. This means that in the EBF, the coke reactivity can increase due to catalytic

influence of alkali even when the carbon structure becomes more ordered.

4. Conclusions

The study showed that coke properties in the EBF continuously change as coke moves down

from thermal reserve zone towards cohesive zone which influences the CO2 reactivity. In the

EBF, carbon structure of coke becomes increasingly ordered leading to increased

graphitisation which is attributed to increasing temperatures experienced by coke as it

descends into the blast furnace. In the same coke samples, silicon and iron concentration was

slightly decreased while alkali content, particularly potassium, was found to increase. Non-

isothermal reactivity based on TGA measurements as well as DTA/DDTA information clearly

establish that coke reactivity towards CO2 is increased as coke moves from thermal reserve

zone to cohesive zone and onward in the EBF despite increasing order in carbon structure.

Increased reactivity of cokes at lower parts of furnace could be related to increased

concentration of alkali components in cokes, as also suggested by other authors [14, 15]. The

study suggests that the potential adverse effect of coke graphitisation due to thermal annealing

could be more than compensated by catalytic influence of alkali components of coke in the

present test campaign. Most likely, the increased alkali components in coke might be related

to condensed alkali from recirculating gases.

5. Acknowledgement

The author would like to thank MEFOS, SSAB and LKAB for providing the samples and the

opportunity to conduct this research. Special thanks to Professor Veena Sahajwalla, Dr

Nourredine Menad and Dr Sushil Gupta as well as other staff at the University of New South

Wales and Luleå University of Technology for their assistance in using equipment and

facilities. Thanks to JERNKONTORET and STEM for financial support.

6. References

[1] M. H. Best, J. A. Burgo and H. S. Valia, Iron and Steel Society, Nashville, TN, United States, (2002), 213-239.

[2] I. Christmas, Iron and Steelmaker (1999), 41-44. [3] K. H. Grosspietsch, H. B. Lungen, G. Dauwels, Ferstl, T. Karjalahti, P. Negro, B. van

der Velden and R. Willmers, Paris, (2000), 2-11. [4] K. H. Grosspietsch and H. B. Lungen, Brussels, (2001), 101-118. [5] J. J. Poveromo, McMaster Ironmaking Course. 2003: Hamilton. [6] B. van der Velden, McMaster Cokemaking Course. 2003: Hamilton. p. 22.1-2.112. [7] Coke - Determination of Coke Reactivity Index (CRI) and Coke Strength after

Reaction (CSR), ISO TC 27/SC 3. 2001. [8] Standard Method for measuring Coke Reactivity Index (CRI) and Coke Strength after

Reaction (CSR). 1996. [9] P. Arendt, F. Huhn and H. Kühl, Cokemaking Int., 2(2001), 50-53. [10] A. Dahlstedt, M. Hallin and M. Tottie, Luleå, Sweden, (1999), 235-252. [11] A. Dahlstedt, M. Hallin and J.-O. Wikström, Paris, (2000), 138-145. [12] L. Hooey, J. Sterneland and M. Hallin, ISS, Baltimore, (2001), 197-208. [13] T. Hilding, V. Sahajwalla, S. K. Gupta, B. Björkman, R. Sakurovs, M. Grigore and N.

Saha-Chaudhury, Luleå, Sweden, (2004) [14] S. Forsberg, Essen, (1987), C6.1-C6.19. [15] B. van der Velden, ISS, Baltimore, (2001), 269-280.

8. Figures and tables.

>700 (C5)

900 – 1200 (C10, C15, C20)

1200 –1500 (C30)

> 1500 (C35)

3035

25

2015

10

54.0

5.0

6.0

7.0

8.0600 800 1000 1200 1400 1600 1800

Temperature of furnace zone (oC)

Dis

tanc

e fr

om to

p (m

eter

s)

Figure 1 a) Schematic of EBF illustrating the locations of coke samples; b) Temperatureprofiles of coke samples from EBF based on assumptions generated from vertical, horizontal and inclined temperature probe measurements.

Gas outlet

Furnace

Sample carrier

Radiation shieldProtective tube

Vacuum

Reactive gasProtective gas

Inductive displacementtransducerElectromagneticcompensation system

Vacuum tight casing

TG/DTA carrier

Thermostatic control

Evacuation system

Figure 2. Schematic of TGA/DTA furnace used for non-isothermal reactivity measurement of coke samples.

LECO C %

40

50

60

70

80

90

100

4 5 6 7 8 9 10

Distance from top (m)

% C Centre

Wall

Figure 3. Displayed in figure is result of carbon content measurement of cokes from campaign 10 by LECO method. Includes hearth samples.

TGA results of centre line cokes EBFC10

15

25

35

45

55

65

75

85

95

105

800 900 1000 1100 1200 1300Temperature °C

Wei

ght %

KL01C

KL10C

KL05C

KL15C

KL20C

KL25C

KL30C

KL35C

Figure 4. TG results from selected coke layers from EBFC 10.

TGA results of centre line cokes EBFC11

15

25

35

45

55

65

75

85

95

105

800 900 1000 1100 1200 1300

Temperature °C

Wei

ght

%

KL01CKL05C

KL10C

KL25C

KL30CKL40C

Figure 5. TG results of selected coke layers from EBFC 11.

Figure 6. DTG results for selected coke layers from EBFC 10.

Figure 7. DTG results for selected coke layers from EBFC 11.

Figure 8. CO evolution for selected coke layers from EBFC 10.

15

30

45

60

4 5 6 7 8


Lc v

alue

s of

cok

e (A

ngst

rom

)

600

800

1000

1200

1400

1600

1800

2000Tem

perature of coke bed layer ( oC)

Central Layer Temperature

Lc values of central layer cokes

Figure 9. Increase in Lc values of coke during its journey towards cohesive zone in the EBF and associated temperatures based on assumptions generated from vertical, horizontal and inclined temperature probe measurements. Includes hearth samples.

Table 1. Differences and similarities for EBFC 10 and 11.Prior to quenching EBFC 10 EBFC 11

Injection fuel Oil. ~100kg/thm Coal. ~105kg/thmFerrous burden LKAB Pellets LKAB PelletsCoke SSAB coke spring 2002 SSAB coke spring 2003 CRI & CSR 23.2 & 68.8 respectively 19.4 & 71.6 respectively.

Table 2. Chemical composition of EBFC10 coke samples.XRF(SSAB) Sum ox. SiO2 Al2O3 Fe2O3 CaO MgO K2O Na2O TiO2 P2O5 SO3Coke samples from Pilot-scale Furnace (LKAB) Campaign 10KL01C 11.4 6.50 3.15 0.58 0.15 0.10 0.37 0.15 0.17 0.04 0.18KL05C 11.59 5.76 2.63 1.20 0.01 0.06 0.17 0.10 0.16 0.03 1.47KL10C 12.53 6.39 2.77 1.32 0.04 0.06 0.35 0.16 0.17 0.03 1.25KL15C 13.26 6.30 2.65 1.06 0.00 0.06 1.24 0.43 0.15 0.02 1.35KL20C 12.98 5.58 2.57 0.92 0.02 0.07 1.78 0.61 0.14 0.02 1.27KL25C 13.77 5.83 2.61 0.92 0.02 0.08 2.31 0.67 0.14 0.02 1.17KL30C 13.89 5.97 2.66 0.97 0.04 0.08 2.07 0.68 0.14 0.02 1.25KL35C 14.80 5.81 2.64 0.77 0.00 0.08 3.21 0.85 0.12 0.22 1.10

Table 3. Chemical composition of EBFC11 coke samples. XRF(SSAB) Sum ox. SiO2 Al2O3 Fe2O3 CaO MgO K2O Na2O TiO2 P2O5 SO3Coke samples from Pilot-scale Furnace (LKAB) Campaign 11 KL01C 10.10 6.13 2.63 1.12 0 0.06 0.18 0.05 0.17 0.026 1.45 KL05C 16.59 6.27 4.07 5.63 0.39 0.11 2.35 0.22 0.11 0.025 1.10 KL10NC 10.28 6.23 2.74 1.09 0 0.04 0.16 0.05 0.16 0.032 1.52 KL15NC 10.59 6.31 2.74 1.32 0 0.05 0.26 0.06 0.17 0.03 1.50 KL20NC 11.44 6.66 2.76 1.54 0.14 0.11 0.43 0.07 0.17 0.029 1.45 KL25NC 10.04 5.61 2.44 2.06 0.07 0.05 0.29 0.07 0.16 0.029 1.42 KL30NC 13.93 6.21 2.74 0.83 0 0.06 3.59 0.42 0.12 0.026 1.20 KL35C 18.98 4.63 1.86 17.84 1.94 0.19 1.01 0.19 0.08 0.025 0.65 KL40C 12.44 6.03 2.53 0.97 0 0.05 2.58 0.34 0.13 0.023 1.27

Paper III

Degradation Behaviour of a High CSR Coke in an Experimental Blast Furnace:

Effect of Carbon Structure and Alkali Reactions

Tobias Hilding, *Sushil Gupta, *Veena Sahajwalla, Bo. Björkman and **Jan-Olov Wikström

Division of Process Metallurgy, Luleå University of Technology, S-971 87 Luleå, Sweden

*Cooperative Research Centre for Coal in Sustainable Development, School of Materials Science and

Engineering, The University of New South Wales, Sydney, NSW 2052, Australia

**MEFOS, Luleå, Sweden

• Contact addresses of authors

Tobias.Hilding: [email protected]

Sushil Gupta: [email protected]

Veena Sahjawalla : [email protected]

Mailing and correspondence address

Professor Veena Sahajawalla Cooperative Research Centre for Coal in Sustainable Development, School of Materials Science and Engineering,The University of New South Wales, Sydney, NSW, 2052, Australia

3

Synopsis

A high CSR coke was tested in the LKAB’s Experimental Blast Furnace (EBF) at Luleå. The evolution of

physical and chemical properties of the centre-line coke samples were analysed by Light Optical Microscopy

(LOM), BET N2 absorption and SEM/XRF/XRD. Alkali distribution in the EBF cokes was examined by

XRF/SEM and EDS. Thermo Gravimetric Analysis (TGA) was used to measure isothermal and non-

isothermal CO2 reactivity of the cokes. The crystalline order of carbon and the concentration of alkalis were

found to increase as the coke descended through thermal reserve zone to the cohesive zone of the EBF. The

crystallite height (Lc) of EBF coke carbon displayed a linear correlation with the measured EBF

temperatures demonstrating the strong effect of temperature on carbon structure of coke in the EBF. Alkali

concentration of the coke was increased as it descended into the EBF, and was uniformly distributed

throughout the coke matrix. The CO2 reactivity of lower zone cokes was found to increase when compared to

the reactivity of the upper zones cokes, and was related to the catalytic effect of increased alkalis

concentration. The deterioration of coke quality particularly coke strength and abrasion propensity were

related to coke graphitisation, alkalization and reactivity. Coke graphitisation is shown to have a strong

influence on the coke degradation behaviour in the EBF.

