Embed Size (px)
Citation preview
Loughborough UniversityInstitutional Repository
The carbon texture ofmetallurgical coke and itsbearing on coke quality
prediction
This item was submitted to Loughborough University's Institutional Repositoryby the/an author.
Additional Information:
• A Doctoral Thesis. Submitted in partial fulfilment of the requirementsfor the award of Doctor of Philosophy of Loughborough University.
Metadata Record: https://dspace.lboro.ac.uk/2134/10950
Publisher: c© Alan Walker
Please cite the published version.
This item was submitted to Loughborough University as a PhD thesis by the author and is made available in the Institutional Repository
(https://dspace.lboro.ac.uk/) under the following Creative Commons Licence conditions.
For the full text of this licence, please go to: http://creativecommons.org/licenses/by-nc-nd/2.5/
i .j -
.~ , 1 ;J ,
'. 1
B)""'OSC r\O :-'0 <a31i3 L0UGHBOROUGH
UNIVERSITY OF TECHNOLOGY LIBRARY
AUTHOR/FILING TITl.E
WA\...~Ita.. A ----------------------1-----------------------
ACCESSION/COPY NO.
--VOL~NO~------- -c~Iit-~-ft-~---------------- I
- 5 jUL 1991
=!~ .. ' .' - 1 J Ul 1994
.'.
-;. ..
, .
THE CARBON TEXTURE OF METALLURGICAL
COKE
AND ITS BEARING ON COKE QUALITY PREDICTION
by
ALAN WALKER
A Doctoral Thesis
Submitted in partial fulfilment of the requirements
for the award of
Doctor of Philosophy
of the Loughborough University of Technology
May 1988
~ by Alan Walker 1988
LourhbO~Go::h U"iver:oity
01' T ~.!:. • t icrUY l--~""-'" ___ -':'-1 ~~~\'!~:O.'6_._-I ~~~ __ ~. ~ ___ .-.,-----l
~<,;' ()?loQ5\o~
ABSTRACT
The carbon in metallurgical coke is composed of textural units, varying
in size and shape depending on the rank of coal carbonized. These induce
a characteristic texture to coke surfaces. This thesis describes a study
of the bearing of this texture on coke strength, particular emphasis
being placed on investigating the feasibility of using textural
composition data, determined by either scanning electron microscopy (SEX)
of etched surfaces or polarized-light microscopy (PLX) of polished coke
surfaces, as a basis of predicting the tensile strength of cokes produced
from blended-coal charges from the behaviour of individual blend
components.
Scanning electron microscopy (SEM) of fractured coke surfaces revealed
differences in the mode of fracture of textural components which implied
variations in their contribution to coke strength. The tensile strengths
of pilot-oven cokes, produced from blended-coal charges, could be related
to their measured PLM textural compositions using equations derived from
consideration of simple models of intergranular and transgranular
fracture.
The coke strengths could also be related, with greater precision, with
texturat data calculated from the coal blend composition and either the
SEX or the PLX textural data for the cokes from the individual blend
components. It was further found that the strength of blended-coal cokes
were additively related to the blend composition and the tensile
strengths of the single-coal cokes. Such relationships are useful, at the
very least, for predicting the strength of cokes from other blends of the
same coals carbonized under similar conditions. The various approaches to .
coke strength prediction have potential value in different situations.
ACKNOWLEDGEMENTS ~~'~o ~~ P::~ ft~1 vv;, CA4J>Cf-A~
o 0 ~--;P--(l-P"~,/' <t.-/..-t~ I offer my gratitude and thanks to Dr. J. W. Patrick, Director of
the Carbon Research Group,( collegue and friend for more years . f! -,ifw;'
than either of us would now wish to count'lor his guidance and
encouragement throughout this project. A. ;t:<~--tt:r __ I
The experimental assistance of Mr. Douglas Hays is gratefully
acknowledge&. Many were the occasions when we conspired to ! thwart the provisions of the Health and Safety at Work Act.
It was a pleasure to have the co-operation of Miss Angela
Moreland, hopefully soon to earn a Doctorate herself, during the
establishment of the technique of PLM texture analysis.
I thank
0'" 1-:"[.;;
my wife/' Frances, for
, #i-' her support throughout this study
and especially for hef'forebearance during the latter months.
-;tN,~,,'
Finally, I am grateful for the financial support of the European
Coal and Steel Community for the various programmes of which "
this study formed a part.
\
PUBLISHED PAPERS
The following papers, based on the work described in this thesis,
have been published or accepted for publication.
Hays, D., Patrick, J. W. and Walker, A.
'An SEM study of fractured and etched metallurgical coke
surfaces,'
Fuel 1982, 61, 232
Patrick, J. W. and Walker, A.
'Preliminary studies of the relationship between the carbon
texture and the strength of metallurgical cokes.'
Fuel 1985, 64, 136
Walker, A.
'Laboratory coal carbonization oven'
Fuel 1985, 64, 1327
Patrick, J. W. and Walker', A.
'An SEM study of the tensile fracture of metallurgical coke'
Journal of Materials Science 1987, 22, 3589
Moreland, A., Patrick, J. W. and Walker, A.
'Optical anisotropY'in cokes from high-rank coals'
Paper accepted for publication in Fuel.
CONTENTS
1. INTRODUCTION
2. LITERATURE REVIEW
2.1 The production and use of blast-furnace coke
2.1.1 Production
2.1.2 Use of coke in the blast furnace
2.2 Aspects of the science of cokemaking
2.2.1 The nature of coal
2.2.2 The classification Df 60als
2.2.3 The coal to coke transformation
2.3 The strength of coke
2.3.1 The fracture of brittle materials
2.3.2 The influence of porosity
2.3.3 Drum tests
2.3.4 The tensile strength of coke
2.3.5 The microstrength test
2.4 The texture of metallurgical coke
-f.-2.4.1 Polarized-light microscopy as applied to carbons
2.4.2 Early studies ·of coke texture
~ 2.4.3 The formation of graphitizing carbons
2.4.4 Development of the texture in metallurgical cokes
2.4.5 The mechanism of the development of coke texture
2.4.6 The classification of coke textural components
2.4.7 The application of coke textural data
2.5 The prediction of coke quality
2.5.1 The prediction of drum indices
2.5.1.1 Coal petrography in coke strength prediction
2.5.1.2 Methods based on the dilatometric behaviour
of coal
2.5.2 The predi~tion of the tensile strength of coke
Page No.
1
3
4
4
6
12 , 12
13
16
20
21
23
24
26
28
30
30
34
35
37
39
43
44
48
48
48
51
54
2.5.2.1 The use of pore
~-----. -2-.-5:2',-2 -Using -the-" bond
2.6 Outline of research
3. EXPERIMENTAL STUDIES
structural parameters
strength'- approach---- -
3.1 An.SEX study of fractured and etched coke surfaces
3.1.1 Introduction
3.1.2 Experimental procedures
3.1.2.1 Cokes used
3.1.2.2 Carbonization procedures
3.1.2.3 Tensile strength determination
3.1.2.4 Specimen preparation for SEK examination
3.1.3 Results
3.1.4 Discussion
3.2 An SEX study of the tensile fracture of coke
3.2.1 Introduction
3.2.2 Experimental procedures
3.2.2.1 Coke used
3.2.2.2 Specimen preparation
3.2.2.3 SEM examination
3.2.3 Results
3.2.4 Discussion
3.3 SEX texture and coke strength prediction
3.3.1 Introduction
3.3.2 Experimental procedures
3.3.2.1 Blends carbonized
3.3.3 Results
3.3.4 Discussion
3.4 PLX texture and coke strength prediction
3.4.1 Introduction
3.4.2 Experimental procedures
3.4.2.1 Determination of PLK textural data
3.4.3 Re·sul ts
3.4.4 Discussion
55
-- 56 --
57
58
58
58 ... - 58
58
59
60
61
61
64
68
68 68
68
69
70
70
72
78
78
79
79
80
81
90
90
91
91
91
92
4. GENERAL DISCUSSION
4.1 The nature of coke textural components -----~--
4.2 The influence of blending on coke textural data
4.3 The influence of texture on coke strength
4.4 Coke tensile strength prediction
4.5 Recommendations for further work
5. CONCLUSIONS
REFERENCES
APPENDIX I :Fitting of INTER(31), TRANS(33) and
INERT(35) equations to data
APPENDIX 11 :Triangular diagrams: iso-strength line
calculation
APPENDIX III :Derivation of INTER(31), TRANS(33) and
INERT(35) equations
TABLES
FIGURES
99
99
103
107
116
121
122
127
135
139
141
- 1 -
1. INTRODUCTION
The world wide production of hard coke in the first half of this decade
exceeded 330Mt annually [1]. In the same period, the average annual
consumption in the U.K. was 6.75Mt of which 5.22Mt and O.26Mt were used
in blast furnaces and foundry cupolas respectively [2]. Thus the coking
industry is both large and economically important. In the U.K. and
Western Europe generally, the demand for metallurgical coke has decreased
in recent years but at the same time there has been an increasing
requirement for coke of superior quality. Metallurgical coke is produced
by the carbonization of blends of coals to about lOOO·C in slot-type
ovens [3]. The different specifications of blast furnace and foundry cokes
impose differences in both the blends carbonized and the carbonization
conditions adopted [3]. The study described in this thesis is concerned
primarily with coke suitable for blast furnace use.
Faced with recurring changes in the pattern of coal supplies and/or the
requirments of the iron-maker, to maintain or improve the quality of his
coke, the coke-oven manager has little alternative but to alter the
composition of the coal blend he carbonizes. Major changes in blend
composition involve testing a limited number of the many possible blends
available in pilot ovens ranging in size up to that of a one-half length
commercial oven requiring a coal charge weighing 17t [4], Such testing
programmes are expensive, hence methods of predicting the quality of
coke, in particular its strength, from the laboratory testing of coals
and/or cokes are continually being sought. In the U.S.A. the petrographic
examination of coals forms the basis of widely used methods of coke
quality prediction but this approach has been less successful in the U.K.
and Europe in general [5].
Smooth, efficient blast-furnace operation requires that the coke should
resist size degradation as it progresses down the the stack [6]. Thus
current specifications for blast-furnace coke invariably include some
specification for the strength of the coke. Frequently, coke strength is
defined in terms of indices derived from the reduction in size of lump
- 2 -
coke in standardized drum tests [7], However, coke .is a brittle material
and thus, despite the mode of the imposed stress, breakage is considered
to occur as a result of induced tensile forces [8J. Consequently, at least
on the research level, the tensile strength of coke is gaining acceptance
as an indicator of coke quality [9J.
The tensile strength of coke has been related to pore structural
parameters by equations derived without regard to possible variations in
the properties of the coke carbon [10J. This is composed of textural
units, building blocks, which vary in size depending on the rank of the
coal carbonized [llJ. These induce a cliaracteristic texture to coke
surfaces when viewed microscopically. The first part of this study
consisted of an investigation of the use of scanning electron microscopy
(SEM) to examine the texture of carbon in metallurgical coke. This led to
an investigation of the bearing of the SEM texture of coke on
considerations of coke strength, particular emphasis being paid to
assessing the feasibility of using textural compositional data as a basis
of predicting the strength of cokes, made using blended coal charges,
from the behaviour of individual blend components. Since scanning
electron microscopes are not usually found in coke-quality laboratories,
an attempt was also made to assess the applicability of the approach
developed to textural data obtained using polarized light microscopy
(PLIO.
This thesis is devided into five sections. The relevant literature is
reviewed in Section 2. Experimental studies and the results obtained are
described in Section 3, this being divided into four parts which logically
reflect varying phases of the work. Each part contains a discussion of
the results obtained. A general discussion follows in Section 4 and
finally the conclusions are presented in Section 5. References and three
appendices follow. Tables and figures are to be found at the end of the
thesis in numerical order on unnumbered pages.
':- 3 -
2. LITERATURE REVIEW
The production of coke for use in metallurgical processes originated some
260 years ago with Abraham Darby's successful use of coke in a blast
furnace at Coalbrookdale in Shropshire (121. This has lead to. the growth
of the present large coking industry operating on a world-wide basis.
Accordingly the available literature is too extensive to be reviewed in
its entirety here, so that this review is, of necessity, selective. Its
objective, therefore, is to present a critical assessment of that
background information, considered most appropriate to the present study,
which is available in English or as a readily accessible translation.
A brief description of the production and use of blast-furnace coke
identifies those coal properties essential for metallurgical coke-making
and those coke properties important in blast-furnace operation. Scientific
aspects of the coking of coal are then discussed and the quality criteria
for blast-furnace coke reviewed. There follows an appraisal of methods,
previously suggested or established, for relating coal properties and coke
quality. The development of the texture of the carbon in metallurgical
coke is then considered in the light of modern views on the formation of
graphitizing carbons by the pyrolysis of carbonaceous precursors which
soften on heating. A consideration of the use made of coke textural data
follows. Finally, methods, established or suggested, for predicting coke
quality.from coal or coke properties are appraised.
The review leads to an outline of the research programme.
- 4 -
i /2.1 The production and use of blast-furnace coke
2.1.1 Production
Metallurgical coke is now predominantly produced by the pyrolysis of
blends of suitable coals to about 1000'C in slot-type ovens [31, the
previous beehive ovens having been completely superseded and formed-coke
briquettes not yet having gained wide acceptance. Blast-furnace and
foundry coke are made in a broadly similar manner but differences in
blend composition and carbonization conditions are necessary to meet
different coke specifications.
The coking of coal consists of charging powdered coal, typically 80% by
weight less than 3mm, into a hot oven with side walls heated to about
1250'C [31. The coal blend used must exhibit some degree of fluidity or
plasticity so that the essential feature of the coking process, the fusion
of particulate coal into massive, porous coke, can take place. This
normally occurs in the temperature range 350-500'C, depending on the rank
of the coal used, before resolidification ensues. Thus two plastic layers
are formed within the coal, charge parallel to the oven walls and these
move progressively towards the oven centre as heat transfer from the
walls takes place. Carbonization is considered complete when the centre
charge temperature attains an arbitarily selected temperature, usually in
the range 900-1000·C.
Coke ovens can have a useful life exceeding 25 years. Existing ovens
therefore vary markedly in size, modern ovens being considerably larger
as a result of continued development of construction materials and
techniques. A modern U.K. installation at the Llanwern'works of'the
British Steel Corporation has ovens 14.6m long, 6.25m tall and 0.45m
average width [131. These are charged with 30.5t of wet coal ( moisture
content approximately 8% by weight ) to a levelled height of 5.98m.
- 5 -
Charges are carbonized to a centre-charge temperature of 950·C at 25mm/h
in 17.3h. There are fifty-three ovens in the battery, total design output
being 1570t/day.
In any coke-oven battery, each individual oven has removable doors at
each end for use during coke discharge [3)-. In the top of each oven are
three or four circular openings through which the prepared coal is
charged into the oven in weighed amounts from a mobile charging car, the
larry-car. Most ovens have two other openings in the top, one at each
end, for conducting volatile by-products, tars and gases, -through cast
iron stand-pipes into collecting mains. These extend the whole length of
the battery and conduct the by-products to the recovery plant.
Blends of crushed coal are stored in overhead bunkers which feed the
required weight of blend into the larry-car. This travels the whole
length of the battery to charge individual ovens through the charge
holes. If wet, the charge is then levelled mechanically. To minimise
pollution, modern practice is to charge on-main, i.e. with suction on the
stand-pipes.
Coking is complete in 12-30h depending on the oven width, the wall
temperature and the type of coke being produced, 25mm/h being a commonly
used mean carbonization rate. Doors at either end of the slightly-tapered
oven are then opened and a ram, inserted from the narrow end, pushes the
coke, normally shrunken from the walls, from the wider-end of the oven
into a quenching car. This moves under a quenching tower where the
incandescent coke is rapidly cooled by a water spray. The cool~d coke,
containing about 5% water by weight, is then graded for use. A modern
alternative is to 'dry quench' the coke in a closed container under
flowing nitrogen.
a~<
The pattern of oven pushing is so organized that there~several full ovens
between successive ovens being pushed. This avoids uneven heating of the
- 6 -
battery. Ovens are not allowed to stand empty nor are they allowed to
cool since this would have an adverse effect on the silica brickware.
Ovens are gas heated with their own product except where lower quality
gas, for example blast-furnace gas, is available.
Shrinkage of the charge, so necessary for ready pushing, also leads to
fissures within the coke, the pattern of fissuring governing the coke size
[14J. Larger coke is required for foundry use [3J and this is achieved by
including anti-fissuring agents [15J, coke breeze, non-fusing high-rank
coals and anthracites, in the blend and by using low carbonization rates.
The carbonization rate at the Cwm coking plant of the National Coal Board
is 15mm/h [16J, i.e. two-thirds the rate used at Llanwern to produce
blast-furnace coke.
Coke quality can usually be improved by increasing the bulk density of
the charge within the oven [17J. Care is necessary however with coals
prone to imposing high pressures on the oven walls during carbonization
since increasing the charge bulk density only aggravates the situation
[18J. Increased bulk density can be achieved by stamp-charging [19J, i.e.
compressing the charge in a large mould before charging through the ram
side door, partial briquetting [20J, charging a mixture of crushed coal
and tar- or pitch-bonded briquettes, or by preheating the charge to about
250'C [21J. Preheated blends may be charged either from a larry-car [22]
or-by pipeline to a position above the ram-side door normally occupied by
the leveller door [23J. Preheated charges are self-levelling. _
2.1.2 Use of coke in the blast furnace
The iron blast -furnace is a tall vertical shaft furnace [24J which makes
use of the carbon in metallurgical coke, directly and indirectiy, to
reduce iron oxides to 'pig iron', containing 4-5wt % carbon and O.5-1wt %
silicon, which is suitable for refining into steel. Like coke ovens,
currently operating blast furnaces vary considerably in size, the largest
- 7 -
being ~some 50m in _total height and 16m in external diameter. _A furnace of
this size produces 10,OOOt of iron/day, consuming 4,500t of coke in the
process.
The solid raw materials for the process, iron ore, fluxing agents, and
coke are charged via a bell at the top of the furnace and distributed
evenly across the diameter of the shaft. Coke, sized 20-80mm, and
mixtures of sintered, pelleted, and/or sized ore with calcium and
magnesium oxide fluxing agents, are charged alternately forming layers
O.5-1m thick. Air, preheated to 1200'C, and perhaps enriched with oxygen,
is injected through tuyeres near the bottom of the shaft to react with
the coke producing carbon monoxide, the principal reductant, and the heat
necessary both for the endothermic reduction reactions and to melt the
iron and slag produced. Gaseous, liquid or solid hydrocarbons may be
injected at the tuyeres to provide additional reducing capacity.
The principal product, molten pig iron, is tapped intermittantly from the
bottom of the furnace. Two by-products are formed. Molten slag, the
product of the reaction of impurities in the burden with the fluxing
agents, is also tapped from the hearth, while gases, containing dust
particles, exit via the the gas collection system at the top. Furnace
operation is usually stable, continous operation for five to eight years
being possible before refractory wear forces a shut down.
The reactions which occur in the blast-furnace shaft are many, varied,
and complex. It is therefore intended to concentrate on the roles played
by the coke in the changing chemical and thermal environment it
encounters as it passes down the furnace shaft, and to identify those
properties which enable it to fulfil its various functions.
Present understanding of the behaviour of raw materials after they are
charged to the blast furnace stems largely from the studies of quenched
- 8 -
blast furnaces carried out in Japan [25], As Fig. 1 illustrates, moving
down the shaft, several distinct zones can be identified :-
1. a region of slowly rising temperature ( 750-1000·C ) in which the
alternate layers of burden materials retain their form,
2. a fusion zone ( 1100-1400·C ) where the iron and slag form
soften and melt,
3. an active coke zone ( 1400-1700·C ) in which loosely packed coke
moves down to be burnt in front of the tuyeres,
4. a raceway ( 1700-1800'C ) in front of the tuyeres where incandescent
coke is thrown about violently in the blast and is burnt in the
oxygen it contains,
5. a static coke bed ( 1400-1700·C ) extending down to the hearth,
through which molten iron and slag percolate.