KEY WORDS: Coke, CSR, Abrasion, Graphitisation, XRD, Gasification, TGA reactivity, Alkali

4

1. Introduction

The Blast Furnace (BF) is the most dominant process of ironmaking worldwide. Coke is the most

important and expensive raw material in the BF and has a strong influence on the process efficiency and hot

metal quality. Due to increasing environmental concerns and recent shortage of coke, there is an increasing

trend to replace coke by other fuels such as injecting pulverized coal through tuyeres in order to decrease

reliance on coke. Coke performs several functions in a blast furnace namely thermal, chemical and

mechanical: a fuel providing the energy required for endothermic chemical reactions and for melting of iron

and slag; a reductant by providing gases for iron oxide reduction; a permeable grid providing the passage for

liquids and gases in the furnace. At low coke rate BF operation, less amount of coke is available in the

burden to maintain sufficient permeability of bed. As the coke moves towards lower zones of a blast furnace,

it degrades and generates fines, which affects the bed permeability and the process efficiency. Therefore,

superior coke quality is critical for a stable and efficient blast furnace operation under low coke rate

conditions.

The intensity of coke degradation in a blast furnace is strongly influenced by coke properties and

associated phenomena such as solution loss reaction, thermal stress, mechanical stress and alkali

accumulation. A range of laboratory tests and procedure have been developed to characterize the coke

quality in order to assess their degradation potential in a blast furnace. The cold/hot strength of coke is often

estimated by laboratory tests conducted under a specified set of reaction conditions. The CRI (Coke

Reactivity Index) and the CSR (Coke Strength after Reaction) developed by Nippon Steel Corporation

(NSC), Japan is one of the most popular such test, which is the most widely recognized or seriously

considered test around the world1, 2. High CSR value of coke is believed to prevent coke from breaking

down, improve the permeability of gas and liquid and increase the productivity as well as decrease the

specific coke consumption3. Many empirical correlations based on ash chemistry and the CSR tests have

been developed in order to predict coke degradation behaviour in an operating blast furnace. Often individual

steelworks relies on their own empirical experience for the interpretation of CSR measurements and some

times modify the coke reactivity test conditions in accordance to their individual blast furnace operations.

Despite widespread popularity, there are well known concerns about the reproducibility of CSR

measurements as well as the validity of the predicted coke behaviour4. Alternatively, a detailed

comprehensive understanding of coke behaviour can be made by testing coke behavior in an experimental

5

blast furnace. Even though the EBF tests are time consuming, tedious and highly expensive, the information

generated is of great value in terms of their reliability suitability due to simulation of more realistic

conditions of a blast furnace process.

Coke degradation in a blast furnace occurs due to chemical, mechanical and thermal effects. At

higher flame temperature, cracking of coke is mainly attributed to thermal stress while at lower temperatures,

the degradation behaviour is influenced by coke reactivity which is dependent on other coke properties. The

coke reactivity can be influenced by its three major properties namely porosity, carbon structure and

constituent minerals. Coke pore structure is modified by growth and/or coalescence of pores, which is often

related to fluidity and swelling characteristics of parent coals 5, and modifies the available carbon surface

area for gas reactions. Coke displays graphitisation behaviour particularly at temperatures exceeding 1200oC,

which is influenced by the catalytic effect of minerals such as iron, and is believed to weaken the abrasion

resistance6. Iron in coke is also believed to catalyze gasification reactions7 which could have different

implications on coke behavior in the EBF. The influence of other minerals particularly those containing

alkali on the coke degradation is less certain. The main aim of this study is to investigate the effect of alkalis

on coke behaviour in the experimental blast furnace. Therefore, it is imperative to discuss further various

aspects of alkalis influence on coke behaviour.

In an operating blast furnace in a temperature range of 800-850°C6, alkali carbonate deposition is

limited to coke surface without affecting its size or strength or causing any additional stress in the bulk coke.

Alkalis are known to influence the solution loss reaction by initiating gasification reaction at considerably

lower temperatures (750-850°C) compared to classical gasification temperatures of 950°C, and also increase

the reaction rates as a function of potassium content 8, 9. The catalytic effect of alkalis and iron phases on

coke gasification was also reported in other studies10. Potassium content of coke varies at different locations

in a blast furnace e.g. alkalis are completely vaporised at raceway and their concentration becomes low in

hearth coke while alkali concentration is significantly high in the deadman coke such that potassium content

reaches up to 30% of total coke ash 8, 11, 12. Potassium is generally believed to weaken the coke strength at

high temperatures8. Concentration gradient of alkali in coke lump was reported to be responsible for stress

abrasion of alkali-rich layer of coke14. In other study, up to 5% variation of alkalis in coke did not indicate

any adverse impact on the coke strength (CSR)15.

6

Potassium adsorption by carbonaceous materials such as coke could vary in many ways depending

on the nature of the carbon phase as well as the treatment temperature13. Non-graphitic carbons such as coke

are believed to exhibit increased degree of potassium absorption with increasing temperature up to 1000°C,

and was related to modification of macro pore size and surface area due to activation13. Potassium species

can interact with coke by two mechanisms namely by diffusing/adsorbing in the carbon micropores or by

intercalating with carbon13. Potassium adsorption in coke can cause irregular swelling, increased stress,

modification of surface area, microstrength, and possibly size degradation13 without affecting the crystallite

height of coke carbon signficantly13.

Despite several studies indicating possible relationship between alkali content and coke strength,

there are uncertainties regarding their extent of impact and their role on the mechanisms of coke weakening6.

The influence of alkalis on coke gasification in a blast furnace also raises the concerns about the suitability

and interpretation of the conventional CSR test results 12 due to its inability to account for the effect of

recirculating alkalis. Therefore, there is a need to understand the influence of alkalis on modifications of

coke properties and its association with coke strength in an operating blast furnace. For this purpose a high

CSR coke was tested in an Experimental Blast Furnace (EBF) at Luleå, Sweden. The main aim of this study

is to study the effect of alkalis on the modification of physico-chemical properties of coke during its descent

in an experimental blast furnace and their association with coke strength as well as abrasion. Measurements

of carbon structure, porosity and alkali distribution of the centreline EBF coke samples are presented.

Evolution of coke strength and abrasion resistance of coke in the EBF was related to their carbon structure

and alkalis concentration.

2. Experimental

2.1 Experimental Blast Furnace Campaign

The current study is based on the coke samples from a dissection study of tenth campaign conducted

in 2002 in the Experimental Blast Furnace (EBF) situated in Luleå, Sweden. The fuel and coke rates were of

the order of 500 kg/thm and 350 kg/thm respectively while blast temperature was around 1200oC. The EBF

has a working volume of 8.2 m3 and a diameter of 1.2 meters at the tuyere and is 6 meter high from the stock

line to the tuyeres 16-18. The campaign was conducted to test a range of blast furnace materials including the

performance of a high CSR coke. This coke was specially prepared at SSAB Luleå coking plant by blending

of 67% low to medium volatile Australian coals, high volatile US coals and a small percentage of pet coke.

7

After completion of the campaign in 8 weeks, the EBF was quenched by purging nitrogen continuously for

ten days from the top. Top quenching also restricted the upward movement of heat flux and retarded any

subsequent reactions of the burden constituents16-18. The reacting gases were removed from the tuyeres in

short interval of less than a minute so that subsequent reactions ceased. Approximately 20-35 coke pieces

were collected from several vertical and radial locations of the EBF after quenching.

FIGURE 1

All coke layers were sampled at three radial positions namely centre line (C), middle (M) and close

to wall (R). The selection of layers was based on previously conducted laser measurements. Only centre line

cokes were selected for this study. Physical locations and codes of the EBF cokes are illustrated in Figure

1a. Coke samples from various locations represent different zones of the EBF. For example, sample KL10C

indicates a center line coke sample at 10th layer. Figure 1b illustrates the temperature of coke layers based on

separate thermal probe measurements. Based on thermal profile coke KL05C represents the stock line coke;

Samples KL10C, KL15C and KL20C represent the thermal reserve zone coke while samples KL25C and

KL30C represent the cohesive zone cokes. Based on physical conditions of layers during excavation, sample

KL35C is used to represent the extreme lower end of the cohesive zone. Chemical composition of the feed

coke is provided in Table I, while CSR (68.8) and CRI (23.15) values of the feed coke are shown in Table

II.

TABLE 1 & TABLE II

2.2 XRD of Cokes

Carbon structure of coke is often related to the reactivity as well as graphitization7, 19, and can be

measured by the X-ray diffraction. Graphitization degree of the EBF coke was measured in terms of the

crystallite height of coke carbon (Lc). Siemens 5000 X-ray diffractometer at the University of New South

Wales (UNSW), Australia was used to obtain the XRD patterns of cokes. Two small coke lumps

(approximately 6-8cm3) were selected from each layer and crushed to powder (< 75 micron). The XRD

patterns were obtained by recording the scattering intensities of coke powder by using Copper K radiation

(30 kV, 30 mA) as the X-ray source. Coke powder was packed into an aluminium holder and scanned over

an angular range from 5-105° using a step size of 0.05° and collecting the scattering intensity for 5 seconds

at each step. The carbon crystallite height (Lc) of EBF cokes was evaluated by analysing the 002 carbon

peak of the XRD pattern20 following classical Scherrer’s equation21.

8

Lc = 0.89 / B Cos

where is the wavelength of the X-ray radiation, B is the full width at half maximum intensity (FWHM) of

the 002 carbon peak and is the 002 carbon peak position. A sharper 002 peak represents a larger carbon

crystallite and a greater degree of ordering of the carbon structure of coke or graphitisation19. The carbon

content of EBF cokes was also measured by using carbon analyser (LECO) at the University of New South

Wales.

2.3 TGA Reactivity of Cokes with CO2

Non-isothermal reactivity was measured by using ~ 70 mg of coke powder in an Al2O3

crucible with the help of Netzsch STA 409 Thermal Gravimetric and Differential Thermal Analyser

(TGA/DTA) located at Luleå University of Technology. Figure 2 shows the schematic of the TGA/DTA. All

the EBF cokes were reacted under dynamic heating up to 1300°C with a heating rate of 10oC/minute and a

CO2 flow rate of 100 ml/min. A custom built TGA situated at the University of New South Wales, Australia

was used measure isothermal reactivity of cokes. The TGA furnace was made of recrystallised vertical

alumina (60 mm ID) tube in which the sample temperature was controlled by an internal thermocouple

located close to the sample holder. Approximately 0.2 g coke powder was placed on a square alumina

crucible (30X 30 mm) holder at room temperature, and the assembly was suspended with a high temperature

stainless wire to a balance to measure the weight loss with an accuracy of the order of 1 microgram

(Precisa® 1212 M SCS). The assembly was kept at a low temperature zone in the furnace followed by

heating up to 900oC at the rate of 2oC/minute while 5 l/minute of N2 was continuously purged through

Brooks 5850E mass flow controller. As the furnace reached 900oC, the reaction chamber was raised into hot

zone followed switching of gases to four l/min of N2 and one l/min of CO2. The weight loss of coke was

continuously recorded to data logger to calculate the carbon conversion.

FIGURE 2

2.4 SEM /EDS Analysis

Two sets of coke samples were examined by using Scanning Electron Microscope (Philips XL

30) equipped with Energy Dispersive X-ray Analysis (EDS). Coke pieces were mounted in an epoxy slow-

setting resin in plastic moulds (40 mm diameter). Surfaces were successfully ground on four different grades

of silicon carbide paper (120, 500, 800, and 1200 grit) with distilled water (without any impurity) and

9

polished with three different grades of paper with diamond paste of particle sizes of 15 μm, 9 μm, 3 μm, and

1 μm and lubrication fluid. The polished coke samples were mounted on aluminium support and coated with

a thin layer of gold-palladium alloy using a Bal-tec MCS 010 sputter coater. In each coke sample, the

elemental composition of mineral phases at several spots was also analysed with EDS analysis with

particular focus on the variation of potassium in the aluminosilicate phase and carbon matrix.

2.5 Light Optical Microscope (LOM) and BET N2 Adsorption

The coke porosity was measured by using Light Optical Microscope (Olympus BH-2-UMA)

equipped with motorized microscope stage (Marzhauser). Previously prepared polished coke stubs used for

SEM analysis, were also used for the optical analysis. The stubs were mounted in a special holder for

microscopic examination. The images of coke surface were acquired at two magnifications of 130x and 520x

in order to improve the reliability of data, and were processed by using a computer software (Leco 3001) to

obtain the macro and micro pores data. The BET surface area of EBF cokes was measured by using a

FlowSorb 2300 by adsorbing N2 at 77 K.

2.6 Coke Strength (CSR) and Abrasion (I-drum Test)

The CSR value of the feed cokes was measured after standard CRI test. THE CSR and CRI

values of feed coke are provided in Table II. The variation of coke strength in the EBF was estimated by

conducting the I-drum test on the quenched EBF coke. In order to distinguish the I-drum test values of the

EBF coke from the standard “CSR” test measurements, a new strength index namely Coke Fine Index (CFI)

was defined. On this basis, CFI values will represent the percentage of coke retained on 10 mm sieve after

tumbling for 600 revolutions for 30 minutes. The CFI values of the EBF reacted coke samples were obtained

by directly tumbling the cokes without further reacting with CO2 in order to assess the effect of EBF reaction

environment, temperatures and alkali exposure. The CFI values of the EBF cokes were much higher (~80s)

than the CSR value of feed coke (65) due to omission of the CRI step.

The percentage of coke passed through a 0.5 mm sieve after conducting I-tumbling test

directly on the EBF cokes was used to represent abrasion index. It should be noted that the values of this

abrasion index` might also vary from conventional abrasion index which is defined as the lack of resistance

to abrasion of the coke after reaction with carbon dioxide in a standard CRI test. The main objective of the

10

coke strength and abrasion measurements is to demonstrate the change in coke behaviour due to the EBF

reactions.

3. Results & Discussion

In the EBF, coke undergoes many changes such as carbon structure and constituent minerals including alkali

phases and porosity as discussed below. Each coke property could influence the coke behavior in an

operating blast furnace particularly its strength.

3.1 Evolution of Carbon Structure

Coke carbon reacts with upcoming CO2 as coke descends into lower part of the EBF.

Consequently, coke carbon content is changed at different locations in the EBF as illustrated in Figure 3.

Carbon content of lower zone coke (KL35C) is approximately 5% less than carbon content of the stock line

coke (KL05C). Lower carbon contents of cokes lower EBF zones can also be attributed to increased ash

content due to carbon loss as well as increased alkali uptake (Table III).

FIGURE 3

Figure 4a compares the XRD patterns of cokes from three representative locations of the EBF and

shows that width of 002 carbon peak of sample KL35C (cohesive zone coke) is sharper compared to the peak

width of the sample KL05C (stock line coke) and KL15C (upper zone coke) of the EBF. The background

intensity of sample KL35C (lower zone coke) is less than those of upper zone cokes. Low background

intensity is indicative of less proportion of amorphous carbon19. The amorphous carbon content of coke was

found to decline sluggishly up to cohesive zone, and changed rapidly as the coke descended below the

cohesive zone of the EBF. The results indicate that carbon structure of coke becomes increasingly ordered as

the coke moves into the lower part of the EBF while the amorphous carbon is increasingly depleted.

FIGURE 4

Figure 4b shows that crystallite height (Lc) of coke carbon increases exponentially with the EBF

height from stock line. Figure 5 shows that the Lc values of the EBF coke increases linearly with EBF

temperatures. It may be noted that, normally, Lc values are not expected to increase significantly up 1200oC

11

due to similar range of thermal environments already experienced by coke in the coke oven. In this

particularly case, it may be related to either low temperature of coking process or variation associated with

estimated temperatures of the EBF in this temperature range (< 1200oC). Many factors such as alkali and

iron species could influence the changes in carbon structure of coke. However, alkalis are not expected to

have a significant effect on the improvement of Lc values with increasing temperature13. A linear correlation

in Figure 5 clearly demonstrates the strong influence of temperature on the Lc values of the coke compared

to impact of increased alkali concentration of cokes in the EBF.

FIGURE 5

3.2 Evolution of Coke Ash Chemistry

In addition to carbon structure, ash chemistry of the EBF coke is also continuously changed.

The chemical analysis of the EBF cokes is provided in Table III. It may be noted that alkalis might not

necessarily occur as oxides as expressed in Table III, which is a common practice of indicating the elemental

concentration of carbonaceous materials. Figure 6 illustrates the variation of alkali content of the coke ash

against the EBF height. Figure 6 shows that K2O and Na2O of coke increases as the coke moves into the

cohesive zone of the EBF. For example, K2O of KL35C coke was approximately 20 times of that of upper

zone coke sample KL05C while the same cokes displayed approximately 10 times increase in Na2O content.

Our observation of remarkable pickup of alkalis by the cokes is in agreement with previous studies11-13.

Alkali could influence the surface area, chemical structure of coke by interacting with coke in two ways

either assimilating in the pores or intercalate with coke carbon as discussed before13. In order to further

understand the mode of adsorbed alkali in coke, three layers of coke samples from two locations at the

opposite ends of the EBF were examined in detail by SEM/EDS.

TABLE III

FIGURE 6

Alkalis in coke are generally associated with aluminosilicate phases due to their origin from

potassium containing clays of parent coals. Figure 7 shows the SEM images of the upper zone coke KL10C

in which the alkali distribution is considered to be similar to that of the feed coke. Figures 7a, 7c and 7e

shows the features of outer, middle and inner core of the coke while Figures 7b, 7d and 7f provide the

magnified views of the selected portion of the same parts of coke KL10C. Table IV provides the EDS

12

analyses of many locations inside the coke as shown by crosses and numbers in Figures 7 and 8. Careful

examination of alkali content of mineral phases of coke KL10C indicated that their alkali range was less than

4%, which is typical to that of potassium aluminosilicate often seen in the feed cokes (Figure 7b, 7d, 7f &

Table IV). The EDS analyses further suggested that range of alkali distribution of coke was not significantly

altered in different sections of the coke KL10C, which could be attributed to less influence of recirculating

alkalis due to less adsorption of alkalis at lower temperatures13.

FIGURE 7

TABLE IV

Figure 8 illustrates a similar kind of the alkali distribution in different parts of the lower zone coke

KL35C. The range of alkali composition of the aluminosilicate of different sections of the coke KL35C was

significantly higher being 10 to 15% (Fig. 8b, 8d, 8f & Table IV) particularly compared to similar phases in

sample KL10C. Similarly to coke KL10C, the range of alkali concentration of aluminosilicate in three

sections of the coke KL35C was of the same order. On the other hand, the range of alkali concentration of

carbon matrix of the coke KL35C appears to have increased significantly compared to that of coke KL10C

(Analysis Table IV with C > 80%). It appears that recirculating alkalis vapours might have been trapped by

aluminosilicate of coke during penetration from external surface into the inner core of coke. At the same

time, alkalis originally present in the aluminosilicate of coke might have been released and diffused

uniformly into the bulk of the coke matter. Therefore, at each location of the EBF, alkali reactions appeared

to have occurred throughout the carbon matrix of coke rather than preferentially accumulating at the outer

layer. However, due to complexity of the heterogenity aluminisilicate composition and distribution and

limited scope of the present study, further investigations are necessary to conclusively confirm the suggested

alkali behavior in coke.

Visual inspection of SEM images of KL10C and KL35C did not display any apparent crack or

significant differences in their macro pores. The chemical composition of the abraded portion of the

externals layers of cokes was found to be similar to alkali content of the bulk coke, which implied that alkalis

were as not preferentially accumulated in the external layers. Therefore, it is reasonable to infer that abrasion

behaviour of the coke in this study might not necessarily be related to the increased alkali absorption by coke

in the different EBF locations.

FIGURE 8

13

3.3 Evolution of Coke Reactivity in the EBF

Figure 9 compares the non-isothermal reactivity of the EBF coke and shows that weight loss of lower

zone cokes was consistently greater compared to the weight loss of upper zone cokes. This implies that

reactivity of coke increases as it descends into lower parts of the EBF. In order to further confirm the effect

of furnace reactions on coke reactivity, isothermal reactivity of two representative cokes was also measured

at 900oC as illustrated in Figure 10. The isothermal reactivity of the cohesive zone coke (KL35C) was found

to be higher when compared to the reactivity of KL05C coke (stock line coke). The coke reactivity is often

related to the carbon structure, surface area and coke minerals. In previous section it was shown that carbon

structure of lower zone cokes was more ordered. Increased reactivity of lower zone cokes can not be

attributed to the carbon structure alone as increased crystalline order of carbon often is often believed to

influence the reactivity adversely19.