Reduction of iron oxides is essentially complete by the beginning of the
fusion zone [24], Higher oxides of iron are first reduced to wustite, the
reduction of wustite itself only becoming thermodynamically feasible at
higher temperatures and at higher concentrations of carbon monoxide.
Below about 900·C, the reaction of the coke carbon with carbon dioxide is
relatively slow so that up to this temperature reduction of iron oxides
occurs by reaction with carbon monoxide formed further down the shaft.
At higher temperatures, the carbon dioxide formed from oxide reduction is
capable of regenerating carhon monoxide by reaction with the coke carbon.
It is considered that perhaps 20-30% by weight of the coke carbon is
consumed in this way [261. The carbon-carbon dioxide reaction is
catalysed by the alkali metals and their compounds [27J. Potassium and
sodium enter the furnace as. impurities in the burden, largely in the form
of complex aluminates and silicates. At tuyere level, these release
potassium .and sodium metals which then take part in a complex series of
reactions which result in their circulation, in a variety of chemical and
physical forms, within zones 1-4 [281. In addition to a· catalytic
influence, the metallic vapour of potassium in particular is capable of
- 9 -
reacting with the coke carbon lattice causing its expansion as
intercalation compounds are formed [29).
F'rom the fusion zone to the tuyeres, the principal role of the coke, being
the only solid present, is to provide a porous bed for the upward flow of
gases and the downward flow of liquids [24), Little coke gasification
takes place in the active coke zone but the coke is subjected to
progressively higher temperatures so the size of the graphitic
crystallites increases [30). Coke entering the raceway is burnt to carbon
monoxide by the oxygen in the blast. Temperatures in the raceway are so
high that the rate of combustion of the coke will be governed by
diffusion of oxygen across the boundary layer [31) rather than by
properties of the coke.
It is now evident that coke fulfils three major roles in the blast
furnace, as a fuel supplying heat for endothermic reactions and to melt
the iron and slag formed, as a source of carbon for the generation and
regeneration of the principal reductant, carbon monoxide, and as a
refractory material maintaining, at high temperatures, a bed permeable to
gases and liquids.
The output of a blast furnace is dependent upon the amount of oxygen
burnt at the tuyeres and the efficiency with which the reductants are
used within the stack [24), Optimum output therefore requires sufficient
permeability within the stack to permit high gas flows evenly distributed
through the burden. Improvements in output have stemmed from careful
sizing of all the raw materials fed to the furnace [24). However once
fusion of the metal and slag occurs, gas flow is governed by the
permeability of the coke bed. This depends on the mean size and size
distribution of the coke [32). Thus, excessive size degradation within the
stack adversely affects furnace performance.
- 10-
Several factors contribute to the size degradation which takes place
wi thin the furnace shaft [26]; the mechanical shock impo<:ed by the fall
from the bell to the stockline, the abrasion by other raw materials, and
the differential shrinkage imposed by alkali intercalation [33] and heat
treatment above the carbonization temperature [34]. Resistance to these
effects is dependent upon the coke possessing adequate initial strength , and the extent to which this is influenced by reaction with carbon
dioxide. Thus some measure of coke strength appears in all specifications
for blast furnace coke. There is also a current trend towards specifying
a maximum reactivity towards carbon dioxide. Resistance to alkali attack
is not specified but control of the alkali loading of blast furnaces [24]
is recommended.
The specifications for the coke used at three U.K. blast furnaces of
differing hearth diameters [35] are given in Table 1 together with, for
comparison purposes, the specification for a foundry coke [361. The
strength of the blast-furnace cokes is specified in terms of Kicum
indices, the determination of which will be described later. As the table
indicates, the larger blast furnaces require higher quality coke ( lower
Kicum KlO index ). For the largest furnace considered, the coke must also
have low reactivity towards carbon dioxide and high strength after
reaction. These properties are determined according to an empirical
method [37] originally· developed in Japan. For foundry use, coke size is
the most important physical parameter [36], the mean size of the coke
used being twice that of coke fed to blast furnaces.
This section of the review shows that the cokemakers principal task is
to choose coal blends which, when carbonized in his ovens under standard
operating conditions, produce a coke strong enough, and of sufficiently
low reactivity towards carbon dioxide, to withstand the environment
within a blast furnace without suffering excessive size degradation. The
coke must also meet the specification for ash and sulphur but these can
usually be estimated from the chemical analyses of the coals carbonized.
-11-
However the cokemaker operates within a business environment so that
rarely- is-he-pel'mitted to produce -the-best achievable coke. Economic -
considerations might, for example, dictate the inclusion of a low cost,
but inferior, local coal, possibly with high sulphur content, as a
component in any blend he carbonizes.
- 12-
2.2 Aspects of the science of cokemaking - - - ---- - --
In order successfully to select coal blends capable of meeting his coke
specification, the cokemaker must possess a sound knowledge of the nature
of coals and their behaviour on heating. The object of this section is to
~ present an outlinE! of those aspects or coal science most relevent to
cokemaking. Space limitations dictate that discussion of many detailed
scientific aspects, in particular the chemical structure of coals,
chemical studies of coal carbonization, and theories of the softening and
swelling of coals, be excluded. Consideration is given to the general
philosophy of coal blending for coke production but a detailed review of
attempts to predict coke quality from the laboratory examination of coals
and/or cokes is giveh later. Except where otherwise referenced, this
section is based on two extensive reviews [38,39],
2.2.1 The nature. of coal
Bituminous coals, along with peat, lignite, and anthracite, belong to a
group of fossil fuels derived from plant material. They are therefore
organic in nature and consist primarily of carbon in chemical association
with hydrogen, oxygen, and sulphur. They also contain significant amounts
of inorganic mineral matter.
The origins of coals go back many millions of years, approximately 300
million for U.K. coals, to the coverage of a peat bog by an impervious
sediment. Slow chemical reactions then led to the coalification of the
peat, forming .progressively lignite, bituminous coal, and anthracite as
the carbon content increased. The temperature associated with increasing
depth is considered the factor most important in enhancing coalification.
Because of its origin, coal is not a mineral of constant composition but
an organic rock whose chemical composition changes as coalification
-13~
advances. Thus, within a single U.K. coalfield, large variations in the
rank of the coal, as measured by the carbon or volatile matter content,
are common, and minor variations occur in the output of a mine working a
single seam. Thus, chemical analysis provides one method of describing
coals.
Alternatively coals maybe described in terms of the proportions of the
various rock types present. That coals are not homogeneous is evident to
the naked eye, bands of bright and dull coals being visible in even a
single piece of coal. Stopes classified these rock types into four
categories, vitrain and clarain being bright coals, durain dull, and
fusain powdery. Microscopic examination, now generally carried out using
light reflected from polished coal surfaces, reveals these rock types to
be composed of mixtures of more or less homogeneous maceral types.
Macerals can be classified according to their origin. Vitrinite, fusinite
and semi-fusinite are derived from woody tissues, exinite from plant
tissues, while the origin of micrinite is less clear. Vitrinite and
exinite, in particular, are each d~vided into several sub-categories.
Empirical relationships between the chemical analysis of vitrinites and
their reflectance being available, it is possible to assess the rank of a
coal from microscopic examination. Thus, in petrographic analysis of
coals, both the maceral composition and the vitrinite reflectance are
quoted.
2.2.2 The classification of coals
Although international classifications for coals exist, internally coal
producing countries persist in using their own national classifications.
Following this precedent, the following remarks pertain to U.K. coals.
In the U.K., a scientific coal classification, based on elemental analysis,
was originally drawn up by Seyler in 1899 and later extended. In this,
coals were allocated into four principal groups, anthracitic (A),
- 14-
carbonaceous (B), bituminous (C) and.lignitious (D) according to their
carbon content. A simplified version of Seyler's coal chart, showing the
position of these coals in terms of their carbon and hydrogen contents is
given in Fig. 2. More complicated versions include sub-axes for calorific
value, and volatile matter and oxygen contents. Drawn on the figure are
lines linking those coals having B.S. swelling numbers of 4, 6 and 8.
Coals with swelling numbers less than about 4 are unsuitable for
metallurgical coke production.
Primarily because it gave no sound indication of the swelling and fusing
properties of coals so important to the coking industry, Seyler's
classification never achieved practical acceptance. The N.C.B. coal
classification system now used in the U.K. is based on two criteria, coal
rank and agglutinating properties. Rank is assessed by the volatile
matter content, on a dry, mineral-matter-free basis, and the agglutinating
properties by the Gray-King assay. The latter involves the heating, at
5K/min, of a horizontal tube half-filled with powdered coal, or for a
highly swelling coal a mixture of coal and sand, and comparing the coke
produced with standard coke types. Both being empirical tests,
repeatability is dependent upon the strict adherence to standard test
conditions [40],
The N .C.B. classification system is illustrated diagrammatically in Fig. 3.
Unless heat-altered, U.K. coals fall within the numbered rectangles. The
coals of primary interest to the blast-furnace cokemaker are the prime
coking coals in classes 301a and 301b, the coking-steam coals in class
204, and the. high-volatile caking coals in classes 401/2 to 601/2, ie. the
coals which show the strongest agglutinating properties.
Additional information on the suitability of coals for coking is obtained
from dilatometry and plastometry. Within the U.K., the accepted
dilatometer test [40] is based on the Ruhr dilatometer, a development of
the Audibert-Arnu version. In this, a pencil of compressed coal is heated
- 15-
at 3K/min while a plunger, resting on the pencil, indicates changes in
pencil length as a function of temperature.
Usually, an initial apparent contraction of about 30% of the original
pencil length.occurs as the pencil deforms to fill the width of the
dilatometer tube. Thereafter, except for non-swelling coals, dilatation
occurs reaching a maximum value when the plunger movement ceases. The
percentage dilatation normally quoted is measured from the original
position of the plunger. The total dilatation is the sum of the
contraction and the dilatation. It is considered necessary for blends for
coking to have a total dilatation in the range 50-150% [321, lower values
resulting in an inadequately-fused structure and higher ones a highly
porous coke.
An alternative dilatometer, the Chevenard-Joumier high-temperature
dilatometer, permits measurement of the contraction of the semi-coke
pencil after resolidification. The rate of contraction~temperature curve
shows two maxima, one near the resolidification temperature, the other
near 700·C. These are associated respectively with the end of active
decomposition and the liberation of hydrogen [411. As will be seen,
contraction behaviour plays an important role in the formation of
fissures in coke.
Most plastometers used in the U.K. are based on the A.S.T.M. version of
the Gieseler constant-torque plastometer. In this, powdered coal is
compacted into a cylindrical capsule equipped with a stirrer bearing
rabble arms. The capsule is heated at 3K/min while the rate of rotation
of the stirrer is recorded. The fluidity, recorded in terms of angular
rotation of the stirrer, rises to a maximum before falling to zero at the
resolidification temperature. Apparent viscosity values can be obtained by
calibrating the system with liquids of known viscosity.
- 16-
As Fig. 4 shows, coals soften and dilate in the temperature range of
active~decomposition" ie ... when :the~rate of. vola-tJle~matter release_is.~,
high. With increasing rank, as Fig. 5 shows, the maximum rate of volatile
matter release progressively decreases while the temperature at which it
occurs increases [381. In contrast, the dilatation and fluidity attain
maximum values in the middle of the rank range . Those macerals which do
not soften and dilate, eg., fusinite, micrinite, ·etc, ·are .termed inert .. Of.
the fusing, reactive macerals, exinites are generally higher in volatile
matter content, dilate to a greater extent and exhibit higher Gieseler
fluidity than corresponding vitrinites.
It is now evident that the chemical composition and other chemically
dominated properties of coals change progressively with coal rank, while
swelling and agglutinating properties, only occur in coals in the centre
of the rank range. Thus, although it is possible, from empirical
relationships, to identify coking coals from their chemical analysis, no
direct linear relationship exists between chemical composition and coking
propensity. Also, the chemical nature of the solvent-extractable fusing
component of a heated coking coal is believed to be similar to that of
the parent coal and thus to the non-fusing residue [42], Thus, the
difference between coking and non-coking coals is considered to be
essentially physical in nature, with chemistry playing a role in the
formation of materials with the appropriate physical properties.
2.2.3 The coal to coke transformation
The mechanism of the transformation of powdered; compacted coal into
fused, porous coke has been demonstrated most clearly using small,
single-Vlall ovens [43,441. In such studies, a temperature gradient is
established in a coal charge and" after cooling, the relative positions of
the products are fixed using epoxy resin. Sections can then be removed
and polished for' microscopiC examination.
- 17-
The quantitative information on the development of the porous coke
structure obtained in one such study [44] is shown in Fig. 4. The process
took place essentially within the plastic temperature range, three stages
being evident. Initial pore formation within larger coal particles led to
their swelling, enhanced interparticle contact and to fusion. Growth of
pores to maximum size near the Ruhr dilatometer temperature of maximum
dilatation then occurred. Finally, in a compact ion stage, complete near
the Gieseler temperature of resolidification, the pore size decreased to
the value found in the semi-coke. The necessary criteria for complete
fusion was found to be that the volume of the closed, spherical pores
within the plastic layer should exceed the original volume of the voids
in the charge. This conforms with the view that a total dilatation
greater than 50% is needed for coke formation. The compaction process is
obviously important in determining the coke pore size, and thus, as
explained later, the coke strength, but the factors controlling this
process are not fully understood.
Although indicating the mechanism of coking, such small-scale studies
provide no information on some important aspects of coal carbonization,
for example, the possibility of dangerous wall pressures being developed
during coking, and the size and quality of coke any particular coal blend
will produce.
During the coke-oven heating cycle, isothermal surfaces, parallel to the
oven walls, move progressively towards the oven centre, the various
stages of the coking process thereby occuring layer by layer [41]. Thus a
layer of plastic coal travels through the charge leaving behind it a
layer of semi-coke. This has two important e~cts. Firstly, the swelling ~
of the plastic layer can result in a pressure being exerted on the ~
oven walls. For most coals, the effect of plastic layer swelling is offset
by the contraction of the send-coke so that the wall pressure stays·
within acceptable limits. However, for some coals with volatile matter
contents near 20 wt% ( dmmf ), the wall pressure can be high enough to
- 18-
cause danger of wall damage. There has been some discussion whether high
swelling or low contraction plays the most important role in the
development of these high wall pressures but there is an association of
high wall pressures and unshrunken coke, difficult to push from the oven.
Another result of plastic layer swelling and semi-coke contraction is the
setting up of alternating compression and tensile forces within the coke
( Fig. 6 ) [14]. When the associated strains exceed the breakage strain
fissures are formed. It is considered that the high rate of contraction
observed in the high-temperature dilatometer near the resolidification
temperature is associated with the main fissure network formed within the
coke charge and that this controls the size of the coke obtained. The
second contraction peak has been related to a secondary fissure network
important in considerations of coke impact strength [41l.
The quality of coke produced in a coke oven is dependent upon many
factors, oven conditions, ego wall temperature, final centre-charge
temperature, soaking time, etc., charge conditions, ego particle size
distribution, charge density, temperature of blend charged, etc. and the
composition of the coal blend charged to the oven. Only the latter factor
will be considered here.
The purposes cif blending are manifold, to incorporate cheap coals, to
achieve a charge volatile matter content within the capacity of the by
product plant, to alter the impurity content of the coke, to avoid
dangerous oven-wall pressures, but primarily to achieve the necessary
coke quality. Where possible, only the prime coking coals were originally
used for coking but, as has occurred in the U.K., as their supply
diminished, maintainence of appropriate coke quality, using indigenous
coals, depe·nded on finding a suitable coal blend as a replacement. For
blast-furnace coke, where the important quality criteria is coke strength,
the aim of blending is to emulate the swelling and contraction behaviour
of prime coking coals. Highly-swelling, high-volatile coals show high
- 19-
rates of semi-coke contraction and thus tend to give small, highly
porous, weak cokes. Low-volatile coking-steam coals may etther induce,
high wall-pressures or, due to inadequate fusibility, produce readily
abradable cokes. Fortunately, blending the two tends to ameliorate their
adverse properties and cokes of size and strength suitable for use in
medium-sized blast furnaces can be obtained [45], However, to achieve the
higher coke quality specified for larger, harder-driven, modern blast
furnaces a proportion of prime-coking coal is incorporated into the
blend. Such additions are especially helpful, as a bridging coal, where
lack of mutual fusibility between the high- and low-volatile coals is
suspected [45], Pitch additions may also be used to combat inadequate
blend fusibility [18], Additions of inert non-fusing'materials;',eg. coke
breeze, some low-volatile coals, and petroleum coke, curbs excessive
swelling. Such additions also have another effect important in foundry
coke production. By reducing excessive contraction rates, they act as
anti-fissuring agents [15] and facilitate the production of coke large
enough for foundry use.
The choice of a suitable blend for, coking clearly involves many options.
Even when the blend components have been selected, their relative
proportions still need to be determined. This can be achieved either by
an exhaustive testing programme or, if a suitable method is available, by
calculating the effect on coke quality of blend composition variations.
- 20-
Between the coke-oven wharf and the blast-furnace bell, coke suffers some
size reduction due to mechanical stresses during 'handling. In the blast
furnace itself, the compressive stresses imposed on the coke by the
, overburden. are insufficient in, themselves to cause .fracture [71.
Nevertheless, significant size reduction does occur between bell and
tuyeres, and only some of this occurs as coke falls from bell to
stockl1ne [461. As was explained in the previous section, it is now
recognised that excessive size degradation within the blast-furnace shaft
reduces the permeability of the stack and adversely influences furnace'
output. The strength of coke is important in so far as it influences size
degradation. Within the blast furnace, oxidation and exposure to high
temperatures contribute to the size reduction. Nevertheless, coke strength
is routinely assessed at ambient temperature.
Although the present drum tests for assessing the strength of coke
involve subjecting large coke pieces to mechanical treatment and
expressing the 'coke strength' in terms of indices based on the reduction
in lump size, they were not devised from a sophisticated view of the role
of coke in a blast furnace. Nevertheless, correlations between blast
furnace output and drum indices have been reported [61. Although, within
the coking industry, drum tests are commonly referred to as strength
tests their principal use is not to measure the strength of coke per se
but to monitor variations in coke quality. All drum tests avoid the
necessity of producing mechanical test pieces of specific size and shape
in sufficient quantity to reflect the inhomogeneity of commercially
produced metallurgical coke.
At least on the research level, two further tests of coke strength have
gained measures of acceptance in the U.K. The adoption of a more
fundamental approach to coke strength resulted in the application of the
- 21 -
diametral compressive test as a measure of the tensile strength of
fissure-free coke [471, while a 'microstrength' test has been used to
assess the strength of small coke particles free from visible pores [481.
In this section of the review, after considering a simple treatment of
relevant fundamental aspects of the fracture of brittle materials, these
various coke strength tests are described and their relation one to
another discussed.
2.3.1 The fracture of brittle materials
The carbon in metallurgical cokes can be regarded as being organized into
small, defective, randomly orientated graphitic crystallites [49], This
structure permits neither the dislocation movement which confers
plasticity nor the extension of coiled molecules which confers
viscoelastic behaviour. Metallurgical coke is thus a stiff, brittle
material. Despite the mode of applied stress, breakage is therefore
considered to occur as a result of induced tensile forces [81.
Brittle failure of metallurgical coke has been discussed in terms of
simple flaw theory [101 according to which it is considered that once a
flaw of critical size begins to propagate i~ will continue to do so until
failure occurs. Crack initiation at flaws in brittle materials depends on
the fulfilment of two conditions. Firstly, the local stress level at the
flaw must exceed the theoretical strength of the material, and secondly,
the energy available must exceed the surface energy of the two surfaces
created during crack propagation [501.
The concept of stress concentration necessary to fulfil the first
condition was originally developed by Inglis [511 to explain the failure,
at stresses well below the theoretical strength, of plates bearing holes
or hatches. Stress analysis of an elliptical hole in a uniformly stressed
- 22-
plate showed that the stress concentration factor, i.e. the ratio of the
local stress to_the applied stress, was
er lero= ( 1 + 2c/b ) <1>
where c and b are the major and minor semi-axes of the ellipse, and er
and ero are the local and applied stresses. Thus, the stress
concentration factor depends on the shape of the hole and not its size.
For a narrow ellipse, 1.e a crack, where c >>> b, then the equation
reduces to :
rrlrro= 2( clp )"2 <2>
where p is the radius of curvature at the crack tip.
Although these studies made a step forward in the understanding of
fracture, they failed to explain why, in practice, large cracks propagate
more readily than small ones. Yet they did provide the key to
understanding the difference between theoretical and practical strengths
in terms of the behaviour of flaws.
This view was developed further by Griffiths in two classic papers
[52,531. He recognised the importance of the stress raising capacity of
existing flaws and, considering the energetics of the process, deduced
that crack propagation occurred when the strain energy available exceeded
the surface energy of the two freshly created surfaces. Since ·strain
energy is proportional to the square of the crack length while the
surface energy is directly proportional to it, it follows that above a
critical crack size there is more strain energy available than is required
for creation of fresh surface. Thus, the propagation of a critically-sized
flaw leads inevitably and directly to failure. It was deduced for plane
stress that the failure condition was
er,= ( 2E~h(c )"2 <3>
- 23-
where 0', is the failure stress, E the Young's Modulus, ~ the surface
energy and 2c the critical crack size. Thus the adverse effect of large
flaws was explained.
An equation of similar form can be deduced by combining equation <1>
with Orowan's estimate (501 of a material's theoretical strength i.e. :
where a is the interatomic spacing. Equating 2p with a gives
0',= ( E~/8c ) 1/2 <5>
where 2c is again the critical crack size. ·This defines the criterion for
failure obtained from considerations of the magnitude of the local stress
at the crack tip.
Since, according to this second criterion, failure would occur at a lower
stress than that deduced by application of the Griffiths concept
< equation <3> ), it has been concluded that, for brittle materials, the
Griffiths criterion is both a necessary and sufficient criterion for
failure (501.
2.3.2 The influence of porosity
. Since metallurgical coke, as well as being brittle, is also highly porous,
it is relevant to consider the influence of porosity on the strength of
other brittle materials.
The variation of the strength of ceramics with porosity was appreciated
before being precisely stated by the empirically-derived Ryshkewitch
Duckworth equation (541:
S= So exp< -bp) <6)
/
- 24-
where S is the strength of a porous polycrystalline body, So that of a
. _similar_ non,,-.porous body,. _p .isthe _fractional _porosity .and b is a
constant.
The form of this equation was explained by Knudsen [55) from theoretical
consideration of a material composed of sintered spheres. He assumed that
-the variation· of strength with porosity reflected changes in the load
bearing area within the specimen, the critical area being that traversed
by an irregular cross-sectional surface passing through the areas of
contact between the spheres.
Knudsen further drew together the Ryshkewitch-Duckworth equation and an
earlier equation of Orowan [56) relating strength with grain size i.e.
S= kG-l/", (7)
to derive an equation involving both grain size and porosity
S= SoG-l/2 expC -bp) (8)
The variation of strength with grain size is consistent with Griffiths
views provided either the critical flaw size is equal to the grain size or
that the stress to propagate a crack within a grain is less than that
required to propagate it across a grain boundary [55]. The equation was
tested for thoria specimens fired to various temperatures in the range
1650-1850·C. The observed decrease in b with increasing firing
temperature was ascribed to spheroidization of the pores as sintering
proceeded [55). Studies of the application of this approach to
metallurgical coke will be reviewed later.
2.3.3 Drum tests
The aim of the designers of drum tests was to replace the laborious
successive drops in the earlier shatter test by mechanical means [7).
Smooth drums produced little size degradation so that either flighted
drums or drums. whose cylindrical surfaces were composed of steel bars
- 25-
were introduced. A barred drum still forms the basis of a Russian test
but in the western "world Bighted-drums are the more popular [71. Three
basic designs of flighted drums are in common use, the A.S.T.M., the Micum
and the J .1.S. ( Japanese standard ) drums. Details of these drums and the
standard mode of operation [57J are given in Table 2.
Several variations of the standard Micum test are in use. In" the U.K., the
half-Micum drum, 500mm long, is used to test a 25kg coke sample. This
drum is also used in the extended Micum test when the coke is sized at
100 revolution intervals up to 1000 revolutions, all the sample being
returned to the drum after each sizing. The reciprocal of the square of
the coke mean size is then plotted against the number of revolutions and
the 'Micum slope' obtained from the slope of the linear portion of the
line evident at high revolution values. In the 1.R.S.I.D. variation of the
Micum test, 20-80mm coke, a size range commonly charged to blast
furnaces, is given 500 revolutions in the Micum drum and the coke quality
is then expressed in terms of the percentage by weight of the product
retained on a 20mm sieve.
In general principle, the three drum tests are similar in that in each a
drum, charged with a specific weight of lump coke, is rotated at fixed
speed for a set number of revolutions and the 'coke strength' then
assessed in terms of indices based on the size analysis of the pr~duct.
However, as Table 2 indicates, the three tests differ in all important
respects ie. the drum dimensions, the number of revolutions, the nature of
the test sample and the sieve size used to assess the strength indices.
These variations in test parameters result in differing mechanisms of
coke size reduction [57). In the J .1.S. test, the flights are large enough
individually to accomodate the entire coke charge. Therefore, as the drum
rotates, the coke lumps are picked up on, and dropped from, the flights.
The predominant mode of size reduction therefore involves shatter and, as
a result, the coke pieces tend to remain sharp edged. In contrast, the
- 26-
flights in the A.S.T.M. drum, being too small to lift coke lumps, merely
induce a rotational motion. Thus abrasion is the primary mode of size
degradation, the tested coke consisting of rounded lumps with much small
debris. The appearance of the coke subjected to the Micum testing
suggests a mechanism of size degradation involving both shatter and
abrasion.
Detailed studies of the behaviour of individual coke lumps rotated in a
half Micum drum [58] show that breakage of large coke pieces can be
described in terms of volume breakage, ie. the breakage of a large lump
into two or three smaller pieces, and surface breakage which produces a
small product less than 5mm in size. Surface breakage arises either as a
result of high local stresses generated on impact of one coke piece on
another or as a result of contact stresses due to the tumbling motion of
the coke charge. In the Micum test, the two forms of surface breakage are
considered to contribute equally to the formation of the material less
than 10mm in size and therefore to the M10 index. The volume breakage of
coke depends upon the impact stress level and the length of pre-existing
fissures. Stresses less than a threshold level have no effect. Repeated
stressing at higher levels induces the incremental propagation of
fissures on each impact until a critical crack length is attained when
immediate propagation to failure occurs. Coke lumps with pre-existing
fissures greater than a critical size fail on initial impact. The critical
size is estimated to be about 20mm [58],
2.3.4 The tensile strength of coke
The measurement of the tensile strength of coke by the diametral
compression method was originally introduced by workers at the British
Carbonization Research Association ( B.C.R.A. ) to obtain a measure of the
strength of coke, more fundamental than that provided by Micum testing,
to help in the understanding of the fissuring of coke in a coke oven.
Recourse was made to a test previously used extensively to measure the
- 27-
strength of other materials difficult to fabricate into the usual
dumbbell-shaped tensile test piece, ego concrete (59] and rock (60],
The test is based upon the stresses developed when a cylindrical specimen
is loaded along a diameter ( Fig. 7a ). Ideal line loading produces a
biaxial stress distribution within the specimen (61], the maximum tensile
stresses acting normal to the loaded diameter and having the constant
magnitude ( Fig. 7b )
O't= 2y/ IrrDt <9)
where Y/ is the load and D and t the specimen diameter and thickness
respectively. For valid tensile results, the stress distribution should
approach the ideal and the fracture be initiated by the induced tensile
forces. Over the central portion of the diametral plane, the theoretical
compressive stress is 6y//rrDt so that here flaw propagation by the
induced tensile stress is expected. At the loading points, the compressive
stress in theory rises to infinitely high values but, in practice, real
loading fixtures distribute the load over an area and this has the effect
of reducing the high compressive forces near the ends of the loaded
diameter. Local edge crUShing, often encountered near the platens as a
result of the high compressive forces, tend only to increase the area of
the applied load. However, failure of the specimen may still result from
either tension or shear. Failure due to shear stresses results in cracks
intersecting the loaded diameter at a high angle. Thus, a diametral
fracture is indicative of a valid tensile test. A typical fracture is
illustrated in Fig. 7c.
No officially-recognised standard procedure for the diametral-compression
measurement of the tensile strength of coke has been established.
However, for many years, the test was used extensively at the B.C.R.A.
There, the initial application in connection with fissuring of coke in a
coke oven was overtaken by its use as an additional indicator of coke
- 28-
quality particularly useful, as discussed later, in the development of
coal blends suitable for the production of high quality coke ..
The test procedure most frequently adopted (81 involved the use of fifty
lOmm diameter by lOmm long cylindrical coke specimens. To obtain
representative strength values from large samples of industrial cokes,
these were cut from numerous coke lumps, carefully chosen by sample
division techniques. Strength measurements were made using a universal
testing machine with a crosshead speed of O.5mm/min. Below a critical
level, the rate of loading has no marked effect on the strength value
obtained but increasing the specimen size leads to a progressive decrease
in the measured strength (621.
Reported values for the tensile strength of coke lie in the range 2.5-7
MPa but the extent to which these values represent valid tensile
strengths is not certain. Not all cokes fulfil the theoretical requirement
of being a homogeneous elastic material and for some cokes marked
deviations from the theoretical stress distribution have been reported.
For this reason, it has been suggested that coke tensile strengths should
be regarded primarily as comparative, not absolute, values (471. Also,
experience suggests that some weak cokes, ie. those with strengths less
than about 3.0MPa, do not simply fail in tension since no diametral
fracture is observed. Nevertheless, the test does give strength values
acceptably reproducible for a material as heterogeneous as coke. Also,
unlike the Micum test, it enables the same procedure to be applied to
cokes made in very different scales of oven thus permitting their direct
comparison. The coke tensile strength has therefore proved a very useful
additional parameter for assessing the strength of coke.
2.3.5 The microstrength test
This test was originally devised in an attempt to obtain a measure of the
mechanical properties of coke produced in a small sole-heated oven (481.
- 29 -
In the test, a 2g sample of coke, sized 600-1000~m, is charged to a
---stainless .steel_tube,300mm. long by _25mmdiameter, _togetheLwJth. twelve
Bmm diameter steel balls. The tube is then rotated, end over end, at
25rpm for BOO revolutions. The microstrength is quoted in terms of two
indices based on the percentage by weight of the product greater than
600 and 212~m. Despite the original claim that the test gave a measure of
those fundamental mechanical properties of coke which·- influence its
mechanical strength, within the coking industry the test has found little
application other than for laboratory cokes.
The earlier discussion of the fracture of brittle materials linked a
materials strength with the nature of the flaws present imd the stress
concentration they induce. In the Micum test, the critical size of
fissures which govern the volumetric breakage, and thus the M40 index, is
about 20mm. ( The nature of the flaws controlling the production of small
debris and thus the M10 index is not clear ). Although the size of the
critical flaws in the tensile strength test has not been directly
measured, from studies of the relationship between tensile strength and
pore structural parameters, it has been inferred (10) that the larger
pores, sized about 400-1000~m, are the critical flaws. No suggestions
regarding the nature and size of flaws controlling size degradation in
the microstrength test have been put forward but the coke particle size
used precludes the presence of large pores. It is evident, therefore, that
the flaws controlling the 'strengths' measured in the three tests differ
markedly in both nature and size. For this reason, it would be unrealistic
to expect a simple relation to exist between the test results and, indeed,
no generally applicable relationships have been reported.
- 30-
2.4 The texture of metallurgical coke
The term texture is used to refer to the appearance of metallurgical coke
surfaces when viewed microscopically, the texture reflecting the variety
of form of the structural units present. The type of microscope used to
examine the texture is identified by using the terms PLM texture or SEM
texture. Polarized-light microscopy has frequently been used to study coke
texture since, in common with cokes from most other precursors which
soften on carbonization, metallurgical cokes exhibit optical anisotropy. A
brief explanation of this effect, based on two standard texts [63,64], and
its application to carbons is given before discussing further the
development of the texture of cokes.
2.4.1 Polarized-light microscopy as applied to carbons
When ordinary light passes through a polarizing filter or prism, its
components at right angles become out-of~phase, linear polarization
resulting when the phase difference is a multiple of one-half wavelength
( >.12 ). Such light progresses by vibration in one direction. Circularly
polarized light, progressing in a spiral, results when the phase
difference is an odd multiple of A/4. Other phase differences give
elliptically polarized light.
Optical properties of crystals are described in terms of reflectance,
refractive index and absorption index. For present purposes, reflectance
has the most relevence. The optical properties of isotropic crystals have
the same values in all directions, but if a crystal surface is anisotropic,
it will have at least two values for each property, ie. it exhibits /
bireflectance, birefringence and bisorbance.
The optical axis of a birefringent crystal corresponds to an axis of
structural symmetry such that a section normal to the axis is isotropic.
I .
- 31 -
A crystal with only one optical axis is described as uniaxial. When an
anistropic crystal is exposed to linearly polarized light along a
direction perpendicular to the optical axis, the optical properties are
maximised when the vibration direction is either parallel or
perpendicular to the optical axis. Such crystals are, by definition,
positive or negative. The direction of maximum properties is defined as
the slow direction. J' . /' - ~ §:~ .. ,' ~/: .'
The stacked layer planes of the graphite lattice are illustrated in Fig.
8. The optical axis lies along the single axis of symmetry, in the Z
direction, and the basal layers, in X-Y planes, are optically isotropic.
Linearly-polarized light is reflected from the prismatic edges with
maximum intensity when the vibration direction is parallel to the layer
planes and perpendicular to the optical axis. According to the
conventions given in the preceding paragraph, graphite is therefore a
uniaxial negative crystal.
Incident-light, polarizing microscopes are commonly used with crossed
polars, ie. with two polarizing filters, the polarizer and analyser in
incident and reflected beams respectively, set at right angles to each
other. The following remarks refer initially to effects observed under
these conditions when no retarder plate is used.
,
I/under these conditions, isotropic-crystal surfaces appear grey whatever
\1 their orientation on the microscope stage. Anisotropic-crystal surfaces
I however exhibit a variation in shading depending on the orientation of
the slow direction of the crystal relative to the vibration direction of
the incident light. If the slow direction is perpendicular to the
(
vibration direction, minimum reflection occurs. If parallel, reflected
light is blocked by the analyser filter in the reflected light path.
I, Crystals so orientated thus appear dark while those aligned at 45· to the
vibration direction appear bright. Intermediate alignments result in grey
shades.
- 32-
The variation in shading observed when a polished surface of an elongated
piece of petroleum needle coke was rotated under polarized light with
crossed polars is illustrated in Fig. 9a, the vibration direction of the
incident ,linearly polarized light being as shown. The angles indicated
refe~ to the orientation of the longest dimension of the coke particle '
relative to the vibration direction. When the longest dimension was
parallel or perpendicular to the vibration direction it appeared dark.
Thus the slow direction of the structure, ie. the prismatic edges were
aligned parallel to the longest dimension.
From this variation of shading with orientation of the carbon layer
planes relative to the vibration direction of incident polarized light, it
can be deduced that the appearance of a carbon composed of small
graphitic areas, randomly orientated with respect to one another, will be
as illustrated in Fig. 9b. Within each area, drawn square, the direction
of the parallel constituent layers is indicated by the dashed line. The
shading applied to each area corresponds to that of similarly-aligned
layers in Fig. 9a. The resultant mottled appearance gives rise to the term
mosaics [651.
In carbons from aromatic pitches, the layer planes are more extensive but
exhibit a variety of faults. A doubly folded structure is illustrated in
Fig. 9c, the dashed lines representing the carbon layers. Over most of the
area these layers lie at 45' to the vibration direction and, as Fig. 9a
indicates, will appear light. However, for each line drawn, a dot on the
line indicates where the layer lies perpendicular to the vibration
direction. This position will therefore appear dark under crossed polars.
The' total effect will be that two black lines, linking the dots along
A--A and B--B, will be visible against a light background. Such lines are
termed extinction contours [66],
Pleochroistic or reflection interference colours can be introduced into
images obtained in a polarizing microscope by using a retarder plate, the
- 33-
thickness of which is so chosen that a phase difference exists for
_____ --linearly=polarized--light--passing _ in -the _fast_and slow _directions. __ Either a
X-plate or a X/4-plate is used with crossed polars and a X/2-plate with
parallel polars. The retarder plate is usually positioned so that its fast
and slow directions lie at 45· to the polarizeI'.
-- When -linearly-polarized light-is -reflected -fromei ther -an -isotropic
surface or an anisotropic surface aligned with either its fast or slow
direction parallel to the polarizeI', the reflected beam is similarly
polarized. On passing through the retarder, due to the plates
birefringence, the emergent beam is elliptically polarized. When these
rays are resolved into a single plane by the analyser, destructive
interference of yellow light occurs so that the surface appears purple.
For anisotropic crystals, the purple is light when the slow direction of
the crystal is aligned parallel to the incident polarization and dark
when the crystal is rotated through 90·.
Other orientations of the anisotropic surface result in an elliptically
polarized, reflected-light beam. The phase difference of the orthogonal
rays is then superimposed onto that induced by the retarder plate. If the
anisotropic section has a slow direction parallel to the slow direction
of the retarder then the phase difference resolved by the analyser is
increased. This results in destructive interference at the red end of the
spectrum so that the surface appears blue. On rotation of the surface
through 90·, destructive interference of the blue end results in the
surface appearing yellow or orange. These colours, however, are very
sensitive to the orientation of the retarder plate. Minor departures from
the ideal 45· angle can change the colours markedly.
For the three examples discussed earlier, the -effect of using a X-retarder
plate with crossed polars is shown in Fig. lOa-c. The result is that
varying orientations of prismatic edges are now indicated by variations
in tint rather than shading. Thus, when the surface of a carbon or
- 34-
graphite is viewed under these conditions, each isochromatic area visible
is composed of aligned layer planes, the colour being dependent upon
their common orientation relative to the polarizer and analyser, However,
. under high contrast conditions, extinction contours can 0' ,',,,/ ,-' , ,','J /~·tkj. ...... -... '- Jf-R<f!""C4 ;/.~rt:,. ......... ',' a"", ... rc·,-.J.~ ..{A.< 'L ... "", '
observed. ..L",. /. ! 'J " .. , '. ./., ( . . , _ .. ~ //.......---1 L 'If\\..t'-'''
2.4.2 Early studies of coke texture
f;b..A~
still be readily .. ~~_,--"":-C"ti..-': t:~r[
Although earlier studies [67,681 had revealed the presence of anisotropic
constituents in metamorphosed coals and metallurgical cokes,~~he present
understanding of the origin of the texture of cokes and other
graphitizing carbons stemmed from the study. by Taylor of a heat-affected
coal seam in New South Wales [661. As the igneous intrusion was
\ approached, progressively larger optically-anisotropic spheres were
\ observed in the isotropic vitrinite matrix. The structure shown in Fig, 11
was one of two proposed to account for the variation in appearence of the •
spheres when the specimen stage was rotated~Sim11ar spherical features,
although smaller in size, were-observed wh~unalter~d coal was heated to /' .
temperatures in the plastic temperature range [661. When the isotropic
" vitrinite was completely eliminated the mosaic coke structure was fully
developed, the whole process being completed within the temperature range
460-490 ·C.
Following this pioneering study, investigations were extended, in the
early nineteen-sixties, by Brooks and Taylor [691 to the study of many
carbonaceous feedstocks. The~e findings opened a new phase in
carbonization research which attracted many workers, notable
contributions to the literature being made by White [70,711 for his
studies of petroleum residues, and by Marsh [72,731 for his wide-ranging
studies of various materials including coals. From such studies, it is now
evident that the carbonization behaviour of many materials which form
graphitizing carbons falls within a general pattern. It is proposed to
CJ1,"'/" ";i(.r-. ~ L 74-, 11/"'/- .-1- _ I _
}1- 35 - . !.,~.IA'" '1',' c(>L /L" A Cl, f~, " <' . . 1(' , IY v ~ , _ '" '''-'. '~, ,1-
~:;, I. 7,lc" ~-;;£-. ::"',-1 " '."', _6< / .. , • "'~~'-J.- .••• - t.~-·I;I'~··:'> ' ...
review this behaviour before consideriAg further the development of the
anisotropic texture of metallurgical cokes.