FIGURE 9

FIGURE 10

Figure 11 shows that surface area of coke carbon initially increases as the coke moves towards the

lower parts of the EBF, however, it starts decreasing as it approaches to the cohesive zone.

FIGURE 11

Due to non-uniform growth of surface area of the EBF coke, the consistent increased reactivity of the

coke samples can not be fully attributed to carbon surfcae area. Higher carbon surface is often related to

higher anisotropic carbon conent of the sample22. This implies that the isotropic part of the EBF cokes was

preferentially consumed due to its higher reactive nature, particulraly in the shaft region. In cohesive zone,

the reactivity of anisotropic carbon of coke can be related to catalytic effect of alklais, as the alkalis content

of EBF cokes increased consistently with increasing temperatures of the EBF following the same trend as

displayed of the reactivity measurements of the EBF coke. It may be noted that at increased temperature

greater proportion of alkalis could release from aluminosilicate slag formed inside coke. It is interesting to

note that reactivity of the EBF coke could increase due to catalytic influence of alkalis even when the carbon

structure is more ordered. This implies that alkalis are playing a strong role in accelerating the coke

reactivity under the EBF reaction environment. These results further suggests that alkalis could increase the

reactivity without modifying the crystalline order of coke carbon.

14

3.4 Effect of Graphitization, Alkalization and Reactivity on Coke Strength

Figures 12 illustrates that the coke strength (CSR) decreases as coke moves towards lower parts of

the EBF. It may be noted that the CSR values of the EBF reacted cokes are higher than CSR value of feed

coke. This is because during the CSR measurement of the EBF samples, cokes were not reacted with CO2

prior to tumbling step of the standard CSR test. The CRI component of the CSR test was omitted mainly to

evaluate the impact of EBF reactions on the coke strength. Figure 12 further shows that abrasion index of

lower zone cokes are higher compared to upper zone cokes. This suggests that erosion tendency of coke also

increases in the EBF at higher temperatures.

FIGURE 12

Coke strength could depend on many factors including porosity and carbon structure23-24. Figure 13

compares the optical images of KLC05 and KLC35 cokes as measured by Light Optical Microscope (LOM).

Coke KLC35 displays significantly large number of bright phases (Figure 13b) when compared to similar

phases in the coke KLC05 (Figure 13a). Greater proportion of bright phases in the cohesive zone coke could

be attributed to increased degree of oxidation of isotropic carbon as the coke moved towards lower part of

the EBF. The LOM images were processed to estimate the pore size data. Figure 14 shows that the

percentage of macro or micro pores is not significantly decreased during the coke descent into the EBF.

Despite known limitation of porosity assessments from optical image analysis, it can be clearly inferred that

the porosity of coke did not increase during coke reaction inside the EBF. Consequently, it is reasonable to

conclude that any subsequent reduction in coke strength might not be related to modification of coke

porosity. It may be noted that total alkali load and the actual duration of contact period of coke with alkalis in

a full-scale blast furnace might be different from that experienced by coke in the EBF, and hence may not

necessarily have the same impact on coke strength.

FIGURE 13

FIGURE 14

Figure 15 shows the relationship between coke strength indicator i.e. coke fine index (CFI) and coke

abrasion index with Lc values of coke at different locations inside the EBF. Figure 15a shows that the CFI

decreases linearly with increasing Lc values i.e. increasing crystalline order of coke carbon. Therefore, it can

be concluded that coke graphitisation has a strong influence on coke weakening inside the EBF. Figure 15b

shows that abrasion index does not change until the Lc values exceed more than 30 Angstrom. This means

that the relationship between carbon structure and abrasion is more apparent at higher temperatures of the

15

EBF as the Lc value of coke exceeds more than 30 Angstrom in a temperature range of 1400oC (Figure 5).

This implies that coke might not abrade until it goes through certain minimum degree of graphitisation.

Figure 16 shows the relationship between coke strength indicator namely coke fine index (CFI) and coke

abrasion index with potassium content of coke at different locations inside the EBF. Figure 16a shows that

initially coke strength i.e. CFI decreases with potassium content of coke. However, potassium content of

coke can not be clearly related to reduction in coke strength once it exceeds 20% i.e. which occurs in the

lower parts of the EBF. The abrasion index is not influenced by the potassium content of coke until it

exceeds more than 20% as shown in Figure 16b. On the basis of results in Figures 15 and 16, one can

conclude that carbon structure of coke provides a consistent correlation with coke deterioration in the EBF

such that the increased graphitisation accelerates coke weakening as well as abrasion. On the other hand, the

reactivity of coke was consistently increased as it descended into the EBF, however it did not display any

evidence of cracks/fissures or change is porosity. This implies that coke reactivity might not have a

significant impact on the coke weakening in the EBF. The EBF observations are consistent with previous

studies in which graphitisation was linked to coke degradation24-25.

The study suggests that coke alkalis catalyse the reactivity but might not have a strong effect on coke

graphitisation and hence on coke strength. The study has significant implications for understanding the effect

of coke properties on its performance in a blast furnace particularly in understanding the influence of coke

minerals on coke performance in an operating blast furnace. While iron species are well known to

simultaneously catalyse coke reactivity and coke graphitisation24, this study implies that alkalis could

catalyse coke reactivity without graphitising. However, further studies are required to establish the

predominant factors affecting coke weakening in an operating blast furnace. The strong effect of coke

graphitisation on fine generation needs to be validated by blast furnace operating experiences based on wide

variety of cokes.

FIGURE 15

FIGURE 16

4. Conclusions

A high CSR coke was tested in an experimental blast furnace. Physical and chemical properties of cokes

samples from the EBF were measured. Evolution of coke properties particularly carbon structure and alkali

16

uptake were related to CO2 reactivity as well as coke behaviour (e.g. CSR/abrasion). On the basis of this

study, following conclusions were made.

1. The order of carbon structure and concentration of alkali species were increased, and were found to

be the most notable changes in the coke properties as it passed through thermal reserve zone to the

cohesive zone of the EBF.

2. The height of the carbon crystallite (Lc) of coke was increased while amorphous carbon content was

decreased in the hotter zones of the EBF. A linear correlation between the Lc values and the coke

bed temperature was established to demonstrate the strong effect of temperature on the carbon

crystallite (Lc) of coke in the EBF.

3. The alkali concentration of coke was increased with increasing temperature of the coke bed such that

most of the alkalis were evenly distributed in the bulk coke rather than in the external coke layer.

4. The CO2 reactivity of coke was found to increase during progressive movement of the coke from the

thermal reserve zone to cohesive zone of the EBF, and was related to the catalytic effect of increased

alkali concentration in coke. The results imply that alkalis could catalyse the coke reactivity without

having any strong effect on graphitisation or possibly any adverse effect on fines generation.

5. The deterioration of coke quality particularly coke strength and abrasion propensity are related to the

coke graphitisation, alkalization and reactivity. The coke graphitisation is shown to have a strong

effect on the coke degradation behaviour.

Acknowledgements

The authors would like to thank MEFOS Metallurgical Research Institute AB and LKAB for providing the

samples and the opportunity to conduct this research, Jernkontoret and Swedish Energy Agency for financial

support. A part of this work was undertaken as part of the Cooperative Research Centre for Coal in

Sustainable Development (CCSD) Research Program 5.1 (Ironmaking). Authors would also like to

appreciate help provided by Mr N M Saha-Chaudury and other staff at the University of New South Wales

for their assistance in using equipment and facilities.

17

References

1) Coke - Determination of Coke Reactivity Index (CRI) and Coke Strength after Reaction (CSR):

ISO/DIS 18894, (2001).

2) Annual Book of ASTM Standards, Section 5: Petroleum Products, Lubricants and Fossil Fuels, vol.

05.05 (1996).

3) Grosspietsch, K.H., H.B. Lungen, G. Dauwels, Ferstl, T. Karjalahti, P. Negro, B. van der Velden and

R. Willmers: in Proc. 4th European Coke and Ironmaking Congress, Paris, 1, (2000), 1-1.

4) Arendt, P., F. Huhn and H. Kühl: Cokemaking International, 2, (2001), 50.

5) Sato, H., J. W. Patrick and A. Walker: Fuel, 77, (1998), 1203.

6) Willmers, R. R. and C. R. Bennington: 2nd International Cokemaking Congress, London, UK (1992),

260.

7) Feng, B., S. K. Bhatia and J. C. Barry: Carbon, 40, (2002), 481.

8) Helleisen, M., R. Nicolle, J. M. Steiler, N. Jusseau, C. Meltzheim and C. Thiriom: 1st International

Cokemaking Congress, Essen, Germany, (1987), C2.1.

9) Forsberg, S: 1st International Cokemaking Congress, Essen, Germany (1987), C6.1.

10) van der Velden, B: McMaster Cokemaking Course, Hamilton, (2003), 22.1.

11) Steiler, J. M., R. Nicolle, P. Negro, M. Helleisen, N. Jusseau, B. Metz and C. Thirion: Ironmaking

Conference Proceedings, ISS, Washington DC, (1991), 715.

12) Tucker, J. and J. Goleczka: 1st International Cokemaking Congress, Essen, (1987), C5.1.

13) Chan, B. K. C., K. M. Thomas and H. Marsh: Carbon, 31, (1993), 1071.

14) Beppler, E., B. Gerstenberg, U. Jahnsen and M. Peters: Cokemaking International, 2, (1994), 15.

15) Gudenau, H. W: First International Congress of Science and Technology of Ironmaking, Sendai,

Japan, (1994), 348.

16) Dahlstedt, A., M. Hallin and M. Tottie: SCANMET, Luleå, Sweden, (1999), 235.

17) Dahlstedt, A., M. Hallin and J.-O. Wikström: in Proc. 4th European Coke and Ironmaking Congress,

1, (2000), 138.

18

18) Hooey, L., J. Sterneland and M. Hallin: in 60th Ironmaking Conference Proceedings, Baltimore, ISS,

USA, (2001), 197.

19) Lu, L., V. Sahajwalla, C. Kong and D. Harris: Carbon, 39 (2001), 1821.