2.4.3 The formation of graphitizing carbons
Studies of the formation of graphitizing carbons have usually involved
heating the chosen precursor at constant rate to selected temperatures
within the plastic range, cooling, and examining the product after
embedding in resin and polishing [741. However for carbonizing pitches,
similar results have been obtained spectacularly by combining cine
photography with hot-stage microscopy [751. Materials reported to conform
to the general pattern of behavour include coal-tar pitches [69,731,
petroleum tars and residues [701, solvent-refined coals [731, some
vitrains [661, some polymers ego poly-vinyl chloride [69,731, and poly
aromatic compounds ego anthracene and phenanthrene [731. Although quite
different in character, all these materials decompose during pyrolysis to
give optically-isotropic pitch-like materials which are fluid at elevated
temperaturesvl731. The chemica~vari~ty of precursors alsD suggests that ~ OA,;''O - ;/0 C Co
mesophase formation is not sensitive to details of·the structure of the
constituent molecules [731. - .-'
The chemistry of the liquid phase pyrolysis of such materials is complex
[761 but is considered ·to involve the' initial elimination of aliphatic
side chains. Polymerization reactions ensue so that, with increasing
temperature, products of increasing size, molecular weight and aromaticity
are produced [731. Eventually, the concentration of large, approximately
1000 amu, essentially planar, aromatic molecules exceeds a critical level
and, by a process of homogeneous nucleation, these planar molecules
assembie together to form, within the isotropic pitch, spherical,
optically~anisotropic, nematic liquid crystals, in which the planar
molecules are stacked parallel to one another [731. These are detectable
under polarized light as pinpricks of light when their size exceeds about
! 0.21'm [661. Large Van der Waals forces and dipole interactions stabilize
- 36-
the liquid crystal [731. Further chemical reaction, polymerization and
_ cross linkage formation, within the liquid crystals leads irreversibly to
their transformation into fluid mesophase spheres [73). The variation of
electron diffraction patterns across ultra-thin sections of spheres - - -
confirms that the constituent lamellar molecules, as shown in Fig. 11, are - ---
aIign~d basically parallel t~ ~quitoria~ plane but curvft to meet the ----pitch -interface with high angle [771. In contrast, non-graphitizing ---- - - --.- -
'Carbci~are formed either from non-fusing precursors or from materials
of such high chemical reactivity that polymerization and cross-linkage
reactions cause resolidification before liquid crystals can develop [73).
With increasing time and temperature, existing spheres increase in size
and more spheres nucleate [781. As the proportion of the isotropic phase
diminishes, continued sphere growth inevitably leads to sphere contact
and coalescence, the latter process being assisted in certain instances
by bubble percolation [79). The processes of sphere nucleation and
coalescence, and the subsequent deformation of the coalesced bulk
mesophase determine the size and shape of the anisotropic structures in
the resultant carbon.
Two extremes of behaviour have been described [80):-
1. Highly aromatic pitches ( high temperature coal-tar pitches and
petroleum pitches ) contain slow reacting components which
precipitate mesophase spheres under severe pyrolysis conditions
( above 460·C [74) ). Slow sphere nucleation permits their
growth to large sizes before coalescence. Low viscosity
then facilitates the rapid reimposition of sphericity and ready
realignment of the constituent lamellar molecules. Complete
coalescence, on elimination of the isotropic phase, results in
a coarse mosaic plastic mesophase which may undergo
deformation by mechanical means, bubble percolation or
convection currents. Uniaxial and biaxial tensile deformation are
considered to result in lamellar or plate-like structures [74).
- 37-
Deformation above 460'C is considered essential for the formation
of needle coke.
2. Less aromatic pitches ( petroleum bitumens and low-temperature coal
tar pitches ) contain more-rapidly reacting components. Thus ready
precipitation of nUmerous anistropropic spheres occurs under less
severe pyrolysis conditions ( under 430'C [74] ). Little growth
occurs before complete coalescence. Low viscocity then restricts
molecular realignment so that a fine mosaic of small, randomly
aligned m'icroconstituents is formed. Resolidification follows
before much deformation of the structure can occur,
Cokes may-pe characterized according to their extinction contour spacing
~~e.size of the isochromatic areas present. The lamelliform morphology
of pitch cokes has been traced by study of their extinction contours
[65], black lines defining the loci of points where the layer planes lie
either parallel or perpendicular to the plane of polarization of the
\ incident light.lBe;ds, splays, folds and lamellar stacking defects are
~d but nothing resembling a' grain boundary can be detected [74J.
. Alignment of lamellae circumferentially to pore surfaces occurs [71l,' F
,~ine non-fusing constituents, ego quinoline-insoluble or carbon black ,~- -- ---.-- -particles congregate at the periphery of mesophase spheres and can -.- -
interfere with their coalescence [69]. The presence of certain foreign
\
' r
atoms, ego sulphur and oxygen, in the pitch leads to lower viscocity and
I small mosaic structures in the coke [81,82J.
2.4.4 Development of the texture in metallurgical cokes
Apart from the original study by Taylor, the development of optical
anisotropy during the carbonization of coals at atmospheric pressure has
been studied in detail only byPatrick' etal [11 ,83,84]. Crushed sainplEis'
of vitrains, ( ie. concentrates of reactive vitrinites ) hand-picked from
single seam coals, were carbonized in open boats at 5K/min to selected
- 35-
temperatures covering the plastic temperature ranges of the coals and to
1000·C. Polished surfaces were examined using crossed polars and a
retarder plate. Results obtained confirmed the earlier reported variation
of the size of the anisotropic entities with coal rank [651. These
entities were allocated, according to size or shape, into various
categories termed fine, medium and coarse mosaics, sized approximately
0.3)lm, 0.7)lm and 1.3)lm respectively, granular-flow and flow-type.
Hicrographs illustrating the appearance of the various forms are
available [111. The term flow~type refers to components with extensively
elongated isochromatic areas, but is not necessarily associated with high
plasticity, detectable by Gieseler plastometry, during carbonization.
Although differentiation between components was acknowledged to be
subjective, by the application of point-counting techniques it proved
possible, both for semi-cokes formed within the plastic temperature range
and for laboratory cokes formed on heating to lOOO·C, to assess
quantitatively their anisotropic composition ie. the proportion of the
various anisotropic components present.
The data obtained from laboratory cokes prepared at lOOO·C from vitrains
of varying maximum reflectance ( Ro max ) are summa,ized in a novel
manner in Fig. 12. This shows that low-rank vitrains'remain isotropic on
heating, but as the reflectance increases from 0.5% to 1.46%, fine,
medium/coarse and granular-flow anisotropic units progressively become
evident in the cokes. The cokes obtained from the prime-coking-coal
vitrains, with reflectances in the range 1.25 to 1.46%, contain granular
flow units as the major component, but at least a small proportion of
material in all the other classes, including flow-type, is present. A
marked change occurs above a vitrinite reflectance of 1.46%. Higher-rank
vitrinites exhibit basic anisotropy in the unheated form and the
principal anisotropic component in the . cokes' ·is· flow-type, ·although about
20% by volume of granular-flow is present in cokes from vitrains with
vitrinite reflectances of 1.45 to 1.55%. As these higher rank vitrains
become progressively less reactive in a coking sense, an increasing
- 39-
proportion of material exhibiting basic anisotropy, unaltered in form
from that in the vitrinite, but more intensely coloured, becomes evident
in their cokes.
Examination of products formed at lower temperatures showed that the
development of anisotropic textures occurred within the expected plastic
temperature range ( no measurements were reported) (l1,83,84J. As the
rank of the vitrains exhibiting basic anisotropy decreased, the basic
anisotropy remained unchanged or was converted, either directly or via
the intermediate formation of a type of fine mosaic component, into flow
type anisotropic material. Isotropic vitrains from prime-coking coals,
having Ro max values in the range 1.25-1.46%, initially developed
considerable concentrations of fine mosaic components which were
converted at higher carbonization temperatures into coarser-grained
mosaics and flow-type structures, the concentration of the latter material
in the 1000'C cokes rising from approximately 5% to more than 70% by
volume as the reflectance of the vitrain decreased. Growth of fine-grained
mosaic units, initially formed, into medium-grained units was observed
for vitrains from the highly-fluid, high-volatile coals in N .C.B. classes
401 and 402. Vitrains from lower rank coals formed fine mosaic units or,
as their softening properties diminished, remained progressively
isotropic.
2.4.5 The mechanism of the development of coke texture
The studies of Patrick et al clearly demonstrate that the development of
optical anisotropy during coal carbonization occurs at temperatures where
plasticity could be expected. Nevertheless, in contrast to studies of an
Australian coal (66], at no stage in the carbonization of U.K. coals was
any clear evidence obtained of spherical.bodies as a precursor to mosaic
or larger anisotropic units (84]. However it was acknowledged that such
units could be present in a size range below the limit of resolution of
the optical microscope. Nevertheless a role by liquid-crystals was not
---- -------
- 40-
excluded and it was suggested that the degree of structural order, as
measured by optical microscopy, was dependent upon the nature of the
coal, ie. its aromaticity, the ordering of lamellar constituents and the
degree of cross-linkage between molecules. This was considered to
influence the balance attained during the plastic temperature range
between the loss of volatile matter required to enable the necessary
molecular rearrangements to take place and the retention of sufficient
plasticity, when appropriate molecular constituents are available, to
allow the formation of the 'liqUid-crystals' which form the optically
anisotropic bodies.
Coals of the highest rank, already having a high degree of structural
order, were considered to require relatively little change in the
transformation from basic to flow-type anisotropy, a process assisted by
volatile matter release [84J. Fine mosaic units observed during
carbonization of coals of slightly lower rank were associated with the
disruption of the original vitrinite structural order. Otherwise, the
process of texture development was enVisaged as a progressive
development of fine to coarser mosaics and to flow-type structures, the
extent of the process being dependent upon the distortion of the mosaic
units as a result of the balance between gas evolution and the viscosity
of the system.
Expressing views which differ in detail from those of Patrick et a1,
Marsh and co-workers [72,73,85] have discussed coal carbonization within
the general carbonization context. Marsh has not reported a study of the
texture of coals heat-treated at atmospheric pressure to temperatures
within their plastic temperature range. Nevertheless, coal carbonization
is described in terms of nematic liquid crystal development and
mesophase sphere growth and coalescence [72] .. High fluidity is considered
a major factor favouring the growth of the isochromatic units visible in
polished surfaces but the chemical reactivity of the constituents in the
plastic mass is also felt to play an important role [861. High reactivity,
- 41-
by inducing cross-linkage between molecules, leads to early
resolidification. Thus, high chemical reactivity and the presence of
heteroatoms are considered to be the factors responsible for the small
mosaic units observed in cokes from low-rank coals. These units were
shown to be polygonal, not spherical, and this was associated with a
limited ability of the mesophase spheres to coalesce [87]. Confirmation
that medium mosaic units, less than 5)1m in size, were merely compressed
t~gether was obtained by X-ray difraction methods [88].( As discussed
below, it is difficult to reconcile the sizes of mosaic units quoted by
different workers.) With increasing rank, as the aromatic character of the
coal increases, the reactivity is lower and higher fluidity is observed at
higher temperatures [89]. These are considered to be optimum conditions
for the development of liquid crystals. Thus the flow-type anisotropy of
coking coals is considered to result from the coalescence of small
mesophase units, the coalesced bulk mesophase retaining sufficient
mobility to respond to shear forces set up by thermal convection currents
and bubble percolation [89],
A reappraisal of Patrick's data suggests that granular-flow and flow-type
components are the final products of different development routes. Thus,
during the development of a coke composed predominantly of granular-flow
anisotropic units ( Fig. 13 ), the progressive rise and fall in the
concentration of fine, medium and coarse components, being reminiscent of
the variation in concentration of intermediates in a consecutive chemical
reaction scheme, indicate their role as intermediates in the formation of
granular-flow components. For the vitrains examined, there is no
consistent evidence, in terms of a rise and fall of concentration with
temperature, that granular-flow material acts as an intermediary in the
formation of flow-type structures. Furthermore, flow-type components were
often first observed at lower temperatures than granular-flow units (84].
Hence, it is here considered that flow-type components are formed from'
basic-anisotropic vitrinite either directly or via the intermediate
formation of a type of fine mosaic constituent. On this basis, when both
- 42-
granular-flow and flow components are produced from a coal, the vitrinite
present is considered to be heterogeneous to the extent of permiting the
two development routes to take place Simultaneously.
As regards the formation of mosaic-anisotropic textural components the
views of Patrick and Marsh can be reconciled as follows. Since isotropic
vi trains soften on carbonization and form graphitizing carbons, a
behaviour totally different from that described above for pitches seems
unlikely. Thus a mechanism involving the formation of liquid
• crystal/mesophase spheres as a precursor to mosaic anisotropy can be
inferred. That individual spheres are not observed can be due either to
their small size [84], or, by analogy with the behaviour of some polymers
[90], to their very rapid formation. The latter view is consistent with
the observation that initially-formed fine mosaic units exist in discrete
areas [91]. Such areas can be regarded as aggregates of mesophase spheres
simultaneously nucleated within a small volume of homogeneous vitrinite,
possibly the size of the vitrain particles carbonized. The gradual fall in /.3
concentration of isotropic material in Fig. )2 is explicable if the
vitrain is considered to contain vitrinites of differing reactivity which
are converted into fine mosaic material over a range of progressively
rising temperatures. If resolidification follows before mosaic coalescence
can occur then, as Marsh explains [88], the final coke will consist of
small anisotropic units, akin to grains, compressed together. Growth, by
coalescence of still fluid units, to larger mosaic units and to granular
flow components occurs during carbonization of vitrains from coals of
higher rank but large mosaic size is not directly associated with coals
of highest Gieseler plasticity. As Fig. 14 indicates, the most fluid U.K.
coals are found in N.C.B. class 401 but vitrain cokes from such coals
contain only medium-sized mosaics as their largest anisotropic component.
/3
It is evident from Fig. ~ that mosaics increase in size when negligible
quantities of isotropic material are present. Mosaic gr·owth is therefore
regarded essentially as a process of coalescence, but whether this
-43-
process leads ultimately to granular-flow components being fully
coalesced is not clear. It is considered doubtful whether the minimum
viscosity exhibited during carbonization of even the most fluid British
coals is sufficiently low that the coalescence process can be assisted by
bubble percolation per se. However, pore formation and growth, by
stretching pore walls, could cause alignment of mosaics and deformation
of still viscous coalesced structures.
The behaviour of those vitrains which exhibit basic anisotropy does not
readily fit the general pattern of formation of graphitizing carbons.
Vitrains whose basic anisotropy is converted into flow-type structures
directly, show negligible Gieseler fluidity on carbonization. Thus the·
mechanism of formation of these flow-type structures is clearly quite
different from the mesophase growth and coalescence mechanism outlined
by Marsh [89). It seems more likely [84) that such vitrains are composed
of molecules whose planarity and alignment are such that, compared to the
mesophase growth mechanism, relatively little change is necessary to
transform their basic anisotropy into flow-type structures. In view of
their different ultimate fate, it has been suggested [91) that the fine
mosaic material, observed during the transformation of the basic
anisotropy of some vitrains into flow structures, is a manifestation of
some disturbance of the original structural order of the vitrain and
differs, in some unspecified way, from the fine mosaics formed in low
rank coals. Without more information concerning the nature of this
material, it is difficult to explain its direct conversion into flow-type
anisotropic material.
2.4.6 The classification of coke textural components
Attention has so far been concentrated on the classification of coke
textural components developed by Pat rick et al (11). However several
other classifications have been published. Summaries of eight of these
- 44 -
are given in Tables 3 and 4, Table 3 containing details of classes of
components in cokes from low-rank coals, and Table 4 the corresponding
information for high-rank components. The terms used to describe the
high-rank components in Table 4 vary widely as do· the quoted sizes cif
the components. The terms isotropic and mosaic are used in all eight
classifications to describe low-rank components. However, based on the
mosaic sizes .quoted, the classifications fall into two groups, one in
which the mosaics are sized in the range 0.5 to 2.5pm, and the other
where a size range up to 10pm is used. Since it appears doubtful whether
mosaics really differ so ·much in size, the discrepancies are considered
to arise from differences in the perceived units in mosaics components.
Unfortunately, few details have been given of the methods of size
assessment or precisely what is being measured. Classifications used to
assess commercial cokes, as opposed to vitrain cokes, also include
classes for non-fusing organic inerts and mineral matter. Details of
fourteen classification systems have recently been published (99).
2.4.7 The application of coke textural data
Marsh and co-workers have made a contribution, too extensive to be
reviewed and referenced in its entirety here, to the literature,
discussion and understanding of the·influence of coke textural components
on those coke properties relevant to blast-furnace use, ie. strength and
resistance to gasification, alkali attack and thermal shock (86). The
approach adopted was intentionally fundamental (86), laboratory studies
of the formation, strength and properties of cokes being regarded as
complementary to technical innovation. The main findings of this approach
are summarized below.
Coke strength is considered to be dependent upon the structural features
of the carbon matrix and their effect on crack initiation and propagation
[100). The importance of fissures in governing the size degradation of
coke was recognised as was the role of textural components in their
- 45-
formation [100). The interfaces between individual mosaic units or
between dissimilar components, being composed of disordered carbon, were
regarded as positions of potential weakness which could favour crack
propagation. However, it was also pointed out that favourably-aligned
interfaces and interlamellar fissures were also capable of acting as
crack stoppers [100). It was shown that the texture of cokes could be
modified by the addition of pitches to the coals prior to carbonization
[101) and this effect was applied successfully to improve the bonding
between blend components. Thus pitch additions to a blend of high- and
low-volatile coals resulted in the formation of an intermediate coke at
the interface between coke components originating from the two coals
[102). Improvements in microstrength of laboratory cokes [102) and the
R10 index of 7kg oven cokes [103) ensued.
The size of textural components can be regarded as being indicative of
their chemical reacti vi ty. In confirmation of this view, in the absence of
catalytic effects, a trend between decreasing reactivity towards carbon
dioxide and increasing content of larger-sized anisotropic components
has been observed [104), However, using the same series of cokes, the
higher structural order of larger units favoured the uptake of potassium,
and the catalytic effect on reaction with carbon dioxide of this variable
potassium content was sufficient to reverse the previous trend.
Gasification in carbon dioxide [105), heat treatment in contact with
potassium hydroxide [104) and thermal shock [105) induced the
development of fissures in coke surfaces which were commensurate with
the size and shape of the textural components present. The fissures were
orientated parallel to the basal layers of the anisotropic units. The
random orientation of the short fissures in cokes exhibiting mosaic
anisotropy were considered to be less detrimental in the context of coke
strength than the longer fissures in flow-type components.
Thus, Harsh regards blast-furnace coke as a carbon-carbon composite
material, the properties of high quality coke being dependent upon the
---- - ----
- 46-
compromise attained in the behaviour of the constituent textural types
[86]. By virtue of the wide range of textural components they contain, an
optimum compromise is attained for cokes from prime-coking coals.
Karsh's studies, while providing an insight into the influence of textural
components on coke properties, were not intended to obtain the
quantitative information necessary for their application in a technical
situation. The number of different classifications for textural components
described in Tables 3 and 4 is evidence of the widespread use of coke
microscopy. Certainly, textural studies can be helpful in the
identification of the nature of the coal blend used to produce a coke.
Nevertheless, only a few attempts to obtain statistical relationships
between the textural composition and the properties of coke have been
reported.