20) Sahajwalla, V., M. Dubikova and R. Khanna: in Proc. Tenth International Ferroalloys Congress, Cape

Town, South Africa, (2004), 35

21) B.D. Cullity: Elements of X-ray Diffraction (1978), Addison-Wesley Publishing, USA.

22) Kerkkonen, O., P. Arendt and H. Kuhl: 61st Ironmaking Conference Proceedings, Nashville, ISS,

USA, (2002), 393.

23) Vandezande, J.A: 44thIronmaking Conference Proceedings, ISS, USA, (1985), 189.

24) Dubrawaski, J.V. and W.W. Gill: Ironmaking and Steelmaking, 1, (1984), 7

25) Y. Kashiwaya, M. Takahata, K. Ishii, K. Yamaguchi, M. Naito, H. Hasegawa: Tetsu-to-Hagane,

Journal of the Iron & Steel Institute of Japan, 87(5), (2001), 259.

19

List of Tables

Table I Chemical composition of the feed coke used in the current study. .......................................20 Table II CSR and CRI values of the feed coke tested in the Experimental Blast Furnace (EBF). ....21 Table III Chemical composition of the EBF coke samples. .............................................................22 Table IV The EDS analyses of different spots of the coke samples KL10C and KL35C illustrated in

Figure 7 and Figure 8 respectively. Analysis spots including more than 80% C are considered as carbon matrix of coke while rest of the points are aluminosilicate phases. ..........................23

20

Table I Chemical composition of the feed coke used in the current study.

Moisture Volatile matter Ash content S N C H 0.12 1.0 11.4 0.55 1.13 87.51 0.16

21

Table II CSR and CRI values of the feed coke tested in the Experimental Blast Furnace (EBF).

CSR 68.8 CRI 23.15

22

Table III Chemical composition of the EBF coke samples.

*XRF analysis was obtained from SSAB laboratories.

Inorganic components in coke indicated as oxides (wt%) Sample code Ashcontent SiO2 Al2O3 Fe2O3 CaO MgO K2O Na2O TiO2 P2O5 SO3

KL05C 11.59 5.76 2.63 1.20 0.01 0.06 0.17 0.10 0.16 0.03 1.47

KL10C 12.53 6.39 2.77 1.32 0.04 0.06 0.35 0.16 0.17 0.03 1.25

KL15C 13.26 6.30 2.65 1.06 0.00 0.06 1.24 0.43 0.15 0.02 1.35

KL20C 12.98 5.58 2.57 0.92 0.02 0.07 1.78 0.61 0.14 0.02 1.27

KL25C 13.77 5.83 2.61 0.92 0.02 0.08 2.31 0.67 0.14 0.02 1.17

KL30C 13.89 5.97 2.66 0.97 0.04 0.08 2.07 0.68 0.14 0.02 1.25

KL35C 14.80 5.81 2.64 0.77 0.00 0.08 3.21 0.85 0.12 0.22 1.10

23

Table IV The EDS analyses of different spots of the coke samples KL10C and KL35C illustrated in Figure 7 and Figure 8 respectively. Analysis spots including more than 80% C are considered as carbon matrix of coke while rest of the points are aluminosilicate phases.

Elemental analysis (wt.%) Code C O Si Al K Na

Fig. 7b – external layer of coke KL10C 1 21.7 24.9 25.9 12.2 1.8 6.12 30.2 15.1 11.4 6.2 0.8 7.53 39.7 26.6 29.5 0.9 0.4 0.34 19.6 33.0 40.3 3.8 0.3 0.75 33.8 21.3 18.0 14.3 2.3 3.76 26.3 27.9 33.8 5.9 0.8 1.97 83.4 2.3 3.0 1.5 0.3 1.18 37.2 27.8 30.2 0.8 0.2 0.49 16.5 32.7 38.2 6.9 0.6 1.5

Fig. 7d – middle layer of coke KL10C 1 97.7 NA 0.3 0.3 0.6 1.12 97.2 NA 0.5 0.4 0.5 1.53 24.4 NA 58.1 14.8 1.5 1.24 94.8 NA 1.6 1.5 0.3 1.85 90.5 NA 5.0 2.8 0.3 1.56 97.6 NA 0.3 0.4 0.4 1.4

Fig. 7f – inner core of coke KL10C 1 32.8 NA 65.9 0.9 0.2 0.22 41.5 NA 57.3 0.8 0.3 0.13 91.5 NA 7.5 0.5 0.1 0.4

Fig 8b – external layer of coke KL35C1 2.6 31.4 25.1 24.7 10.1 2.92 3.3 39.6 55.2 0.6 0.1 0.13 2.5 30.1 25.0 24.8 10.8 3.54 86.8 3.0 0.4 0.5 5.6 1.05 2.1 30.0 26.6 24.5 10.4 3.16 84.8 3.3 0.6 0.6 5.8 0.9

Fig 8d - middle layer of coke KL35C 1 12.4 27.7 25.1 20.3 8.6 3.72 10.5 29.5 23.8 22.1 8.8 3.63 45.6 16.8 4.7 5.0 14.9 1.34 24.2 20.1 20.5 17.4 9.7 3.45 27.5 25.5 16.2 14.7 9.5 2.96 10.2 28.5 24.9 22.6 8.4 3.5

Fig 8f – inner core of coke KL35C 1 4.5 29.9 26.7 22.8 8.5 4.32 9.7 29.0 24.6 23.2 6.1 3.63 85.7 2.1 1.7 1.3 2.8 1.24 86.8 3.1 0.5 0.2 2.6 1.0

24

5 89.1 0.0 0.9 0.8 2.4 0.96 4.9 29.9 21.3 30.7 5.2 3.97 87.4 1.7 0.6 0.3 2.1 1.58 4.9 30.0 23.9 28.2 6.3 3.69 5.2 29.8 25.6 23.9 8.0 3.7

List of Figures

Figure 1 a) Schematic of the EBF illustrating locations of the excavated coke samples; b) The temperature profiles of the EBF estimated from temperature probes. Coke sample codes are also indicated..............................................................................................................................25

Figure 2 a) Schematic of the TGA/DTA-MS reactor used for the measurement of non-isothermal reactivity of cokes; b) TGA reactor used for the measurement of isothermal reactivity...........26

Figure 3 Carbon contents of the cokes and the temperature profile of the EBF are plotted against the EBF height from top on primary and secondary Y axis respectively. .................................27

Figure 4 a) Comparison of XRD patterns of the EBF cokes from three locations, b) Carbon crystallite dimension (Lc) of coke plotted against the EBF height from top. The tentative EBF temperatures are plotted on the secondary Y-axis. ....................................................................28

Figure 5 Correlation between the crystallite height of coke carbon (Lc) and the estimated coke temperatures based on the temperature profile of the EBF estimated from probe measurements.....................................................................................................................................................29

Figure 6 Alkali concentrations of the EBF coke ash plotted against the EBF height from top. Tentative temperature of the corresponding EBF bed layers is also indicated. .........................30

Figure 7 a) SEM image of external layer of the upper zone coke KLC10, b) magnified view of a selected region of external layer; c) and d) similar images of middle layer of the same coke; and e) and f) illustrate the coke matrix of the inner core. Crosses indicate the location of EDS analysis.......................................................................................................................................31

Figure 8 a) SEM image of external layer of the cohesive zone EBF coke KL35C, b) magnified view of a selected region of external layer; c) and d) similar images of middle layer of the same coke; and e) and f) illustrate the inner core of coke matrix. Crosses indicate the location of EDS analysis..........................................................................................................................32

Figure 9 Loss in weight of the EBF coke samples with increasing reaction temperature of TGA/DTA. .................................................................................................................................33

Figure 10 Comparison of carbon conversion of coke samples KL05C and KL35C representing two different zones of the EBF with time during CO2 reactions at 900oC in a TGA furnace. .........34

Figure 11 Variation of surface area (BET N2 adsorption data) of the EBF coke samples with height of the EBF from top. ..................................................................................................................35

Figure 12 Variation of coke strength indicated by CFI (Coke Fine Index) based on I-drum test and abrasion indices of coke with the EBF height from top.............................................................36

Figure 13 a) Optical images of the upper zone coke sample KL05C; b) Lower zone coke sample KL35C. The dark areas in figure b indicate the regions of oxidation........................................37

Figure 14 Variation of porosity (based on LOM measurements) of coke with the EBF height. .......38 Figure 15 a) Coke strength indicator (CFI) of the EBF coke samples plotted against coke carbon

structure (Lc), b) abrasion index of the same cokes plotted against Lc values..........................39 Figure 16 ) Coke strength indicator (CFI) of the EBF coke samples plotted against potassium

content of their ash, b) abrasion index of the same cokes plotted against their potassium content........................................................................................................................................40

25

Figure 1 a) Schematic of the EBF illustrating locations of the excavated coke samples; b) The temperature profiles of the EBF estimated from temperature probes. Coke sample codes are also indicated.

26

Figure 2 a) Schematic of the TGA/DTA-MS reactor used for the measurement of non-isothermal reactivity of cokes; b) TGA reactor used for the measurement of isothermal reactivity.

27

Figure 3 Carbon contents of the cokes and the temperature profile of the EBF are plotted against the EBF height from top on primary and secondary Y axis respectively.

28

Figure 4 a) Comparison of XRD patterns of the EBF cokes from three locations, b) Carbon crystallite dimension (Lc) of coke plotted against the EBF height from top. The tentative EBF temperatures are plotted on the secondary Y-axis.

29

Figure 5 Correlation between the crystallite height of coke carbon (Lc) and the estimated coke temperatures based on the temperature profile of the EBF estimated from probe measurements.

30

Figure 6 Alkali concentrations of the EBF coke ash plotted against the EBF height from top. Tentative temperature of the corresponding EBF bed layers is also indicated.

31

Figure 7 a) SEM image of external layer of the upper zone coke KLC10, b) magnified view of a selected region of external layer; c) and d) similar images of middle layer of the same coke; and e) and f) illustrate the coke matrix of the inner core. Crosses indicate the location of EDS analysis.

32

Figure 8 a) SEM image of external layer of the cohesive zone EBF coke KL35C, b) magnified view of a selected region of external layer; c) and d) similar images of middle layer of the same coke; and e) and f) illustrate the inner core of coke matrix. Crosses indicate the location of EDS analysis.

33

Figure 9 Loss in weight of the EBF coke samples with increasing reaction temperature of TGA/DTA.

34

Figure 10 Comparison of carbon conversion of coke samples KL05C and KL35C representing two different zones of the EBF with time during CO2 reactions at 900oC in a TGA furnace.