Kultilinear-regression analysis has been used to obtain relationships
between the J. 1.8. drum indices of cokes produced in box tests and in
pilot ovens from fourteen single coals, varying in reflectance from 0.79
to 1.67%, and the semicokes formed during testing of the coals in a
Gieseler plastometer [106]. For the DI30 indices ( obtained using 30 drum
revolutions ), the equations obtained for box-test and pilot-oven cokes
respectively were :
and
DI30= 44.428+ 0.3371180+ 0.345tFM+ 0.692tCl[ - 0.547tIF + 6.556tCFt 0.832tIN (10)
DI30=-123.2- 2.031'1180+ 2.254*Fl[+ 2.786KM + 1.408tIF + 7.257 tCFt 1.931t IN <11>
where 180, FM, CM, IF, CF and IN refer to the proportions of isotropic,
fine mosaic, coarse mosaic, incomplete fibrous, complete fibrous and
organic inert components in the semicoke. No explanation was offered for
the differing effect of the isotropic and incomplete fibrous components,
as implied by the difference in sign of the coefficients in the two
equations, in the two series of cokes. The correlation coefficients for
the two equations were 0.67 and 0.78 respectively but a markedly higher
- 47-
correlation coefficient, 0.977, was obtained for a relationship, similar in
form, obtained between the DI150 indices of the pilot-oven cokes and the
textural composition of the semicokes.
Regarding the reactivity of coke towards carbon' dioxide, several attempts
to quantify the relative reactivities of individual textural components
have been reported. One approach adopted was to calculate the
reactivities from the variation of the textural composition as
gasification proceeded [1071. The values so obtained were then compared
with those obtained by applying a modified multi-linear regression
treatment to the data. Over the range of coals studied, wide variations in
the reactivities of the individual textural components were ,observed.
However, for both approaches, average values of the reactivities tended to
fall with increasing size of textural component, but values obtained using
the two approaches were not in particularly good agreement. In another
similar microscopic study [108), although the reactivities of individual
components varied from coke to coke, the ratio of reactivities of any two
components remained remained relatively constant. This was explained in
terms of the variable catalytic influence of the mineral matter in the
cokes. Generally similar trends were observed using vitrain cokes
obtained from British coals [109), but the correlation coefficient of a
regression equation derived from the gasification rates of the cokes and
their textural compositions was low.
Clearly, much further work is needed before the influence of coke textural
components on the properties of industrial metallurgical cokes can be
quantified sufficiently accurately for technical applications.
- 48-
2.5 The prediction of coke quality
The important properties determining the quality of blast-furnace coke
are size, strength and chemical purity. Since these properties are
dependent upon those of the parent coal, considerable effort has been
expended in attempting to predict the quality of coke from the results of
the laboratory examination of coals. The chemical purity of coke, ego ash
and sulphur content, may be calculated with acceptable precision from the
corresponding data for the coal and a knowledge of the coke yield, so
that it is the prediction of coke strength that has received most
attention. Early studies have been reviewed (110). These generally
involved seeking correlations between coke strengths and single
parameters characteristic of either the swelling and softening properties
of coals or their chemical structure. Thus, parameters used include
carbon, hydrogen and volatile matter contents and data from the B.S.
swelling test, the Gray-King assay, various dilatometers, the Gieseler
plastometer and the Sapozhnikov penetrometer. In several cases this
simple approach resulted in reasonable correlations being obtained, but
only for a restricted range of coals. It is not therefore proposed to
review this early work further. Instead, it is intended to concentrate on
two better-established, more-sophisticated methods and on two recently
suggested approaches whose potential has.not yet been fully explored.
2.5.1 The prediction of drum indices
2.5.1.1 Coke petrography in coke strength prediction
Coal petrographic methods of predicting coke strength are based on the
view that the macerals in coal may be devided into reactives, which fuse
on heating, and inerts, which ·do nof, ·and that the strength of coke is
dependent upon the rank of the softening components and the ratio of
reactives to inerts.
- 49-
By extending earlier Russian studies [111], and following extensive small
scale coking tests, these views were quantified into a method of
predicting the stability factor of coke as measured by the A.S.T.M. drum
test [112]. The method required the determination of the proportions by
volume of vitrinite, exinite, resinite, semifusinite, micrinite, fusinite
and mineral matter present in a coal or blend and the proportion of the
vitrinite falling into reflectance classes 1 to 22. ( These classes
reflect variations of maximum reflectance from 0 to 2.19% ). Then, the
total percentage of inerts in a coal blend was taken to be the sum of the
percentages of micrinite, fusinite, two-thirds the semi-fusinite, the
vitrinite in class 22 and the mineral matter. The reactives were
considered to comprise the vitrinite, resinite, exinite and the remaining
one-third of the semi-fusinite. The latter was allocated to the vitrinite
class corresponding to the average reflectance of the blend while the
exinite and resinite were allocated to the vitrinite classes present in
the blend on a pro rata basis.
The two parameters used to obtain the predicted coke stability index were
the Strength Index ( SI ) and the Composition-Balance Index ( CBI ). The
Strength Index was obtained from the equation :
SI= (K,.P,)+ (K".P",)+""""(K,,,.P2')
PT
(12)
where K" K2, etc. are the strength indices of reactives in classes 1,2,
etc. obtained from the curves. in Fig. 15, P" P2, etc. are the percentages
of the reactives in classes 1, 2, etc. and PT the total percentage of
reactives in the blend. The curves in Fig. 15 were obtained from small
scale carbonization experiments, and, curiously, imply that high
reflectance vitrinites, which show inferior swelling and agglutinating
properties, give cokes having high strength indices.
- 50-
The curve in Fig. 16 shows the optimum reactiveslinerts ratio for each
vitrinite class determined from the curves in Fig. 15. These values were
used in the calculation of the Composition-Balance Index :
CBI= Q (13)
p, IM, + P"'/M"'+ ........ P2 , 1M2 ,
where Q is the total percentage of inerts in the blend, and M" M2 , etc.
,are the appropriate optimum reactiveslinerts ratios. The predicted
stability factor could then be read from the Strength Index/Composition
Balance Index graph in Fig. 17 on which are plotted previously-determined
iso-stability lines. It was shown that; for coals containing less than 12
wt% ash and for the standard carbonization conditions used, the stability
factor could be predicted with a correlation coefficient of 0.98 [113J.
This prediction method has been used with success by other workers in
the U.S.A. [114, 115], although slight modifications to the optimum
reactiveslinerts ratio were necessary, possibly to account for the effect
of different carbonization conditions. However, although the approach has
been modified to permit the prediction of Micum [116] and J .1.S. (117]
drum indices, the method has not enjoyed universal success especially
when used for coals differing in character from those American coals on
which the method was originally based. This has been ascribed [5] to the
varying coking capacities of different sub-types of vitrinite in the same
reflectance class and to uncertainties regarding the behaviour of inerts
(116]. The allocation of two-thirds of the semi-fusinite to the inert
category has been questioned and it is now recognised that the behaviour
of inerts depends on their size and distribution. For a world-wide range
of coals, A.S.T.M. stability indices of the corresponding cokes were
calculated by the original method and by a procedure modified so that the
inerts content referred only to the coarser inert material (5J. The
modification resulted in the mean deviation between the measured and
calculated stability factor falling from 14.9 to 4.6 units. It was
- 51 -
concluded that the original method predicts the stability factor with
accuracy only when the CBI is close to unity and that, even when
modified, the formulae are applicable only if the carbonization
conditions, the bulk density and coking rate in particular, are similar to
those originally used.
The American coke-quality prediction method has not been applied with
success to predict the Micum indices of cokes produced from U.K. coals, a
possible cause being the disparate coal fields from which the coals are
mined. It has been reported [5], however, that a method is being
developed. This is based on the variation of the Micum )[40 index with
total inerts content for coal blends having vitrinite reflectances in the
range 0.75-1.25% as shown in Fig. 18. The index is then corrected for the
inerts present as particles greater than 3mm and less than 0.12mm
according to the lines drawn in Fig. 19. The accuracy with which this
method predicts the )[40 index has not been reported.
2.5.1.2 Methods based on the dilatometric behaviour of coal
The most extensive investigation of the feasibility of using the
dilatometric behaviour of coal as a basis of predicting coke strength has
been undertaken in West Germany by Simonis and co-workers [118, 119,
120]. They were able to derive equations, utilising th~ volatile m~iter content and dilatometric characteristics of coals together with the size
consist of the charge and the coking conditions, for calculating both the
Micum M40 and M10 indices.
The dilometric characteristics of a coal were reduced to a single value,
G, calculated according to :
G= ( [E+V]/2 ) • ( [k+d]/[Vk+Edl ) <14)
where E and V are the temperatures of initial softening and maximum
dilatation and k and d are the contraction and dilatation respectively.
- 52-
The coking conditions were also reduced to a single value, the coking
factor K, obtained from the relationship
K= ( pvll ) I 2 <15>
where p, v, and II are the charge density, carbonization rate and oven
width respectively.
The departure of the size distribution of the coal charge from the
optimum size distribution was calculated using :
S=E(p-po) <16>
where p and po are the measured and optimum percentages by weight of the
coal in each of the five size ranges cons1dered for which p> pc,.
The Micum M40 index was then calculated according to
M40= A+ BK+ CS <17>
where Band C are dependent upon both the G value and the volatile
matter content ( wt% dafb ), while A is dependent only upon the volatile
matter content.
From the tabular data presented by Simonis (1181, it is evident that A
varies more markedly with ·the volatile matter than with the G value and
that the ranges of B ( ie. 0 to 6.1 ). and C ( ie. +0.18 to -0.40 ) differ
widely. Since K will usually lie between 20 and 25 while S will rarely
exceed 25, the M40 index will be dependent primarily on the volatile
matter content, through its influence on A and B, and the coking factor,
K.
The use of this equation is illustrated in Table 5. For these examples,
it is assumed that a 45cm wide oven is used to carbonize a charge of
density 0.785t/m'" in 16.8h. Thus K= 23.7. It is further assumed that the
coal size is larger than the optimum with 18.8 wt% being greater than
3.15mm, the optimum percentage for this size range being zero. The
- 53-
proportions present in the other size ranges being less than the optimum
proportion, S= 18.8. The values qf A, Band C given in the table are taken
from the data of Simonis [118], assuming that the G value is unity. As
Table 5 shows, for these conditions, as the volatile matter content of the
charge increases from 20 to 35 wt%, the predicted M40 falls from 83.1 to
43.7 units.
Using data from thirty-one coking plants using coal blends with volatile
matter contents ranging from 20.7 to 29.0 wt% , the M40 index could be
calculated by this method with a standard error of 2.4 units.
The method was subsequently modified to use a G value calculated from the
petrograpic analysis of coal [121]. Since the dilatometric characteristics
can be determined more readily than the petrographic data, such a
modification appears to have limited merit. However, this modified
procedure is recommended, by the originator, as the most useful method of
predicting coke strength from petrographic analysis [5].
Regarding the Micum M10 index, using the parameters defined earlier, the
following relationships were obtained by regression analysis [120]:
<18>
where the coefficients Mo, M" etc. are dependent on the volatile matter
content, V, and the dilatometer G value according to
For the twenty cokes considered, ranging in volatile matter content from
25.1 to 34.2 wt%, the standard error of estimating the M10 index from the
equation was 0.65 units.
Both methods were claimed to apply for 'coals and coal blends' with
volatile matter contents of 18 to 35% w/w, G values from 0.95 to 1.10,
and containing less than 20% by volume of inertinite, carbonized under
conditions such that the coking factor lies between 19 and 24.
- 54-
However, when it was attempted to apply this approach to blends of U.K.
coals [122], it was found that the standard error of estimating the M40
index increased from 2.4 units for blends with volatile contents of 20 to
30 wt% to 13.3 units for those blends containing more than 34 wt% of
volatile material. It was therefore considered that this relationship was
inadequate for blends of high volatile coals. An alternative equation was
therefore derived using data from one hundred and seventy blends of U.K.
coals carbonized in a 250kg pilot oven. This was :
M40= 103.9+ 24.8G- 1.196V5/l0E.+ 2.57V"'/T- 88V/T < 20 >
G being the dilatometer G value, V the blend volatile matter content (wt%
dafb ) and T the carbonization time to a centre charge temperature of
900'C in an 18" wide oven. The standard error of estimating the M40 using
this equation was 2.2 units for blends containing neither coke breeze nor
anthracite. The equation was stated to apply to blends of volatile matter
content from 19 to 41wt%, with G values from 0.89 to 1.13 and carbonizing
times of 13.9 to 19.5h. Values calculated according to this method are
included in Table 5 to highlight the difference between the M40 values
predicted by the German and British methods, especially for blends with
high volatile matter contents.
The German method of predicting the Micum M1D index from the volatile
matter content and dilatometer characteristics has been applied to data
for U.K. coal blends with disappointing results [122]. Attempts to improve
correlations by modifying the mathematical treatment have also proved
unsuccessful. No explanation has been proposed.
2.5.2 The prediction of the tensile strength of coke
No method for the prediction of the tensile strength of coke has been
established but two studies appear to have the potential to form the
basis of a suitable method.
- 55-
2.5.2.1 The use of pore structural parameters
Following extensive careful study using automated image analysis, two
equations were derived to relate the tensile strength of metallurgical
cokes to parameters characteristic of their' porous structure [10]:
(21)
s= 450- Fmax-O.5 exp-[ 2( Fmax/Fmin )0.5_ p ] <22)
where Wand P are wall and pore intercept sizes, N the number of pores
per field examined, Fmax and Fmin the maximum and minimum Feret's
diameter of the larger pores and p the volume porosity.
These equations in themselves are inadequate for use in coke tensile
strength prediction. However, there are firm indications, based on the
study of a limIted number of coals, that the pore and wall intercept
sizes of cokes from blended-coal charges are additively dependent upon
the blend composition and the pore-structural parameters of cokes
obtained from individual blend components [123]. These findings have been
successfully exploited in the formulation of blends for the production of
coke with specific tensile strengths [10], However, if the additivity
principle applies also to the number of pores per field then equation
<21) could be used as a basis for calculating the tensile strength of
cokes obtained from blended charges from a knowledge of the porous
structure of single-coal cokes. It is anticipated that equation (22) could
be used in a similar way.
- 56-
2.5.2.2 Using the 'bond strength' approach
If cokes from blended coal charges are examined under polarized light,
discrete regions consisting of material from the parent coals can be
readily identified, any interaction being limited to a narrow inter
regional layer [124]. On this basis, a model for the fissure-free strength
of coke has been proposed in which the strength is considered to be
dependent on the bonding between blend components and their probability
of contact. The treatment closely resembles the classical treatment of the
interaction between molecules in multi-component liquids [125], For a coke
from a binary coal blend, the strength is then:
(23)
and for a ternary blend :
where S" S2 and Sa corespond to the strength of cokes from the single
coals 1, 2, and 3 respectively and S'2, S,~ and S23 are the strengths of
the bonds between coals 1 and 2, 1 and 3, and 2 and 3 respectively.
Although this approach was considered to refer strictly to the coke
tensile stength, its reported application involved the attempt to
calculate, using equation (24), the Micum 1(10 index of cokes made using
ternary coal blends from the inter-bond indices, obtained using equation
(23), from the determined MI0 indices of cokes from binary coal blends.
For the ternary blends reported, although the predicted MI0 indices
followed the same pattern as the experimental data, the model
consistently underestimated the MI0 index.
- 57-
2.6 Outline of research
The original objective of the study described in this thesis was to
investigate whether the examination of the texture of the carbon in
metallurgical coke by scanning electron microscopy could provide
information additional to that previously obtained using polarized-light
microscopy. In particular, it was enVisaged that, by exploiting the large
depth of focus of an SEM, the study of the fracture surfaces of cokes
might provide further insight into the three-dimensional form of the
textural units present and their mode of failure. The findings of this
study added confirmation to the view [100) that the textural composition
of coke could have a bearing on its strength. Two approaches were
therefore adopted in order to examine this possibility further. Firstly,
the influence of textural components on crack development during the
tensile testing of coke was investigated and, secondly, attempts were
made to derive relationships which quantified the dependence of coke
tensile strength on the textural composition as determined during SEM
examination of coke surfaces etched in atomic oxygen. The latter study
was primarily orientated towards investigating the possibility of
predicting the tensile strength of cokes, made from blended-coal charges,
from the behaviour of the.individual coals. Since the relationships
obtained appeared to be potentially useful in industrial situations where
scanning electron microscopes are not readily·available, the ~pplicability of the form of the derived relationships to textural data obtained using
polarized-light microscopy was also investigated.
- 58-
3. EXPERIMENTAL STUDIES
As explained in the previous section, this study was carried out in four
discrete phases, the conclusions from one providing the 'justification for
the ne,xt. Accordingly, this Experimental Studies section is in four parts,
each of which descrihes the experimental procedures and presents and
considers the results obtained for one of the four phases of the study.
3.1 An SEX study of fractured and etched coke surfaces
3.1.1 Introduction
The literature review described how the examination of polished
metallurgical coke surfaces using a polarizing microscope provided
interesting information regarding the structure of the coke carbon, the
variation in the size and form of the reactive-derived textural
components reflecting the macro-crystallinity, (126) of the carbon. of .
Examination, in an SEM, argon-ion bombarded coke surfaces (127) provided
further insight into the nature of these entities. The original objective
of this study was to establish whether features evident on SEM
examination of fractured coke surfaces, and surfaces etched in atomic
oxygen, could also be identified with textural constituents revealed by
polarized-light microscopy. Atomic-oxygen etching had previously been
used to enhance features in surfaces of carbon-carbon fibre composites
(128),
3.1.2 Experimental procedures
3 .1.2.1 Cokes used
The cokes examined were produced in a small pilot oven from a series of
six coals ranging in rank from a steam-coking coal ( N.C.B. class 204 )
-59-
to a weakly-caking coal ( N.C.B. class 602 ). Analytical data for these
coals are given in Table 6. The coals covered the whole range of coal
rank normally encountered in blast-furnace coke production in the U.K.
3.12.2 Carbonization procedure
The carbonization method developed for this study involves the
progressive immersion of a cylinder of packed, crushed coal into a
furnace so that a plastic layer moves through the coal charge in a
manner which simulates the carbonization of coal in a commercial oven. A
line drawing of this small pilot oven and its ancilliary equipment is
shown in Fig. 20. A full description of the oven has been published [129).
The coal retort is of silica, lm long and 60mm in internal diameter. The
coal charge is located between silica-brick spacers, drilled with a
number of holes to permit gas passage and held in position by spring
loaded silica rods. The ends of the retort are fitted with flanges, the
inlet flange carrying a gas inlet tube and a thermocouple inlet facility,
and the outlet flange a 25mm diameter outlet pipe. The latter is
connected to a tar trap, the plastic outlet pipe of which acts as a
condenser for tars. The furnace is wound in three parts, each being
controlled thermostatically' so that the temperature variation along a
250mm hot zone is less than l5·C. The furnace, fitted with wheels, is
mounted on rails and drawn along by a wire connected to an electric
motor, the rate of movement of the furnace being determined by the size
of the cam on the final drive shaft.
The operating procedures adopted as standard are as follows. With the
retort in a vertical position, the quantity of air-dried coal,
approximately 600g, sized 90% by weight less .than 3mm, required to form
a cylinder 225mm in length at a charge density of 820kg/m3 is added in
six equal increments, the coal being tamped down between each increment
to the required level. The tube is then positioned as shown in Fig. 20
- 60-
and a flow of oxygen-free nitrogen purge gas started. The furnace,
controlled at 1080·C to give a centre charge temperature of 1040·C, is
drawn along at 20mm/h. These carbonization conditions were chosen to
.simulate the temperature regime experienced by a preheated, gravity
charged coal blend in a 450mm wide slot oven. When carbonization is
complete, the retort tube is removed from the furnace and allowed to cool,
the flow of nitrogen being maintained.
3.1.2.3 Tensile strength determination
The product from the oven is a single piece of coke, fissured internally.