35

Figure 11 Variation of surface area (BET N2 adsorption data) of the EBF coke samples with height of the EBF from top.

36

Figure 12 Variation of coke strength indicated by CFI (Coke Fine Index) based on I-drum test and abrasion indices of coke with the EBF height from top.

37

Figure 13 a) Optical images of the upper zone coke sample KL05C; b) Lower zone coke sample KL35C. The dark areas in figure b indicate the regions of oxidation.

38

Figure 14 Variation of porosity (based on LOM measurements) of coke with the EBF height.

39

Figure 15 a) Coke strength indicator (CFI) of the EBF coke samples plotted against coke carbon structure (Lc), b) abrasion index of the same cokes plotted against Lc values.

40

Figure 16 ) Coke strength indicator (CFI) of the EBF coke samples plotted against potassium content of their ash, b) abrasion index of the same cokes plotted against their potassium content.

Paper IV

Investigation of coke properties while descending through an experimental blast furnace

1Tobias HILDING, 1Jan-Olov WIKSTRÖM, 2Urban JANHSEN, Olavi KERKKONEN3

1Mefos Metallurgical Research Institute AB, Box 812, 971 25 Luleå, Sweden 2ThyssenKrupp Stahl AG, Kaiser-Wilhelm-Strasse 100, 47166 Duisburg, Germany

3Rautaruukki Oyj, Ruukki Production, Rautaruukintie 155, P.O. Box 93, 92101 Raahe, Finland

Abstract

A high and a low CRI coke were tested in the LKAB’s Experimental Blast Furnace (EBF) at Luleå. Frequent probe sampling from two shaft probes and an inclined probe going through the cohesive zone area was done. In addition to this, one tuyere core drilling took place for each coke quality. Process differences were recorded and analysed. The evolution of physical and chemical properties of the sampled coke were analysed by XRD. The inorganic phases of the EBF cokes was analysed by XRF. The CO2 reactivity of the coke samples was measured under non-isothermal conditions (1300oC) using Thermo Gravimetric Analysis (TGA). The anisotropic/isotropic relation of the coke carbon texture was measured by a light optical microscope. The tuyere core sample was divided into four segments and analysed, with use of above mentioned methods, as well as porosity studies by light optical microscope.

The order of carbon structure and the concentration of alkali species were increased as coke passed through thermal reserve zone to the cohesive zone of the EBF. The alkali concentration of the EBF coke was increased with increasing depth until the coke reached tuyere level. There was a higher alkali uptake for the high CRI coke. The CO2 reactivity of coke was found to increase with increasing depth of the EBF from top, and was related to the catalytic effect by increased alkalis in coke. The TGA test indicated only small differences between high quality and poor quality coke. Starting from different levels of anisotropy of the feed cokes, the isotropic carbon texture of both coke types increased during its descent in the EBF. The main result of these measurements is that the isotropic coke carbon components are more metallurgical treatment resistant under load of the EBF.

KEY WORDS: CSR, CRI, Alkali, Graphitisation, Gasification, TGA reactivity, XRD, coke carbon texture, porosity

1. Introduction

Blast Furnace (BF) is the most dominant process of iron making worldwide. In addition to ferrous burden coke is the raw material charged into the BF which has a strong influence on the process efficiency and hot metal quality. Both economic and environmental pressures demand a reduction in coke consumption in blast furnace process. At present there is an increasing focus on replacing coke by injecting cheaper coal or other reductants through tuyeres. At low coke rate operation, less amount of coke is available in the BF for melting of ferrous burden, providing gases for iron oxide reduction and to maintain the gas/liquid permeability. Therefore, the optimised coke quality is becoming increasingly critical for the efficient operation of a blast furnace.

A range of quality tests have been developed to characterize degradation potential of coke in a blast furnace. Today the Coke Reactivity Index (CRI) and the Coke Strength after Reaction (CSR), developed by Nippon Steel Corporation (NSC), Japan is the most widespread test to assess the effect of solution loss reaction on coke strength. The CRI/CSR test is adopted by ASTM and is also being considered for ISO standard1, 2. High CSR coke is believed to prevent the coke from breaking down, improve the permeability of gas and liquid and increase the productivity as well as decrease the specific coke consumption3. During the test, 200 g of coke crushed to 19-22.4 mm, reacts with CO2 gas at 1100 oC during 2 hours. CRI is defined as the mass loss after gasification reaction. CSR is measured in “CSR drum” rotating the reacted coke at 20 revolutions per minute during 30 min. CSR is defined as the weight percentage of coke over 10 mm after the rotation. The main factors affecting negatively coke CSR are obtained to be an increase in carbon texture isotropy and ash basicity.

As the coke moves towards lower zones of a blast furnace, it degrades and generates fines, which in each separate case weakens BF permeability and process efficiency. Blast furnace tuyere core drillings during the monthly stoppages of industrial blast furnaces have given new information about coke degradation at higher temperatures than 1100 °C. At tuyere level there is a pronounced influence of thermal impact on coke5. Reactions combined with breakdown of silicate minerals generate new micro-void in carbon textures and is believed to weaken coke resistance. With an increase in temperature carbon textures begin to decompose by graphitization. Speed of texture graphitization is in turn catalysed by molten iron and also slag accretions6.

An exact alkali role on coke degradation has far remained unclear. Alkalis sublimate into coke without affecting its size or strength in an operating blast furnace at temperature range of 800-850°C6. Alkali impregnation is known to speed-up the solution loss reaction such that the CRI values for coke have significantly increased depending on free potassium content and coke properties8, 9.

The potassium content of coke varies at different locations of an operating blast furnace, e.g. at tuyere level alkalis are completely vaporised at high temperatures of the raceway but deadman coke ash in the BF centre could contain up 30 % of potassium oxide 8, 11, 12. First of all alkalis are fixed in aluminosilicates13.

The free potassium adsorption is believed to be a function of carbon textures in coke as well as temperature14. For example, potassium penetration into isotropic textures, compared to high anisotropic domains, is shown to be greater with temperatures 1000 -1200 °C and is related to macropore size and surface area14. Generally, alkalis are believed to cause breakage of coke by the selective reactions with different ordered carbon forms.

The main focus of this investigation is to study different coke qualities in an experimental blast furnace. The measurements of texture and porosity distributions with the ash chemistry of the EBF coke samples from several locations are presented. Changes in low and high CSR cokes formed during passage in the EBF are compared with each others. The differences in the process were also studied.

2. Experimental

2.1 Experimental Blast Furnace

The EBF has a working volume of 8.2 m3 and a hearth diameter of 1.2 m and is 6 m high from stock line to tuyere. It has been described in details elsewhere15-17.

2.2 EBF trial with poor quality coke

The trial took place in the spring of 2004 and consisted of two parts i.e. a two days reference period with a low CRI coke followed two days of operation with high CRI coke.

Figure 1 Illustration of the Experimental Blast Furnace and the approximate positions of the probes

Solid sampling occurred at three positions, i.e. upper shaft, lower shaft and through the cohesive zone, see Figure 1. In addition a tuyere core drilling was done.

The sampled coke material was separated from slag fluxes and pellets. Prior to x-ray diffraction, XRF and TGA reactivity measurements, small coke lumps (approximately 6-8 cm3) were selected from each probe and crushed to powder (< 75 micron).

The tuyere drill core was divided in four equally large segments and labelled Centre, Mid 1, Mid 2, and Wall. Thereafter the samples were sieved to fractions of -19 mm, 19-22.4 mm and +22.4 mm. The samples labelled Centre thus represent coke from the centre of the furnace at tuyerelevel.

The process parameters were altered as little as possible when reductant changed. Same amountof coal injection was used. The cokes were very different in quality, as can be seen in Table I.

Table I: Properties of the feed coke used in the current study

Parameters Low CRI coke High CRI coke CRI 19 48CSR 72 35Fe 0.35 1.05SiO2 6.14 4.72P2O5 0.022 0.053Al2O3 2.82 2.26MgO 0.04 0.2Na2O 0.04 0.11K2O 0.14 0.22TiO2 0.18 0.1

2.3 XRD of Cokes

A Siemens 5000 X-ray diffractometer was used to measure the carbon structural parameters including degree of coke graphitisation. XRD patterns were obtained by recording the scattering intensities by using Copper K radiation (30 kV, 30 mA) as the X-ray source. Coke powder was packed into an aluminium holder and scanned over an angular range from 5-105° using a step size of 0.05° and collecting the scattering intensity for 5 seconds at each step. Carbon structure of coke is often related to the reactivity as well as graphitization7, 18. Graphite structure can be described as regular, vertical stacking of hexagonal aromatic layers. Graphitization degree of the EBF cokes was characterized by the crystallite height of coke carbon (Lc), and was evaluated by analysing the 002 carbon peak of the XRD pattern19. The XRD data was processed to obtain the crystallite height (Lc) of coke carbon. Average stack height (Lc) of the 002 carbon peak was calculated using Scherrer’s equation by using K = 0.89.

A sharper 002 peak indicates a larger carbon crystallite and a greater degree of ordering of the carbon structure of coke or graphitisation18.

2.4 TGA Reactivity of Cokes with CO2

Non-isothermal reactivity was measured by using 60~80 mg of coke powder in an Al2O3 crucible with the help of Netzsch STA 409 Thermal Gravimetric and Differential Thermal Analyser (TGA/DTA). All the EBF cokes were reacted under dynamic heating up to 1300°C with a heating rate of 10oC/minute and a CO2 flow rate of 100 ml/min.

2.5 Micro texture measurement

The change in the coke microstructure passing the EBF was measured by an automated microscopic measuring procedure developed at TKS to quantify the ordering of the coke carbon microstructure. This measuring procedure is based on the optical physics of the bi-reflectance. The dimension of the bi-reflectance is recorded using a linear polarising filter in the reflected light of the sample at various polarisation degrees.

The microscope employed is equipped with a scanning stage, an auto focus system and a power-driven polarizer in the reflected microscopic light. An adapted image analysing system enables quantification of quantify the degree of anisotropic and isotropic components calculated from the optical bi-reflectance.

Using this method at both coke operations (high and low level CRI) the feed cokes were investigated in comparison to the tuyere coke material sampled by the tuyere probe. The material of each tuyere core drilling was split radially into four segments. The material of each segment was screened into three fractions (<19 mm, 19-22,4 mm and > 22,4 mm) and than separated into coke, metal and slag components. From the crushed feed coke and each tuyere coke material a polished section was prepared for determination by this microscopic measurement technique.