Depending on the degree of fissuring, thirty-five to fifty 10mm diameter
by 10mm long cylindrical test specimens were prepared from each charge
for tensile strength testing by the diametral-compression method, the
brittle nature of the coke necessitating the use of diamond-tipped
cutting implements. The coke strengths were determined, according to
equation (9), using a Tensometer universal testing machine, a cross head
speed of 0.5mm/min being employed. These test conditions are based on
procedures informally standardized at the B.C.R.A. [81. Preliminary tests
indicated that the differences between the mean tensile strengths of
cokes from repeat carbonizations lay within 0.2 to 0.3MPA. For the cokes
considered, as detailed in Table 7, the tensile strengths ranged from 4.42
to 6.61MPa, such values being comparable with those of good quality,
commercially-produced blast-furnace cokes [101. As an indication of the
dispersion of strength values of individual specimens about the mean, for
these cokes the standard errors of the mean tensile strengths are
included in Table 7. These lie in the range 0.21 to 0.29 MPa and are
typical of both the values obtained for cokes studied in this work and
values obtained when commerial cokes are examined. Also quoted in the
table are the measured fractional coke yields ( w/w ) and yields
calculated from the analytical data for the coals.
- 61-
3.1.2.4 Specimen preparation for SEM examination
To prepare etched surfaces, epoxy-resin blocks, 50mm in. diameter,
containing. ten to sixteen rectangular pieces of uncrushed coke, were
prepared and polished, by conventional techniques, to a standard suitable
for examination under an optical microscope. They were then etched by the
highly reactive species, presumed to be atomic oxygen, formed in carbon
dioxide by an electrode less ,discharge, power being, supp.l1e~H~....f. _ .. iW w.J;!-~ cP-J!. dj-~ ~~ ---L _. <v.1 0{
~Jtlr~~r:~~~~ ~p-The apparatus used ·is illustrated in Fig. 21. The discha~tube was
""-fitted with a needle valve to control the gas flow, the pressure~ithin
the system being determined by the the gas flow and the unthrottled \ suction of a rotary pump. Pressure and power settings were chosen so that
\ the visible discharge just touched the sample. Under standard operating
~onditions, adequate change in surface topography was attained in about~ ,20min.~mall specimens, each containing one rectangular coke surface, were / ~ then cut from the large block) t;/tJ.r,~-,1 ~ ~'" "
. / (I ~ ~V yvj fot-~t9f:'j _~tched specimens and~hemi-Cynndr:fcal, -broken, tensile=-tes{,?,'Y'ff~:;:L; J
,t"J~ ~~ ,3pecimens·'were cleaned ul tra-sonically.llta-,then"gpld coated, using a· \ ~€.
IY ()~ a<-? ~~ ~ -) Nanotech Semprem 2 sputter coater,Abefore/J'xamination in-a_Cambddge
r --' -i lt ~I nstrum.ents-S604-scann ng_e ec ron .microscope .......
3.1.3 Results
Low-magnification views of fractured and etched coke surfaces are shown
in Fig. 22a and·b. The most obvious feature of the fracture surface shown
in Fig. 22a are the pores, P, their interconnected nature being
immediately apparent. Variations in the form of features in the broken
cell walls only become evident at higher magnification. The atomic oxygen
etchant preferentially attacked the epoxy-resin filling the coke pores,. P,
- 62-
leaving the coke carbon standing slightly proud. Examination of etched
surfaces at higher magnification revealed pitted and channelled surfaces
varying considerably in appearance. By viewing the same area by the
appropriate technique before and after etching, it was established that
the surface texture of etched coke was dependent on the type of
anisotropic component present. This process was facilitated by the cokes,
being made from single coals, possessing relatively simple textural
compOSitions, no coke containing more than two vitrinite-derived
components with fractional contents greater than 0.1 v/v. Textural
components form a series continuously varying in appearance so that their
classification is a subjective process. Nevertheless, it proved possible
to classify textural constituents in etched surfaces in a manner
analogous to that used to classify optically-anisotropic components in
polished surfaces and, by using a point-counting technique, to quantify
the textural composition. This involved examining five hundred positions
on the etched coke surface at a magnification of 10,000 times and
allocating the component present to one of the textural component
categories described in Table 8. Points 0.5mm apart in traverses 0.5mm
apart were examined. Examination of eight to ten small blocks was
necessary to accumulate five hundred counts. It is estimated that the
errors associated with counting vary from .01 at the .01 level of
component to .04 at the .5 ·level.
The results obtained using the six single-coal cokes are given in Table 9
and illustrated in Fig. 23. In this, coke textural components are
indicated both by the flow-type, coarse flow, mosaic nomenclature used in
connection with anisotropic entities and the corresponding terms
lamellar, intermediate and granular which are considered more descriptive
of the components when seen in partial three-dimensional view in etched
and fractured surfaces. However, as will be seen later, a strict· ·one~to-'
one correspondence does not exist between the two methods of
classification.
- 63-
The variation of the SEM textural composition of coke with rank of the
coal carbonized is illustrated in Fig. 23. Cokes from high-rank coals
contain lamellar and intermediate components as major constituents
whereas those from lower-rank coals are composed of intermediate and/or
granular components. The cokes from higher-rank coals also contain a
minor constituent which appears relatively featureless in etched surfaces
and is accordingly termed flat. As Fig. 23 indicates, the six cokes also
contained 0.15 to 0.30 v/v of carbonaceous inert components.
The appearance, at high magnification, of the various textural components
in etched surfaces is illustrated in Figs 24a-28a. After etching, the flat
component,_ illustrated in Fig. 24a, appeared relatively featureless with
either occasional pits or short channels. Lamellar components, Fig. 25a,
exhibited parallel arrangements of ridges and channels, greater than 10~m
in length. Predominantly, the ridges and channels were aligned
circumferentially to the pore surfaces in a manner analogous to that
described for petroleum cokes [71). Intermediate components, Fig. 26a, are
intermediate in appearance between lamellar and granular forms with short
ridges and channels, often branched. Some variation in form was evident
for both lamellar and granular forms, the interchannel spacing varying
from 0.5 to 3~m. This was considered to reflect the angle with which the
constituent lemellae intersected the etched surface so that these
categories were not sub-divided when pOint-counting. Granular carbon,
prevalent in cokes from lower-rank coals, exhibited a uniform pitted
appearance in the SEM after etching, Fig. 27a. When quantifying the coke
texture, such components were divided into four sub-classes; coarse,
medium, fine and very fine, the associated pit sizes being approximately
0.3, 0.2, 0.15 and O.l~m respectively. Carbonaceous inert components were
identifiable either by their woody structure ( Fig. 28a ) or, if small, by
their un fused sharp edges.
Also shown in Figs 24-27b and c are micrographs at two magnifications
illustrating the corresponding textural features when seen in the tensile
- 64-
fracture surfaces. Correspondence is based on the general appearance and
the proportions of the various components present. Fracture surfaces of
flat components, Fig. 24b and c, are characterized by brittle fracture
river patterns but are otherwise qUite smooth. The lamellar nature and
circumferential alignment of lamellae relative to a pore surface is
ilustrated in Fig. 25b and c. Crack propagation during specimen failure
has involved translamellar fracture. However, the thickness of the
lamellae in the fracture surface exceeds the interchannel distance noted
in etched surfaces. The fracture surfaces containing intermediate
components, Fig 26b and c, are again intermediate in appearance between
the fracture surfaces of lamellar and granular constituents. In some
views the material seems to be 60mposed of small distorted lamellae. The
fracture surfaces of granular components, shown in Fig. 27b and c, give
the impression of intergranular fracture with grain sizes coresponding to
the pit sizes in etched surfaces. Some areas containing brittle fracture
river patterns are observed in surfaces of very fine granular components.
The fracture surface of a carbonized fusain particle with a well
preserved woody structure is shown in Fig. 28b. This has fractured giving
a very smooth fracture surface with occasional river patterns.
3.1.4 Discussion
This work has shown that features visible, when fractured and etched coke
surfaces are viewed in an SEK, can be identified with textural units
revealed by polarized-light microscopy of polished coke surfaces.
Generally, the form of the textural units deduced by interpreting the
variation of shading or colour when a specimen is rotated under polarized
light is confirmed by the present SEM studies. Of the two techniques, it
is considered that SEM provides more direct information, in partial
three-dimensional form, which requires little further interpretation. This
is particularly true for lamellar components. SEK examination of etched
surfaces, Fig. 29, can provide immediate evidence of the splays and folds
of the constituent lamellae. Such information can be obtained by optical
-65-
microscopy but only by intensive observation of the movement of
extinction contours as the specimen is rotated (65).
In the present work, atomic-oxygen etching of coke surfaces was carried
out in the,expectation that the less etch-resistant parts of the structure
would be removed leaving the more ordered structure for viewing in the
SEM [128). Argon-ion bombardment has previously been used to enhance the
structural features of vitrain-cokes surfaces [127). Subsequent SEM
examination revealed parallel ridges and channels in cokes containing
flow-type anisotropic components and almost hemispherical nodules in
cokes consisting of mosaic units. The nodules varied in size with vitrain
rank and were apprOXimately one-half the size of the mosaic units visible
by optical microscopy. It was concluded that the mosaic units took the
form of distorted spheres and that, when these 'were aligned and
overlapped, ridges rather than nodules resulted.
Quite different effects were observed in the present study after etching
with atomic oxygen. For those surfaces containing mosaic units, the
pitted surfaces obtained suggest that it is the centres of the mosaic
units rather than the interfaces between them which are the most reactive
to atomic oxygen. This tends to confirm the impression given by
micrographs of atomic-oxygen-etched surfaces of Gilsonite-pitch coke
[85), but contrasts with the interpretation of the appearance of etched
metallurgical coke (87) in terms of the attack on disordered carbon at
the interfaces between mosaics. The size of etch pits appears to
correspond more closely with the size of nodules on argon-ion bombarded
coke surfaces [127) than with the quoted sizes of mosaic units Ill). The
ridged surfaces evident after etching or ion bombardment of cokes
containing flow-type anisotropic units are quite similar in appearance.
However, in view of the contrasting'effect of the' two treatments on
mosaic components, it is possible that the ridges on one surface reflect
the channels on the other. Apart from this different mode of etching, the
results observed with atomic-oxygen etching are consistant with the
- 66-
conclusions regarding the shape of textural components drawn from the
ion-bombardment experiments. However, no surface features corresponding /
to the intermediate components observed in the prese~t work were seen in
ion-bombarded surfaces.
Since atomic-oxygen etching and argon-ion bombardment attack different
parts of the surface of metallurgical cokes, it can no longer be gen::ally
assumed that etchants preferentially attack the-less well-ordered ~-- ---- -- - -- --- ---,.---~
crystalline structures. The alternate arrangement of ridges and channels -' --- - - -- - , -induced into surfaces of lamellar carbon by both techniques is also t - - - - - - - ---- - -- -- -- --
difficult to explain. That this arises as a result of a periodic variation r" "......- • ______ •• _____ _
in reactivity across the surface is clear but the cause of this is not: ~,=-':-::===2..-=-::':=~==-::=-=_ -- .----- ._- . -----___ _ It is evident from the electron micrographs that the appearance of
fracture surfaces varies with.the rank of the coal from which the coke
was produced. In most instances, the textural component responsible for
the fracture feature could be identified readily by the similarity between
the feature and the appearance of the etched surface. The examination of
fracture surfaces therefore lends confirmation to the interpretation of
the shape of textural components in coke carbon deduced from study of
etched surfaces. For the lamellar components, both the lamellar nature and
the preferred alignment of· lamellae relative to the pore surfaces are
evident in fracture surfaces. Because of the circumferential alignment of
lamellae relative to the pore surface and crack propagation from pore to
pore, fracture of this component results in breakage across, rather than
between, lamellae. In fracture surfaces, lamellae appear thicker than the
interchannel distance in etched surfaces, the thickness representing the
spacing of interlamellar fissure formed in coke during carbonization.
Fracture surfaces of cokes .contaIning granular components give the
impression of intergranular failure, the size and shape of the grains
being consistant with the nature of units deduced from studies of etched
and ion-bombarded surfaces. The implication is that, for 'this material; it
- 67-
is the interface between textural units which provides a weakness array
along which cracks may propagate. The textural components responsible for
the flat and very finely pitted textures in etched surfaces of cokes from
high- and low-rank coals respectively both give rise to smooth fracture
surfaces containing 'brittle fracture ~iver' patterns. Similar smooth
surfaces are observed with fractured inert components and are considered
to be associated with very rapid crack propagation through coke carbon
containing only short range order.
The interpretation of the nature of coke fracture surfaces solely in
terms of the structural units in coke carbon is consistent with the
acknowledged brittle character of metallurgical coke [81. Features in
fracture surfaces are thus unlikely to be the result of plastic
deformation. The variation in the degree of roughness of the fracture
surfaces of the various textural components and the variation in the mode
of fracture thus revealed, particularly for the lamellar and granular
components, suggest corresponding variations in the strength of the
carbon matrix composed from them. Any such variation would be expected
to contribute to the variation in strength amongst cokes, the effect
being dependent upon the proportions of the various textural components
present.
,
-- - - ------------~
- 68-
3.2 An SEX study of the tensile fracture of coke
3.2.1 Introduction
It is evident from the study of fractured and etched metallurgical coke
surfaces just discussed that the fracture surfaces of the various textural
components in cokes differ in topography in a manner which, particularly
for the flat, lamellar and granular components, implies a difference in
their mode of fracture which could have a bearing on the strength of
coke. Accordingly, an attempt has been made to study the initiation and
propagation of cracks in coke subjected to diametral compressive loading,
the object being to further the understanding of the breakage of coke and
to assess the influence thereupon of the various coke textural
constituents. EqUipment necessary for the direct observation of coke
specimens under slowly increasing load not being available, the approach
adopted was to compare, by examination in an SEM, central, circular, cross
sections of 'as received' test specimens with similar surfaces of
specimens after diametral-compressive loading.
3.2.2 Experimental procedures
3.2.2.1 Coke used
The coke used was produced in a 17t test oven [4J by carbonizing a wet
charged blend containing a 2: 3: 1: 4 mixture of coals in H.C.B. classes
204, 301b, 501 and 502. This coke was selected on the basis of the
variation of the coke textural composition with coal rank, Fig. 23, in the
expectation that it would contain appreciable quantities of lamellar,
intermediate and medium and fine granular carbon, together with some
inert components. The tensile strength of this coke, determined by the
diametral compression testing of fifty 10mm long by 10mm diameter
- 69-
specimens, was 5.1MPa, a value commensurate with that of good quality
blast-furnace coke.
3.2.2.2 Specimen preparation
The f .
folowing ~
four types of specimen were examined in the SEM
a. 'as-received' specimensj
b. 'stressed' specimens, which had survived a load equivalent
to the mean tensile strength of the coke;
c. 'stress relieved' specimens, the loading of which was
discontinued when a marked fall in the force-displacement
curve indicated marked stress relief;
d. 'fractured' specimens, ie. after loading to failure.
It should be noted that as regards their strength the 'as-received',
'stress relieved' and 'fractured' specimens should be more comparable to
the sample as a whole than the 'stressed' specimens, all of which
belonged to the stronger half of the sample. The latter specimens had
been stressed to between 50 to 100% of their breakage stress.
To prepare specimens for SEM examination, ten specimens of each type
were first embedded in epoxy resin, leaving 5mm of their length standing
proud. After curing the resin, the exposed coke was abraded away and the
surface of the cross-sectional plane at the centre of the specimen
polished by standard techniques. Polished surfaces were then etched in
atomic oxygen and coated with gold prior to examination in the SEM. At
all stages of preparation and examination, care was taken to ensure that
the orientation of the stressed diameter was known to within plus or
minus 10'.
-~
- 70-
3.2.2.3 SE)! examination
The specimens were examined in the SEX-using fourteen equally-spaced
traverses at an instrument magnification of 200 times. Under these
conditions the whole of the surface of the specimen could be viewed with
minimum overlap. The positions along the traverse of any flaws were
noted, as were the textural component through which they passed. To
identify the textural component, the magnification was raised to 10,000
times. For extended microcracks and fracture cracks, ( the terms are
explained below ) this procedure was repeated for each traverse in which
the crack was observed, the textural component at two positions being
noted for each field of view. The texural composition of the coke was
determined as described earlier.
3.2.3 Results
Preliminary examination of specimens in the SEX showed that the flaws
present varied from microcracks, which extended from a pore into or
across the adjoining pore wall, to extended microcracks, larger fissures
at least 300~m in size and evident on more than one traverse, and
ultimately to fracture cracks traversing the whole specimen. Accordingly,
when mapping their distributions, flaws were classified into one of these
three categories. Interlamellar fissures in lamellar carbon were
discounted unless they originated at a pore. Also, minor flaws near the
edges of specimens, which were considered to have arisen during sample
preparation, were ignored.
The appearance of the three types of microcrack is illustrated in the
micrographs in Figs 30-32. In lamellar coke carbon, the lamellae are
usually aligned circumferentially to the pore surface so that crack
propagation from pore to pore results in breakage across the lamellae as
shown in Fig. 30a. The jagged nature of the crack path is attributed to <
the propagating crack temporaily being diverted along relatively easy A
-'71-
pathways between lamellae. Fissures in intermediate and medium granular
components are shown in Fig. 30b and c. The variation in tortuosity of
the cracks shown is commensurate with the variation in the size and
shape of the components present.
The carbonaceous inert components in the coke were invariably enfolded
within reactive-derived constituents. For present purposes, flaws were
regarded as being associated with an inert particle if they extended from
a pore and traversed the pore wall to the inert particle ( Fig. 31 ). More
extensive cracks are illustrated in Fig. 32, a network of extended
microcracks being shown in Fig. 32a and part of a fracture crack in
Fig. 32b.
Flaw-distribution diagrams obtained using the four types of specimen are
shown in Figs 33-36, the orientation of the stressed diameter being from
top to bottom as indicated. The values below each individual flaw diagram
in Figs 33-35 refer to the number of flaws observed in that specimen.
Within each group of specimens considerable variation in the number of
flaws per specimen was observed. However, it is clear that the 'stress
relieved' and 'stressed' specimens contained a larger number of both
simple and extended microcracks than the 'as-received' specimens. The
average number of simple microcracks per specimen for the 'as-received',
'stressed' and 'stress relieved' specimens was 37, 47 and 54 respectively.
The total number of extended microcracks, indicated in the distribution
diagrams by short lines, observed in all ten specimens of each type was
14, 28 and 34 respectively.
Crack diagrams for the fractured coke specimens are given in Fig. 36, the
determined strength for each specimen being given below each individual
diagram. The mean tensile strength of these specimens, 4.71MPa, is
slightly lower, than that of the sample as a whole, but the spread of
results is similar'. Positions of microcracks in these specimens were not
recorded. The figure demonstrates clearly the complexity of the crack
- 72-
networks resulting from breakage of coke under diametral compression.
Multiplanar and branched cracking is the rule rather than the exception.
In one or two instances, it is suspected that the orientation of the
loaded diameter had not been maintained accurately during sample
preparation, but this does not obscure the fact that in certain Cases"a
subsidiary crack had propagated at a high angle to the loaded diameter.
The frequency of observation of the various textural components at
microcracks in the 'as-received', 'stressed' and 'stress relieved'
specimens, and at fracture cracks in the broken specimens, is expressed
as a fraction and compared with corresponding values for the coke
matrix, ie. the textural composition of the coke, in Table 10, the textural
components being identified by their initial letter as in Table 8. The
relatively low number of microcracks associated with inert components
stems from their immersion within the pore-wall material. When account is
taken of this factor, for the reactive-derived components there is a
broad correspondence between the textural composition and the frequency
of observation of textural components at microcracks and at fracture
cracks.
3.2.4 Discussion
Metallurgical cokes are notoriously inhomogeneous. Thus, individual cokes
made using a blend of a number of coals of varying rank are likely to
contain all the microscopic features found in all commercial cokes. In the
present study, numerical data was accumulated in an attempt to make valid
deductions. However, in view of the relatively small number of specimens
examined, it is doubtful whether the microcrack-density data in
particular should be regarded as more than semi-quantitative. Furthermore,
it is recognised that this study represents an attempt to investigate a
three-dimensional effect by examining two-dimensional features.
- 73-
Variations in the mode of fracture of the various coke textural
components were evident from the SEM study of fractured coke surfaces.