2.6 Porosity measurements

Variation in porous structure was measured by using the analySIS 3.2 program and Olympus microscope with 520 magnifications. Coke porosity is calculated as the average value of the pieces measured for the tuyere segment.

3. Results & Discussion

In the EBF, coke undergoes many modification such as carbon structure and constituent minerals including alkali phases and porosity as discussed below. Each coke property could influence the coke behavior in an operating blast furnace particularly its strength. The trial with 100% high CRI / low CSR coke lasted for two days and was unique of its kind. With such poor quality regarding strength, it would have been very difficult, if not impossible, to try this in a full scale blast furnace. But it was possible to try it in the EBF and the interesting aspect was the high CRI value.

3.1 Process analysis

Operational data, compared to a reference period with low CRI coke, is shown in Table II.

Table II: Operational data for the reference and trial period

Blast parameters

Reference low CRI

Trial high CRI

blast volume nm3/h 1595 1588 oxygen addition % 1,3 1,9 blast moisture g/nm3 23 20Reductants coke rate kg/thm 436 468 coal rate kg/thm 103 104 reductant rate kg/thm 539 572 Hot Metal C % 4,39 4,33 Si % 1,99 0,9 S % 0,022 0,061 temperature °C 1451 1385Slag volume kg/thm 147 155 CaO/SiO2 0,98 0,91 Top gas temperature °C 188 135 Eta CO 0,46 0,5 Flue dust amount kg/thm 3 8C % 41 55Fe % 20 14

The most important results are summarised as follows:

- The coke rate was increased by more than 30 kg/thm.- Even with the much higher reductant rate it was not possible to keep an acceptable hot

metal heat level. Silicon content and hot metal temperature was significantly lower andsulphur content much higher. Surprisingly, the carbon content was not so much influenced.

- The gas utilisation was significantly increased, although with very high fluctuations, asshown in Figure 2. The same phenomena occur in the EBF when operating with a ferrous burden with a high degree of swelling and disintegration.

- Higher slag volume and lower basicity, because of lower silicon content in the hot metal. - The temperature distribution in the EBF was changed in such a way that the temperature

in the lower shaft became much higher. The cohesive zone moved upwards which was also verified from shaft pressure measurements.

- The amount of flue dust was almost tripled, but the carbon content only increased by 30 %, indicating a more irregular gas flow, also causing also more iron units to leave the furnace with the gas.

Figure 2 Gas utilization for the reference and trial period. The arrow corresponds to the change from reference coketo the trial coke

3.2 Evolution of Carbon Structure

Figure 3 and 4 compares the XRD patterns of cokes from four representative locations of theEBF. It is seen that the width of the 002 carbon peak becomes sharper as the coke descends. The background intensity of lower zone coke samples (e.g. tuyere level) are less than those of cokesamples from the upper part of the EBF (upper shaft samples). Low background intensity isindicative of less proportion of amorphous carbon. The amorphous carbon of coke was found to decline sluggishly up to the cohesive zone, and changed rapidly as the coke descended in the hearth regions of the EBF. This means that coke carbon becomes increasingly ordered as the coke passes from thermal reserve zone to bosh region while amorphous carbon is increasingly depleted.

The comparison between the high and low CRI coke reveals that the temperature profiles havebeen different. It verifies the fact that the cohesive zone was shifted upwards during the trialperiod with high CRI coke.

Upper probe



Lower probe



(a) (b)Figure 3 a) Comparison of x-ray diffraction patterns between low and high CRI coke from the upper probe b) cokefrom lower probe

Inclined probe



Tuyere samples, centre



(a) (b)Figure 4 a) Comparison of x-ray diffraction patterns between low and high CRI coke from the inclined probe b)coke from the tuyere core drilling, from the centre of the furnace

The so-called Lc value was calculated using Scherrer’s equation:

)cos(89.0Lc , where is the wavelength x-ray radiation, is the FWHM (Full Width at

Half Maximum) and 2 is the centre of the 002 carbon peak. Lc

0

20

40

60

80

100

120

upper lower inclined Tuyere wall Tuyerecentre

Ång

strö

m

Lc Ref cokeLc Trial coke

Figure 5 Calculated Lc values from x-ray diffraction spectrum

The Lc increases as the coke descends which can be seen in Figure 5. The lower Lc value generated from the trial coke from the tuyere core centre can be explained by the lower hot metal

temperature i.e. lower heat level in the hearth. The Lc value is higher for coke found in the centre than for coke found at the wall side, thus indicating higher temperature at the centre.

3.3 Evolution of Coke Ash Chemistry

The XRF analyses of the EBF coke samples are provided in Table III. In addition to carbon structure, coke ash chemistry is also continuously changed in the EBF for both coke qualities.

Table III: Chemical composition. The reference low CRI coke corresponds to number 1 and the high CRI trial cokecorresponds to number 2. C stands for Centre, and W for Wall

XRF SiO2 Al2O3 Fe CaO MgO K2O Na2O TiO2 P2O5 S

Coke samples from EBF Campaign 13

Feed 1 4.66 0.35 0.08 0.15 0.04

Feed 2 4.72 1.05 0.62 0.22 0.10

Upper 1 4.07 2.54 1.04 0.50 0.25 0.18 0.09 0.16 0.10 0.53

Upper 2 5.79 2.68 0.76 0.28 0.13 0.23 0.09 0.13 0.04 0.74

Lower 1 3.92 2.62 0.81 0.40 0.23 1.03 0.26 0.14 0.11 0.55

Lower 2 5.42 2.47 1.54 0.60 0.22 0.89 0.19 0.10 0.06 0.82

Inclined 1 4.15 2.50 0.73 0.38 0.21 1.47 0.26 0.14 0.08 0.55

Inclined 2 5.82 2.49 1.46 1.45 0.41 2.95 0.62 0.09 0.07 0.77

Tuyere 1 C 2.91 2.87 0.98 3.42 1.30 0.10 0.01 0.23 0.006 0.40

Tuyere 2 C 1.93 2.58 1.42 1.18 1.20 0.69 0.16 0.12 0.015 0.40

Tuyere 1 W 8.47 4.83 0.54 0.67 0.20 0.35 0.09 0.24 0.008 0

Tuyere 2 W 6.31 4.12 1.40 2.17 0.40 1.02 0.24 0.18 0.02 0.50

The alkali uptake is higher for the high CRI coke as can be seen in Table III and Figure 6a). It is clear that the increase of potassium and sodium is proportional. However, the amount of potassium is much higher for all samples. Figure 6b) reveals the alkali distribution in the tuyere core probe. It can be seen that the highest amount of alkali is found at the wall side of the EBF. Further, the low CRI coke had very little sodium and potassium when compared with the highCRI coke. The reason to this could be the much different temperature profile in the hearth whichis verified by the Lc measurements.

Alkali as function of probes

0

0,5

1

1,5

2

2,5

3

3,5

Feed Upper Lower Inclined TuyereW

TuyereC

% P

otas

sium

oxid

e

0

0,1

0,2

0,3

0,4

0,5

0,6

0,7

% S

odiu

m o

xide

Low CRI K2OHigh CRI K2OLow CRI Na2OHigh CRI Na2O

Tuyere core samples

0,0

0,5

1,0

1,5

2,0

2,5

3,0

Na2O Na2O K2O K2O

Low CRIcoke

High CRIcoke

Low CRIcoke

High CRIcoke

% in

cok

e

wallmid 2mid 1centre

(a) (b)

Figure 6a) Alkali in coke as a function of samples b) Alkali distribution in tuyere core.

Because of the different size distribution between the drill cokes, the amount of -19 mm trial coke and +22.4 mm reference coke were enough for ash chemical analysis. Iron catalyses noticeable carbon gasification, independent of texture type. A decrease in SiO2/Al2O3 ratio in ash, see Figure 7, refers to an intensive silica vaporisation of silicates from furnace wall tocentre. It is clear that with higher temperature there is an increase in the coke porosity. Silicatebreakdown catalysts dissolution of surrounding carbon and mosaic textures starts to change coloras an indication of graphitization. Proportion of carbon textures, which are enriched in silicate dissemination and secondary graphite, should be decreasing, not necessarily isotropic texturereacting easy with CO2.

Alkalis are known as a low temperature indicator when connected with silicates. For the high CRI coke the increased alkali seems to be connected with be the carbon textures. The more of isotropic carbon phases in the coke, the higher the potassium uptake and, maybe, abrasion of the coke. An influence of alkalis on carbon disappearance is not clear in any way. Even gasified areas inside fused isotropic textures are detected. At the tuyere level there seems to be a moredominant relationship between silica vaporization from silicates and increased coke porosity. However, there seems to be a positive relationship between increased gasification spots ofisotropic texture and coke porosity. A small decrease of the high CRI coke porosity with an increase with alkali content can be seen.

Chemical differences of tuyere coke ash

0,5

1

1,5

2



SiO

2/A

l2O

3

0

2

4

6

8

Alk

alis

(%) Low CRI:SiO2/Al2O3

High CRI:SiO2/Al2O3Low CRI:AlkalisHigh CRI:alkalis

Figure 7 Silica and aluminium oxide and alkali content in coke found in tuyere core

3.4 Evolution of Coke Reactivity in the EBF

A TGA comparison between feed, upper, lower, inclined probes, as well as tuyere core probe for both high and low CRI coke was done. Figure 8 to 10 compares the non-isothermal reactivity of the EBF coke. The weight losses of lower zone cokes are consistently higher compared to upper zones coke samples. This implies that the reactivity of coke increases as it descends into theEBF. The coke reactivity is often related to the carbon structure, surface area and coke minerals.In previous section it was seen that carbon structure of lower zone EBF cokes was increasingly ordered. Therefore, the carbon structure alone could not be responsible for increased reactivity oflower zone coke samples as increased crystalline order of carbon is believed to retard the reactivity.

Figure 8a compares the feed of the high and low CRI cokes. The solution loss reaction startsearlier for the high CRI coke and at 1300°C the weight loss is 53 % for the trial coke and 46 % for the reference coke. However, the trial coke had approximately 5 % higher moisture content.

In Figure 8b) the upper probe samples of the reference and trial coke are compared. It is interesting to notice that the low CRI coke actually reacts faster than the high CRI coke. The mass change is 64 % and 37 % respectively. The same pattern is seen in Figure 9a) and b). At tuyere level the high CRI coke does react faster and has a greater mass loss, see Figure 10a) and b). However, it is a little difference and one would expect a much larger difference.

400 500 600 700 800 900 1000 1100 1200 1300Temperature /°C

50

60

70

80

90

100

TG /%

High CRI coke

Low CRI coke


-46.00 %

400 500 600 700 800 900 1000 1100 1200 1300Temperature /°C

40

50

60

70

80

90

100

TG /%

High CRI coke

Low CRI coke

-36.76 %


(a) (b)

Figure 8 a) Reference and trial coke from feed b) Reference and trial coke from upper probe

400 500 600 700 800 900 1000 1100 1200 1300Temperature /°C

40

50

60

70

80

90

100

TG /%

High CRI coke

Low CRI coke


-49.92 %

400 500 600 700 800 900 1000 1100 1200 1300Temperature /°C

20

30

40

50

60

70

80

90

100

TG /%

High CRI coke

Low CRI coke


-81.74 %

(a) (b)

Figure 9 a) Reference and trial coke from lower probe b) Reference and trial coke from inclined probe

400 500 600 700 800 900 1000 1100 1200 1300Temperature /°C

20.0

30.0

40.0

50.0

60.0

70.0

80.0

90.0

100.0

TG /%

High CRI coke

Low CRI coke


-76.24 %

400 500 600 700 800 900 1000 1100 1200 1300Temperature /°C

20

30

40

50

60

70

80

90

100

TG /%

High CRI coke

Low CRI coke


-72.70 %

(a) centre (b) wall

Figure 10 a) Reference and trial coke from centre of the EBF at tuyere level b) Reference and trial coke from wall side of the EBF at tuyere level

The TGA study implies that the CRI value have little influence when the coke have been crushed, thus suggesting that structure plays a dominant role.

3.5 Isotropic / anisotropic changes in the coke carbon microstructure

In Figures 11a) and 11b) the results are presented as reduced to the Anisotropictotal and Isotropictotal components as mean values of the different fractions. With respect to the measuringmethod the Anisotropics and the Isotropics are detected in close correspondence, i.e. the sum always results in 100%.

The measurements of the feed cokes result in very different levels of the detected anisotropy. The high quality reference coke contains 79 vol.% Anisotropics corresponding to 21 vol.% Isotropics, the low quality high CRI coke is 62 vol.% of Anisotropics and 38 vol.% Isotropics.This result reflects a very different origin of the coals (coal blends) used for the coke typesdetermined.

With respect to the coke texture of the tuyere coke at both operations the Isotropics increase towards the centre of the EBF. In the case of the high CRI coke at the centre an increase of 10% in the Isotropics is corresponding to a 10% decrease of the Anisotropics towards the centre. The changes in the high quality reference coke are less, in the Isotropics an increase of up to 5%corresponds with a 5% decrease in the Anisotropis.

With respect to the behaviour of these cokes during descending in the EBF it can be concluded, that the isotropic carbon components are more metallurgical treatment resistant under load of the EBF compared to the anisotropic components. Apparently the gasification attacks the anisotropiccarbon components more than the Isotropics.

45

50

55

60

65

70

75

80

85


Ani

sotr

opic

tota

l in

vol.%



High CRI feed coke


15

20

25

30

35

40

45

50

55


Isot

ropi

cto

tal i

n vo

l. %



High CRI feed coke


(a) (b)Figure 11 a) Anisotropic coke carbon components b) Isotropic coke carbon components

3.6 Porosity differences

The tuyere drill core was divided in four segments, i.e. Centre, Mid 1, Mid 2, and Wall. It wasthen sieved to fractions of -19 mm, 19-22.4 mm and +22.4 mm. Variation in tuyere cokeporosity by the image analysis was measured using 19-22.4 mm drill coke. The 50 g imageanalysis sample of about 10 coke pieces is constructed at the same weight distribution as insidethe 19-22.4 mm drill core.

Variation in porous structure was measured for two sets of tuyere coke, low and high CRI, see Figure 12 and 13. Coke porosity is calculated as the average value of the pieces measured for the tuyere segment. The image analysis shows a greater porosity percent for low CRI coke which

gently increases towards the EBF centre, see Figure 14. A similar increasing porosity line is true for the high CRI coke but an opposite downward trend in the EBF centre.

Low CRI coke porosity versus pore area

1000

3000

5000

40 50 60 70 80

Porosity (%)

Por

e (u

m2) Wall

0.25-0.5 m0.5-75 mCentre

Figure 12 Pore area vs. porosity for the reference coke

High CRI coke porosity versus pore area

1000

3000

5000

40 50 60 70 80

Porosity (%)

Pore

(um

2) Wall0.25-0.5 m0.5-75 mCentre

Figure 13 Pore area vs. porosity for the trial coke

Tuyere coke porosity versus pore area

40

50

60

70



Poro

sity

(%)

2000

2500

3000

3500

4000

Cok

e si

ze (u

m2)

Low CRI/PorosityHigh CRI/PorosityLow CRI/PoreHigh CRI/Pore

Figure 14 The tuyere coke porosity vs. pore area for both of the coke types

4. Conclusions

A test with a poor coke quality (high CRI/low CSR) was conducted in an experimental blast furnace. Comparison was made with a reference period using high quality coke (low CRI/high CSR). Evolution of coke properties particularly carbon structure and alkali uptake were related to CO2 reactivity. On the basis of this study, following conclusions were made.

1. Process: The cohesive zone was shifted upwards during the high CRI coke trial. The gas utilization was higher but more fluctuating during the trial period. Despite a much higher reductant rate, hot metal temperature was too low. Flue dust generation was significantly higher for the trial period.

2. Graphitisation: The graphitisation increased as the cokes descended through the EBF. The degrees of graphitisation were different for the high and low CRI cokes when comparing feed and tuyere samples which verify different temperature profiles.

3. Alkali uptake is higher for high CRI coke. This could be explained by a higher specific surface area for the high CRI coke.

4. TGA with CO2: The CO2 reactivity of coke was found to increase during progressive movement of the coke from thermal reserve zone to cohesive zone of the EBF, and was related to the catalytic effect of increased alkali concentration in coke. This was the same for both cokes. CRI plays little role when comparing reactivity with CO2 of crushed sample in TGA.

5. Porosity: The tuyere drill results indicate a higher porosity and larger pores for both cokes at the centre where the temperature has been higher. The reference coke has a higher porosity but smaller pores.

6. Isotropic/anisotropic changes: Starting from different levels the anisotropic coke carbon components decreased and the isotropic components correspondingly increased as the coke descended through the EBF. Thus the isotropic coke components seem to be more metallurgical treatment resistant.

ACKNOWLEDGEMENTS

The authors would like to acknowledge RFCS and the Swedish Energy Agency for partly funding the project. Thanks to LKAB for providing the samples and the opportunity to conduct this research. Thanks to Prof. Veena Sahajwalla and Dr Sushil Gupta at the University of New South Wales for their assistance.

REFERENCES

1) Coke - Determination of Coke Reactivity Index (CRI) and Coke Strength after Reaction (CSR): ISO/DIS 18894, (2001).

2) Annual Book of ASTM Standards, Section 5: Petroleum Products, Lubricants and Fossil Fuels, vol. 05.05 (1996).

3) Grosspietsch, K. H., H. B. Lungen, G. Dauwels, Ferstl, T. Karjalahti, P. Negro, B. van der Velden and R. Willmers: Proc. 4th European Coke and Ironmaking Congress, Paris, 1,(2000), 2-11.

4) Arendt, P., F. Huhn and H. Kühl: Cokemaking International, 2, (2001), 50-53. 5) Kerkkonen, O.: Iron & Steel Technology Conference Proceedings, AISTech, Nashville, 1,

(2004), 469-482. 6) Willmers, R. R. and C. R. Bennington: 2nd International Cokemaking Congress, London,

UK 28-30th September 1992. 260-285. 7) Feng, B., S. K. Bhatia and J. C. Barry: Carbon, 40, (2002), 481-496. 8) Helleisen, M., R. Nicolle, J. M. Steiler, N. Jusseau, C. Meltzheim and C. Thiriom: 1st

International Cokemaking Congress, Essen, Germany, C2, 1987. 1-20. 9) Forsberg, S: 1st International Cokemaking Congress, Essen, Germany, C6, 1987. 1-19. 10) van der Velden, B: McMaster Cokemaking Course, Hamilton, Canada, May, 2003. 22.1-

112.11) Steiler, J. M., R. Nicolle, P. Negro, M. Helleisen, N. Jusseau, B. Metz and C. Thirion:

Ironmaking Conference Proceedings, ISS, Washington DC, USA, 50, 1991. 715-723. 12) Tucker, J. and J. Goleczka: 1st International Cokemaking Congress, Essen, Germany, C5,

1987. 1-24. 13) Gornostayev, S., Kerkkonen, O., Härkki, J. Journal of the Iron & Steel Institute of Japan,

Vol 45, (2005), No 1. pp 1-7. 14) Chan, B. K. C., K. M. Thomas and H. Marsh: Carbon, 31, (1993), 1071-1082. 15) Dahlstedt, A., M. Hallin and M. Tottie: SCANMET, Luleå, Sweden, 1, 7-8th June, 1999.

235-252.16) Dahlstedt, A., M. Hallin and J.-O. Wikström: Proc. 4th European Coke and Ironmaking

Congress, Paris, France, 1, 2000. 138-145. 17) Hooey, L., J. Sterneland and M. Hallin: in 60th Ironmaking Conference Proceedings,

Baltimore, ISS, USA, 60, March 2001, 197-208. 18) Lu, L., V. Sahajwalla, C. Kong and D. Harris: Carbon, 39 (2001), No. 12, 1821-1833. 19) Sahajwalla, V., M. Dubikova and R. Khanna: in Proc. Tenth International Ferroalloys

Congress, Cape Town, South Africa, 1-4th February, 2004. 351-356. 20) Kerkkonen, O., P. Arendt and H. Kuhl: 61st Ironmaking Conference Proceedings, ISS,

Nashville, USA, 10-13th March, 2002. 393-404. 21) Vandezande, J.A: 44th Ironmaking Conference Proceedings, ISS, USA, (1985), 189. 22) Dubrawaski, J.V. and W.W. Gill: Ironmaking and Steelmaking, 1, (1984), 7. 23) Y. Kashiwaya, M. Takahata, K. Ishii, K. Yamaguchi, M. Naito, H. Hasegawa: Tetsu-to-

Hagane, Journal of the Iron & Steel Institute of Japan, 87(5), (2001), 259-265.

evolution of coke properties while descending through a ...€¦ · coke has ancient origins and...

Documents