Additional supporting evidence is now available from the micrographs
illustrating crack paths in Figs 30-33. The irregularity of the crack
path through granular components clearly varies depending on the pit
sizes observed in the etched surface, thus supporting an intergranular
fracture mechanism. Fracture of the lamellar components aligned
circumferentially to the pore surface involves breakage across the
lamellae. The jagged fracture path shown in Fig. 30 is typical of a tough
material. It is this difference in the mode of fracture which suggests a
corresponding variation in the porosity-free strength of the different
coke carbon textural components. For the lamellar and larger granular
components, the observed large angular diversion of the crack paths may
be indicative of the involvement of shear as well as tensile stresses in
crack propagation.
As Table 10 indicates, the frequency of observation of the various
textural components at both fracture cracks and microcracks corresponds
to that expected on the basis of the textural composition of the coke.
Thus, in contrast to previously expressed views (lOO], there appears to
be no marked preference for cracks to be initiated in, or diverted
through, any particular textural component. However, textural components
in cokes are' not intimately and homogeneously mixed but, to a first
approximation, can be regarded as existing in discrete volumes which
correspond broadly in size to the sizes of the original coal particles.
Diversion of a propagating crack around the larger of such volumes would
only occur if a very easy route for crack propagation were available.
Apparently, any differences in the strength of textural components is
insufficient to promote such effects. No evidence was found to support
the view £124] that interfaces between different textural components were
positions of weakness.
-74-
Regarding the mechanism of coke breakage, with the techniques available
it was not possible to observe directly the various stages leading to
fracture. Nevertheless, from the data it is possible to construct the
following sequence of events.
Since coke specimens subjected to diametral compressive loading contained
a higher number of microcracks and extended microcracks than the as
received specimens, the implication is that stable microcracks are
introduced into the coke structure by stresses smaller than the breakage
stress. The stresses associated with shrinkage during the later stages of
the carbonization process, thermal shock during quenching and mechanical
shock during handling are considered to be responsible for the
microcracks seen in the as-received specimens as well as for the gross
fissures visible to the naked eye in lump coke. The variation of the
microcrack density in the as-received specimens implies differing degrees
of local pre-stressing.
It is not possible to differentiate between pre-existing microcracks and
the additional ones introduced by diametral-compressive loading. However,
the flaw-distribution diagrams do not suggest a marked tendency for
microcracks to be concentrated in the area of high tensile stress along
the loaded diameter. The ratio of compressive to tensile stress increases
with distance from the centre of the specimen [130), but since the
initiation of microcracks may also depend on other factors, for example
the inclination of the stress concentrating flaw to the direction of the
applied stress, microcracks formed away from the loaded diameter are not
necessarily Induced by compressive forces.
Most simple microcracks were observed to extend from a pore into the
cell wall or to traverse the cell wall to an adjacent pore. This implies
that pores are tpe principal centres for crack initiation. Since so many
simple microcracks were observed to extend only from one pore to another,
pores appear to have the ability both to initiate and stabilize
-75-
microcracks. Stabilization is presumed to stem from the broadened crack
tip no longer acting as an effective stress raiser so that crack
propagation is interrupted. No obvious difference in the shape of those
pores which acted as initiators or stabilizers of microcracks was
apparent in two-dimensional views of etched surfaces. The formation of
stable microcracks will be associated with the release of strain energy.
This effect, together with some edge crushing near the platens, is
responsible for the applied load often increasing in a stepwise manner
and for the audible creaks and groans emitted by coke specimens under
load.
Initially, each further increase in the load can be envisaged to induce a
new generation of discrete subcritically-sized m icrocracks , many quite
remote from those formed earlier, and these too only grow to limited
sizes before being stabilized. At higher stress levels, when the
microcrack density is already high, further loading of the specimen
induces the formation of extended microcracks, this process being
associated with a much higher degree of stress relief and consequent
fallback in the applied load. It is evident from Fig. 35 that the length
of stable extended microcracks can approach 5mm. At present it is not
possible to decide whether such extended features are the result of a
single initiation/propagation event, the extension of a previously
existing microcrack, or the joining together· of a newly~formed microcrack
with a previously-existing one.
Failure is considered to occur when the concentration of microcracks and
extended microcracks is so high that the next increment of the load
results in the joining together of sufficient stable microcracks to form
a large critically-sized flaw. Then, according to simple flaw theory,
failure directly ensues. Graphites, despite having quite different
structures, also are reported to fail by a mechanism involving the
formation of unstable flaws from the stable microcracks induced at lower
stress levels [131).
- 76-
On this failure mechanism, a high subcritical microcrack density in coke
is a precondition for failure. Since the 'as-received' specimens exhibit a
wide variation in the number of microcracks they contain, it is
reasonable to postulate that the dispersion of strength values for a coke
is related to the flaw density in the individual 'as-received' specimens.
On this basis, those 'as-received' specimens in Fig. 33 which contain a
high microcrack density would be likely to fail before a load equivalent
to the mean tensile strength of the sample was attained. Thus data for
the 'stressed' specimens are more comparable with those for the 'as
received' specimens containing fewer inherent flaws.
Fracture crack systems in broken tensile specimens are very complex
( Fig. 36 ). The multiplanar and branched cracking observed results from
. the strain energy released exceeding that necessary merely to propagate
the crack and therefore initiating secondary cracks ahead of the
propagating tip. However, some contribution to the complexity of the
fracture crack system can be expected to arise from secondary cracking
near the loading positions. There is some evidence in Fig. 36 of cracks,
usually subsidiary ones, occurring at high angles to the loaded diameter,
and in such cases it appears inevitable that shear forces were involved
in their propagation. Nevertheless, Fig. 36 does show that the broken
specimens do contain a diametral fracture plane. This is one criterion
necessary for the determination of a valid tensile strength value by this
technique [611. However, previous work [47] showed that, when specimens
of industrial cokes were loaded under diametral compression, marked
deviations from the theoretical stress distribution occurred and it was
concluded that coke tensile strengths determined in this way should be
regarded as comparative not absolute values.
The successful use of the equation (22) to relate the tensile strength of
cokes with their porosity and the Feret's diameter of the larger pores
implies that the larger pores act as the Griffith critical flaws in coke.
Clearly, in a statistical sense, the larger pores control the coke tensile
- 77-
strength and they appear to provide the stress concentration necessary
for the initiation of the microcracks observed in the present work.
However, the direct evidence now available demonstates that there exists
in coke, prior to failure, extended microcracks of linear dimension
substantially greater than the maximum ·dimension of the larg·er·pores.
Thus, the critical Griffith flaws in coke should now be regarded as
consisting of the series of pores and interconnecting microcracks which
constitute these extended microcracks.
-- - - ------------------
- 78-
3.3 SEX texture and coke strength prediction
3.3.1 Introduction
The differences in the mode of fracture of the various coke textural
components evident from the examination of fractured coke surfaces
( Section 3.1 ) imply variations in their strengths and in their
contribution to the strength of coke from blended-coal charges. Any such
effect would be dependent upon the proportions of the various components
present. In support of this view, for a series of twenty-five cokes, a
multi-linear regression equation was obtained which highly correlated the
coke tensile strength to the determined textural composition (132). The
desirability, in industrial situations, of a method of coke strength
prediction has been explained previously. A relationship of this type, in
itself, is inadequate for predictive purposes. However, in cokes made from
blended-coal charges, components originating from individual blend
constituents are usually readily identifiable (124), the implication being
that the textural composition of the cokes is an additive property of the
blend components. The objective of this part of the study was to
investigate whether this view could form the basis of a method of coke
strength prediction. In concept the approach adopted was unreservedly
empirical. It was based on the idea that, if coke textural data is, or
could be assumed to be, additive, then any relationship obtained between
the strength and textural composition data of cokes from hlended coal
charges, calculated from corresponding data for cokes obtained from
individual blend components, could be used for predictive purposes.
Accordingly, using six coals, two- and three-component coal blends were
carbonized in the small pilot oven and relationships were sought between
the measured coke tensile strengths and the textural compositions of the
cokes calculated from the textural behaviour" of the blend "components;
- 79-
The calculations described in this section were carried out using
commercial software on either Commodore 8032 personal or Multics
mainframe computers, or software written by the author in Pascal for use
on an Amstrad PCW8256 computer. This section concentrates on the attempt
to investigate an empirical coke strength prediction method, further
consideration of the form of the relationships used being reserved for
the General Discussion.
3.3.2 Experimental procedures
The coals used were those listed in Table 6. The carbonization procedures
and the methods of determining the tensile strength and textural
compositions of the cokes were described earlier ( Section 3.1 )
3.3.2.1 Blends carbonized
The two- and three-component blends carbonized are listed in Table 11,
the blend compositions being quoted in fractions by weight of air-dried
coal. Their compositions lie on one or more of the five triangular
diagrams in Figs 37a-41a at the centres of the circles bearing the blend
number. On each diagram, the identity of the coals used in the blend are
indicated by the letter in 'the circle at each corner, ie. at the position
corresponding to a charge of unblended coal. The three component blends
contained various proportions of 'the coals A-C-F, A-E-F, A-B-E, A-C-E and
A-D-E. Such blends are of two types; mixtures of low- and high-volatile
coals ( A-E-F and A-D-E blends) or of low-, medium- and high-volatile
coals ( A-C-F, A-B-E and A-C-E blends ). As Figs 37a-41a indicate, the
blends had volatile matter contents of 24, 27, 30 or 33 wt% ( dafb ).
Both the range of volatile matter content and the type of three-component
blend used reflect' industrial practice in the U.K.; but overall' the range'
of blends carbonized exceeds that used commercially. In all, forty-four
blended-coal charges were carbonized.
- 80-
3.3.3 Results
For the single-coal carbonizations, the determined and calculated
fractional coke yields ( wlw ) and- the tensile strengths of the cokes are
given in Table 7. The calculated coke yields were obtained from the
analytical data given in Table 6 according to the relationship
Z= ( 1- I (A+M) 1100) ) ( 1- IV 1100) )+ A 1100 <25>
where M and A are the air-dried moisture and ash contents of the coals
and V the volatile matter content on a dry-ash-free-basis, all values
being in percent by weight. The SEM textural compositions of these cokes
have been given in Table 9, the textural components being identified by
their initials as listed in Table 8.
All the calculations described below were carried out using textural data
calculated to three decimal places and, so that the calculations can be
repeated precisely, all textural data in this thesis are reported to this
accuracy. It is acknowledged, however, that the errors associated with the
measurement of textural data do not justify such precision.
Experimental data for the blended-coal charges are given in Table 11.
Values listed include the blend composition quoted in terms of the
fractional content by weight of air-dried coals, the measured ana
calculated fractional coke yields and the coke tensile strength. The two
calculated coke yields ( Yb and Zb ) were obtained from the blend
composition and the measured ( Y ) and calculated( Z ) fractional yields
of the single-coal cokes, ie.
<26>
<27>
- 81 -
where F. is the fractional content by weight of the ith coal in the
blend, y, and Z. are the measured and calculated fractional yields of the
ith single-coal coke. Clearly, the measured yields of single-coal cokes
provide the more accurate method of estimating the yield of cokes from
blended-coal charges, the mean absolute difference between the measured
and calculated yields being 0.007 w/w. Calculation of the coke yields
from analytical data consistently underestimates the yield, the average
underestimation being 0.038 w /w.
3.3.4 Discussion
To investigate the proposed method of coke strength prediction, the
textural compositions of cokes from blended coal charges were calculated,
by computer, using the following general relationship :
" Ti= l: T..kFk,Ck
\.;. ~ I
q ,
.l: l: Ti.kFkCk t~ \ ><':"1
(28)
where T. is the fractional content of the ith textural component in the
coke from a blended-coal charge, T •. k is the fractional content of the i th
textural component in the kth Single-coal coke, Fk is the fractionai
content ( air-dried basis ) of the kth coal in the blend and Ck is a
correction factor for the kth coal.
The correction factor is equal to 1, Y or Z ( as given in Table 7 )
depending on whether textural data is calculated from the concentration
of the coals in the blend ( Method C ), or coal concentrations corrected
for the measured ( Method Y·) or calculated' ( Method V ) fractional
yields of the single-coal cokes. Clearly, no blend contained more than
three coals but computer program writing was simplified by adopting the
general form of equation <28>. The lower term in the equation, essential
- 82-
only when using correction factors less than one, is necessary to correct
the textural data to a total fractional content of unity.
The calculated textural data for cokes from the blended-coal·charges are
given in Tables 12 to 14. Data calculated using yield corrections are
considered to reflect more accurately the presence of the single-coal
cokes in the coke from the blended-coal charges. Thus, because of the
higher coke yields of high-rank coals, such data (Tables 13 and 14 )
contain higher calculated contents of textural components characteristic
of cokes from high-rank coals than corresponding data calculated without
yield correction ( Table 12 ). The effect is emphasized slightly by the
higher volatile coals having slightly higher moisture contents than the
other coals and the fact that the calculation of textural data by Kethod
C involves no moisture correction.
To relate the tensile strength of coke from blended charges with
calculated textural composition data, relationships of the following form
were investigated
1 s= K+ r A,T, <29) , ..
.
, s= r A,T, <30)
(:\
q 9 s= r r A, . k T i Tk <31>
L=t If",!
<32)
<33)
- 83-
T s= ~ A, T,:;: If R (34)
, "
7
s= ~ A,(T,+ T,"'11 R) (35) l':.\
where S is the tensile strength of coke from a blended coal charge, K is
a constant, T, is the calculated fractional content of the ith textural
component in the coke and A, is the corresponding coefficient in the
equation. Tk is the calculated fractional content of the kth textural
component in the coke ( i may equal k ) and A, ,k is the coefficient
corresponding.to the product T,Tk . The ratio 1/R is the ratio of the
calculated contents of inerts to reactives ( non-inerts ) present in the
coke.
For hypothetical cokes containing only two textural components, the above
equations would take the following forms :-
S= Kt A,T,t A2T", (36)
(37)
(38)
(39)
(40)
(41)
(42)
- 84-
In practice, it proved impossible, using equations (32) and (34) to
obtain any adequate fit between measured and calculated coke tensile
.strengths. Thus these equations are not considered further. The
coefficients in equations (29) and (30) were.obtained using commercial
multi-linear regression analysis software on Commodore PET and Multics
mainframe computers respectively. These two equations will be referred to
as the MLR<29) and NOKMLR(30) equations respectively, the latter name
reflecting the fact that the equation did not contain a constant. For
reasons which be will fully explained in the general discussion, equations
(31), (33) and (35) are referred to as the INTER(31), TRANS(33) and
INERT<35) equations respectively.
The values of the coefficients in the MLR(29) equations, derived by
computer analysis for the three sets of calculated textural data, are
given in Table 15, while Table 16 compares the measured coke tensile
strengths for the forty-four cokes from blended-coal charges with, for
each coke, the three calculated values. It is evident from Table 15 that
the coefficients in the three MLR equations differ. markedly, yet, as Table
16 shows, the three equations predict very similar tensile strength
values for each coke and the degrees of fit between measured and
calculated strengths, indicated by the standard errors of estimation
quoted in Table 15, were almost identical. Furthermore, since the three
sets of calculated textural data in Tables 12 to 14 are quite similar, it
is possible to use any equation with any set of calculated textural data
without materially reducing the degree of fit between measured and
calculated strengths. To illustrate the degree of fit between measured and
calculated strengths implied by a standard error of estimation of 0.443,
Fig. 42 contains a plot of measured strengths against those calculated
using the MLR<29) equation for textural data calculated using method Y.
Although the MLR<29) equations permit the calculation of coke tensile
strengths from calculated textural data with some precision, because of
the variation in sign ( positive or negative ) of the coefficients in the
- 85-
equations, they permit neither the ready identification of the textural
types contributing most strongly to coke strength nor the nature of the
coal blends likely to give good quality coke. Use of the NOKMLR(30)
equation was therefore investigated in a preliminary attempt to overcome.
this problem.
In fact, as Table 17 indicates, the result was unsatisfactory. Again, no
consistancy was observed amongst the three equations in the signs of the
coefficients associated with any particular textural component. In
addition, this software package eliminated one or two textural components
( different ones from each set of calculated textural data ) from
consideration on the grounds that their fractional contents correlated
highly with those of other textural components. However, the standard
errors of estimation obtained showed some improvement to approximately
0.39 MPa. Table 18 compares measured coke tensile strengths with those
calculated using the NOKMLR(30) equations.
Attempts were therefore made to fit the INTER(31), TRAHS(33) and
INERT(35) equations to the data in such a way that all the coefficients
were positive. Since no commercial software could be found which would
meet this requirement, software was written in Pascal for use on an
Amstrad PCW8256 computer. ·The limitations of the algorithm used and the
essential features of the program are described in Appendix I.
For nine textural components, to fit the INTER(31) equation to the
tensile strength and calculated textural composition data with accuracy
would involve the calculation of the values of forty-five coefficients. A
single run of the Pascal program in Appendix 1 for so many coefficeints,
assuming each could take one of three values, would take 10'9 hours.
Thus, while recognising that the procedure could influence the precision
of the fit obtained, it was decided to shorten run times by assuming that
the values of som·e coefficients were equal. How this was achieved will be
explained later when the derivation of the INTER(31) equation is
-86-
discussed. This process reduced the number of terms considered to
nineteen but establishing even approximate values for these was a time
consuming process.
The values of the INTER(31) equation coefficients obtained using the
three sets of calculated textural composition data are listed in Table 19.
For each set of calculated data, the figures at the intersection of each
row and column is the coefficient associated with the product of the
fractional contents ( T ) of the components, identified by their initial
letter. Thus, for textural data obtained using method C, the value 9.5 at
the intersection of the intermediate ( I ) row and the coarse-granular
( Gc ) column is the coefficient associated with the cross-term T,.Tac .
Similarly, the value 8.2 in the column to the left is the coefficient
associated with the squared term TI2. All values falling along the
diagonal from top left to bottom right of the tables are associated with
squared terms and hence individual textural components.
It is evident from Table 19 that the values of the INTER(31) equation
coefficients obtained for the three sets of calculated textural data are
almost identical, as indeed are the standard errors of estimating the
coke tensile strengths from the three equations. Apart from the
coefficients associated with Tac2 , which has a value of only five, high
values of the coefficients lie within the odd-shaped area outlined near
the centre of each table. Generally, these are associated with lamellar,
intermediate and coarse- and medium-granular textural components.
However, since the coefficients corresponding to products of these
components ( cross-terms ) are larger than the squared-term coefficients
for individual components, the equation implies that cokes containing all
the four components would have a higher strength than one containing a
large amount of one component. This is in accord with the finding that
cokes from prime-coking coals, which are usually strong, contain a
complex mixture of textural .types [Ill. Thus, although the INTER(31)
equation is not able to calculate the coke tensile strengths with a
,- 87-
precision equal to that of the MLR and NOKMLR equations, in contrast to
these equations, it does appear to offer a plausible explanation of the
textural requirements necessary for high strength cokes.
The number of coefficients used in the TRANS(33) and INERT(35) equations
were eight and seven respectively. In the former case this was because
the inert components were treated as a single textural category. In the
INERT(35) equation, the inert content was incorporated into the complex
term associated with each textural component. The values of the
coefficients obtained using the two forms of equation and the three sets
of calculated textural data are given in Tables 21 and 23, while the
measured and calculated coke tensile strengths are compared in Tables 22
and 24. As was previously observed for the INTER(31) equation; for both
the TRANS(33) and INERT(35) equations the coefficients obtained using
the three sets of calculated textural data were quite similar, as were the
standard errors of estimating the coke strengths from the three
equations. Both equations ranked the contribution of the textural
components derived from reactive coal constituents to the coke tensile
strength in the order Intermediate > Medium granular ) Lamellar > Coarse
granular ) Fine granular ) Very-fine granular > Flat. In fact, the values
of corresponding
similar. At 0.461
coefficients in the two sets of equations were very
and 0.453' MPa respectively, the mean values of the
standard errors of estimating the coke tensile strengths from the
TRANS(33) and INERT(35) sets of equations were slightly lower than that
obtained with the INTER(31) equation. Providing the textural compositions
of the cokes from the single coals are known, on the basis of the
magnitude of the coefficients in the INTER<31>, TRANS(33) and INERT(35)
equations, ready identification of those coals capable of producing high
strength cokes is possible. Being Simplest in form, the TRANS(33)
equation is probably best suitable for this purpose;
It is now evident that, for the forty-four cokes considered, the coke
tensile strengths can be related to calculated textural data with
-88-
reasonable precision using five different equations. Differences in
precision arising from the three methods of calculating the textural data
and between the five equations being relatively small, choosing between
them becomes a matter of judgement. Of the five equations, the MLR(29)
equation produces the lowest standard errors of estimation while using
the fractional contents of all the textural components present in the
cokes. Also, it is considered that the textural data calculated using a
correction based on fractional yields by weight of the single-coal cokes
( the Y data ) should be the most accurate. Accordingly, this equation
and set of calculated textural data were used in an alternative method of
comparing measured and calculated coke tensile strengths.
For the forty-four cokes studied, the tensile strengths are shown on the
triangular diagrams in Figs 37b-41b, the coke tensile strength being
placed at a position corresponding to the composition of the coal blend
from which the coke was produced. Also shown on the triangular diagrams
are straight dotted lines linking the composition of blends giving cokes
of specified calculated strength. These iso-strength lines were obtained
from the relationship, derived in Appendix 11 between the fractional
concentrations by weight ( F, and F2 ) of coals 1 and 2 in the blend.
The figures demonstrate the correspondence between the measured and
calculated strength values; the degree of fit between measured and
calculated tensile strengths being as expected on the basis of Fig. 42
and the standard error of estimation associated with the MLR(29)
equation.
Thus, this study has shown that several equations, differing in form, can
be used to relate measured coke tensile strengths to textural data
calculated from yhe blend composition and the textural composition of the
single-coal cokes. The errors involved in calculating the coke strengths
from the.equations are approximately twice those associated with the
measurement of coke tensile strength by the diametral-compression method.
At the very least, these equations provide the basis of a method of
-89-
predicting the strength of coke, produced in the small-pilot oven under
the conditions specified, from any blend containing the six coals used in
this investigation. Further study of the method should involve
investigating the general applicability of the relationships. However, the
relationships were derived using coals covering a wide range of coal rank
and the whole range of textural composition likely to be encountered in
commercial coke-making in the U.K.
Two possible modes of use of the equations in coke quality prediction can
be envisaged. To achieve strong cokes from any combination of available
coals would first involve the elimination of less-suitable coals on the
basis of the textural composition of the cokes obtained from them and the
magnitude of the coefficients in the TRANS equation. For potential blends
containing three coals, plotting iso-strength lines, calculated from the
MLR equation, on triangular diagrams would help to identify blend
compositions giving high strength cokes and strengths of these could be
confirmed by direct computation. More than three coals could be
accomodated provided some were to be used in fixed proportions. An
alternative problem often encountered is to achieve a sensible balance
between adequate coke strength, low sulphur content and low cost. For
blends containing three coals, this could be achieved by comparing, using
overlays, triangular diagrams bearing iso-strength, iso-sulphur content
and iso-cost lines.
- 90-
3.4 PLM texture and coke strength prediction
3.4.1 Introduction
The studies described in Section 3.3 demonstrated the feasibility of
using coke textural data, determined during the SEM examination of etched
coke surfaces, as a basis of predicting, from the behaviour of individual
blend components, the strength of coke produced from blended coal
charges. However, since scanning electron microscopes are seen in coal
and coke quality control laboratories considerably less often than
optical polarizing microscopes, and since, as explained in Section 3.1,
textural components revealed by SEM examination can be identified with,
although probably not with one-to-one correspondence, components visible
under polarized light, it seemed worthwhile to establish whether the
method of coke strength prediction developed for use with SEM textural
data could be applied equally well to corresponding data obtained using
polarized light microscopy ( PLM ). This study formed one part of the
work described in this section.
As described in the Literature Review, schemes for the classification of
coke textural components visible under polarized light are many and
varied. However, since a series of laboratory cokes, made from coals
covering a wide rank .range, whose PLM textural composition had previously
been determined using one classification scheme were available, it was
decided to use this scheme in the present work. It was later considered
necessary, however, to modify the classification of components prevalent
in cokes obtained from high-rank coals. The reasons for this have been
described in detail elsewhere [1331. The modified scheme is described
below.
Using this scheme, in addition to determining the PLM textural
composition of the six single-coal cokes produced in the small pilot
oven, data for the forty-four cokes prepared from blended-coal charges
were also obtained. This permitted the investigation of the additivity of
textural data.
-91-
3.4.2 Experimental procedures
3.4.2.1 Determination of PLK textural data
To prepare,coke samples for examination under polarized light, they were
first crushed gently to maximise the yield of material in the 120-600~m
size range. After cleaning ultra-sonically and drying, the coke grains
were mixed with epoxy resin and formed into a 15mm diameter pellet.
This was then embedded into further resin and cured to form a 25mm
diameter by 10mm thick block. The upper, coke-bearing surface was then
polished, using conventional methods, to give a scratch-free, highly
reflecting surface.
PLK textural data were then determined using a Leitz Ortholux polarizing
microscope. Crossed polars, together with a full wave retarder plate, were
used to impart colour to the image and a X100 air objective and X10 eye
pieces were used to give an overall magnification of X1000. Textural data
quoted are based on the examination of 500 positions on the coke surface
of each block. At each position, the textural component present under the
cross-wires was allocated to one of the eleven textural categories
described in Table 25. The appearance of the textural components under
these conditions is illustrated in Figs 43 to 52. The appearance of small
inerts is not shown. They consisted of identifiable fragments of large
inerts ( Fig. 52 ). A Swift mechanical stage and electronic counter were
used to position the block and to accumulate the data. The sizes quoted
in the table for the various mosaics are mean values obtained from a
total of three hundred measurements taken from projected images of six
transparencies, each showing material typical of the component
considered. From each transparency, measurements were taken from
isochromatic areas believed to represent the basic unit constituting the
mosaic.
3.4.3 Results
PLM textural composition data for the six pilot-oven cokes produced from
single-coal,charges are listed in Table 26. The data is quoted in units
-92-
of volume fraction. However, as was explained earlier in connection with
SEM textural data, it is assumed that the densities of the various
textural components are equal so that the composition in volume fraction
is assumed to be equal to that in weight fraction. Future tables will
therefore refer to textural data simply as fractional compositions.
For these cokes, PLM textural data, recalculated to an inert-free basis,
are compared with corresponding SEM textural data in Fig. 53. For this
purpose, the broad- and striated-flow categories have been considered as
a single category labelled flow. For the lamellar/flow,
intermediate/granular-flow and granular/mosaic components, the shapes of
the histograms in the figure confirm the anticipated general
correspondence between the textural components observed by the two
techniques, the differences in textural composition between the two
methods of analysis being explicable in terms of minor differences in the
position of boundaries between components. It was originally thought that
the flat component identified using scanning electron microscopy
corresponded to an anthracitic component. However, using polarized light
no anthracitic components were observed. It is therefore now believed
that the material identified in these cokes as flat is a flow/lamellar
material whose lamellae are aligned almost parallel to the etched surface.
When viewed under polarized light, such material would be classified as
flow material unless the lamellae were aligned exactly parallel to the
polished surface in which case it would appear isotropic.
Measured PLM textural data for the forty-four cokes obtained from
blended-coal charges are given in Table 27.
3.4.4 Discussion
To investigate the general applicability of the method of predicting the
strength of cokes from blended-coal charges from the behaviour of
individual blend components, to textural data measured using polarized
light microscopy, textural data were first calculated for the forty-four
cokes produced using blended coals, from the blend composition and the
measured textural data for the six single-coal cokes. The same three
- 93-
methods of calculation, ie C, Y and V, used when calculating SEI>! textural
data, were again employed. These data were then subjected to computer
analysis to obtain coefficients in the MLR(29) and TRANS(33) equations.
These two equations were chosen for this purpose since, using SEM
textural data, they respecfively gave the best lit between measured and
calculated coke strengths and permitted ready identification, from coke
textural data, of those coals producing high-strength cokes.
Calculated PLM textural data for the cokes from blended-coal charges
calculated using methods C, Y and V are listed in Tables 28-30
respectively. Coefficients obtained by multi-linear analysis of the three
sets of data·are given in Table 31, while measured tensile strengths and
those calculated from the MLR(29) equation are compared in Table 32. As
was observed when using SEM textural data, the coefficients in the
equations obtained using the three sets of calculated data differed
widely. The standard errors, given in Table 31, of estimating the tensile
strength from the equations all lie close to 0.42MPa and are slightly
lower than corresponding values previously obtained using SEM textural
data. Clearly, PLM textural data can also be used in this approach to coke
strength prediction.
In deriving coefficients for the TRANS<33> equation, the inert components
were again treated as a single textural category. The reiterative computer
algorithm previously described was again used to obtain the coefficients.
These are listed in Table 33 for the three sets of calculated textural
data together with the standard errors of estimating the coke tensile
strengths using the equations. Measured and calculated coke tensile
strengths are compared in Table 34. As was observed for SEM textural
data, for each textural component considered, the coefficients in the
three equations were similar. However, the standard errors of estimation,
approximately 0.38MPa, were the lowest yet calculated. The coefficients
ranked the contributions of the textural components to coke stength in
the order Granular flow > Coarse mosaic > Medium mosaic > Striated flow
> Fine mosaic > Broad flow > Isotropic > Inerts, although differences
between the last four components were relatively minor. This ranking is
similar to that observed for corresponding SEM textural components. For
- 94-
PLH data therefore, this equation appears to be the most suitable both as
a basis of coke strength prediction and also as a means of identifying,
from coke textural data, those coals capable of producing high-strength
coke.
Regarding the question of additivity of coke textural data, all three
methods of calculating the PLM textural data used in the equations
derived above were based on the assumption that textural data for cokes
from blended-coal charges were additively dependent upon the blend
composition of the coal charge and the textural data for the single-coal
cokes. Differences between the measured PLM textural data given in Table
27 and the three sets of calculated data listed in Tables 28 to 30
provide a measure of the validity of the assumption. The mean values of
the absolute differences between measured and calculated textural data,
averaged over the forty-four cokes from blended-coal charges and the nine
textural components considered were 0.0357, 0.0354 and 0.0356 for the C,
Y and V methods of calculation respectively. Thus the differences between
the three methods of calculation appear to cancel out so that, from this
point of view, no one method has any marked advantage over the others.
Further examination of Tables 27 to 30 shows that departures from
additivity, indicated by the sizes of the differences between measured
and calculated textural data, vary depending upon the particular textural
component considered and the rank of the coals present in·the blend. This
point is evident more clearly on examination of Table 35 which lists, for
all textural components in cokes from all blended charges, the
differences between measured textural data and that calculated according
to method Y. These differences mean that blending has induced changes
during carbonization such that, for example, a proportion of the vitrinite
in coal A instead of .forming striated-flow components forms broad-flow
material. It is assumed in the following analysis that a vitrinite which,
in a single-coal carbonization, forms a particular textural component, in
a blended-coal charge can normally be influenced only to the extent of
forming a component adjacent to it in the classification scheme, ie. a
medium-mosaic forming vitrinite can be influenced to form either coarse
or fine mosaics but neither granular-flow nor isotropic material.
- 95-
To assist in obtaining a qualitative appreciation of the influence of
blending coals on the textural composition of the cokes produced, the
differences listed in Table 35 have been grouped together according to
the identities of the coals in the blend and averaged over the number of
blends containing those coals. Thus, coal A was carbonized in four two
component blends with each of the three coals D, E and F. For each type
of blend, A-D, A-E and A-F, the measured minus calculated textural data
were averaged over the four blends. This data, together with data
similarly obtained for the other two- and three-component blends studied
are listed in Table 36.
For cokes from two-component coal blends, histograms in Figs 54 to 56
illustrate the effect of blending on the coke textural compositions. In
each figure, histograms in the upper part of the figure show the textural
composition of the cokes obtained from the individual coals used in the
blends, while immediately underneath is a histogram showing, for each
textural component, the departures from additivity, as measured by the
differences between the measured and calculated textural content listed in
Table 36, which result from blending. The letters A to F on the
histograms indicate the coal or blend considered. Hatched bars indicate
that the direction of the departure from additivity was observed for all
the blends examined. For example, in Fig. 54 are shown the data obtained
from the low-volatile coal A and the medium-volatile coal C and blends
therefrom. The histogram at the bottom of the figure shows that the coke
produced from the blend contained higher proportions of broad- and
granular-flow components and lower proportions of striated-flow and
coarse mosaic components than would be expected on the basis of
additivity. Examining the magnitude of the departures from additivity,
this result can be achieved if, on blending, (1) broad and granular-flow
components were produced instead of some striated-flow material and (2)
granular-flow components were produced instead of some coarse mosaics.
In this particular instance, cokes from both coals A and C contain both
coarse mosaic and striated-flow components so it is not possible to say
which coal was influenced to the greatest extent. Nevertheless, the
effects observed can be accounted for by the changes indicated by the
arrows drawn on histograms of the textural composition of single-coal
- 96-
cokes. Thus an arrow pointing to the right indicates a downgrading to a
smaller-sized textural component and conversely one pointing to the left
indicates an upgrading to a component of larger size.
The histograms in Fig. 55 illustrate the effects observed on blending the
low-volatile coal A with the high-volatile coals D, E and F. The high
volatile coals decrease in rank from D to F. In these blends, in contrast
to the behaviour previously described, both broad-flow and striated-flow
components are down-graded to form striated-flow and granular-flow
components respectively. For blends of coals A and E, some downgrading of
granular-flow to coarse mosaic also appears to occur. The effect of
blending on the constituents of high-volatile coals is to upgr·ade
isotropic and mosaic components but the detailed changes depend on the
rank of the coal and the textural components present in the single-coal
coke. Thus for coal F, isotropic and fine mosaics are upgraded while for
coal E medium mosaics are additionally upgraded. The coke from coal D
contains little isotropic and fine mosaic material and in this case
medium and coarse mosaics are upgraded.
The effects on coke textural composition resulting from blending medium
and high-volatile coals are illustrated in Fig. 56. Again arrows on the
histograms of the textural ·composition of the single-coal cokes indicate
the direction of the changes necessary to bring about the effects
observed on bl:nding. These show that, as was observed for blends
involving coal A, blending results in the upgrading of isotropic and fine
mosaic material in cokes from the high-volatile coals E and F to fine and
medium mosaic material respectively. For the medium volatile coals, Band
C, blending primarily influences the striated-flow component. For both of
these coals, a proportion of this is upgraded to broad-flow while for
coal C, an additional proportion is downgraded to granular-flow. It
therefore appears that the upgrading of striated-flow components to broad
flow, previously noted for blends of coals A and C, is a phenomenon
associated with the presence of medium-volatile coal in the blend.
Thus, the departures from additivity of textural composition observed
when two coals of different rank are carbonized together, under these
- 97-
conditions, appear to conform to a pattern, albeit complex. The textural
composition of the cokes from the blended coals indicates that, when
high-volatile coals are blended with a coal of higher rank, whether low
or medium-volatile, upgrading of the smaller-sized mosaic components
formed from the high-volatile coals occurs. The effect of blending on the
textural components formed from the higher-rank coal depends on its
volatile matter content. Thus, if a low-volatile coal is the second blend
component downgrading of both broad- and striated-flow material occurs.
When a medium-volatile coal is the second blend component, flow
components behave differently. Upgrading of a proportion of striated-flow
material to broad-flow is consistently observed while an additional
proportion is downgraded to granular-flow material.
Regarding three-component blends, as Table 36 shows, for blends of one
low- and two high-volatile coals the directions of departures from
additivity, as indicated by the sign associated with the value for each
textural component, were generally in accord with the pattern outlined
above. Also, in the three groups of blends containing low-, medium- and
high-volatile coals, ie. A-B-E, A-C-E and A-C-F, the behaviour of
components from the low-volatile coals was generally as expected.
However, for only the first two groups of such blends were the directions
of the departures from additivity for flow components consistent with the
presence of a medium-volatile coal in the blend. In blends of coals A-C
F, both broad-flow and striated-flow components were downgraded. Thus as
a result of this variation in the behaviour of the flow components
depending on the nature of the other coals in the blend, attempting to
predict the departures from additivity for three-component coal blends
appear especially difficult where a medium-volatile coal is involved.
For coke strength prediction purposes, it· does not appear absolutely
essential that measured textural data for single-coal cokes are used to
calculate the textural data for cokes from blended-coal charges. In
principle it appears equally feasible to use data calculated from measured
data for a small number of blended-coal cokes. To investigate such views,
textural composition data for cokes produced from ten three-component
coal blends ( numbers 5, 11, 13, 17, 21, 25, 29, 33, 39 and 43 ie. two
- 93-
such cokes from each triangular diagram in Figs 37 to 41 ) were first
used to obtain notional textural composition'data for single-coal cokes.
Coal blend compositions were used in this calculation without correction
for the yield of coke from the single-coal carbonizations. The data
obtained is given 'in Table 37. This data was then used to calculate
textural composition data for cokes from blended-coal charges using
equation (28) ( page 81 ) and method C. The data so derived is listed in
Table 38. The mean absolute difference, averaged over the forty-four cokes
and nine textural components, between the measured textural data in Table
27 and the calculated data listed in Table 38 is 0.023. Thus, as could be
anticipated, calculating textural data in this way is more accurate than
using the measured data for'the Single-coal cokes when the corresponding
difference was 0.036.
The calculated textural data for cokes from blended-coal charges in Table
38 were used to derive coefficients in a strength/texture relationship
based on the TRANS(33) equation. Those which gave the lowest standard
error of estimation (0.44MPa) are listed in Table 39 while the calculated
coke tensile strengths are compared with measured values in Table 40.
Although the precision with which this equation predicts the coke tensile
strengths is less than that associated with the corresponding equation
with coefficients derived using textural data calculated from measured
data for the Single-coal cokes, to obtain a strength/texture relationship
in this way does provide a useful alternative method of coke strength
prediction. The potential value, in different situations, of the various
equations for coke strength prediction is discussed later.
4. GENERAL DISCUSSION
The individual experimental sections contain some discussion of the
results described therein. The ,object of this General Discussion is to
consider some wider aspects of this work especially where these involve
the findings of more than one experimental section. Four basic topics are
covered. The nature of the textural components in metallurgical coke are
first discussed before reviewing, in the light of previous work, the
influence of coal blending on coke textural composition. The extent to
which coke tensile strengths can be regarded as being causally dependent
on coke texture is considered and this leads to the identification of
further methods of coke strength prediction. The potential value of the
various prediction methods, developed in this work, in different
situations are then reviewed. Further studies, of both a scientific and
technical nature, necessary to extend this work are recommended.
4.1 The nature of coke textural components
Since the pioneering work of Brooks and Taylor (69) considerable research
effort has been directed towards extending and applying their views on
the development of optical' anisotropy in carbons and in the use of
polarized light,microscopy to investigate their structure. When the
intensive study of extinction contours, as developed by White and co
workers (71), is applied to the study of polished surfaces of carbons,
then the coloured or shaded images evident can be interpreted to
elucidate the structural form of the textural units comprising the carbon.
The present studies demonstrate that, for cokes from coals, in most
instances such structural information can more readily, directly and
clearly be obtained by viewing, in partial three-dimensional form,
fractured and etched surfaces in a scanning electron microscope. The
present work is restricted to studies of metallurgical cokes but the
value, in classifying the textural forms of carbon, of the complementary
information obtained by examination of the two types of surface in an
SEK has been confirmed for carbons obtained from a wide range of'
precursor materials (134).
-100-
It is not however suggested that scanning electron microscopy should be
used to the exclusion of polarized-light studies. Rather the two
techniques should be regarded as complementary. Indeed, during this work
it has been observed that, when those textural components classified as
coarse or medium granular, from SEM examination, are studied under
polarized light at high magnification, the changes in shape of the
isochromatic areas in the mosaic components and the corresponding
movement of extinction contours imply that such materials are not akin to
grains bounded by interfaces, as is implicit in the choice of the name
granular and as had been suggested by Marsh [88). Instead, it appears
that they are continuous lamelliform structures similar to those in
lamellar/flow units, although the smaller size of the isochromatic areas
in mosaic material implies smaller areas of commonly oriented lamellae.
Fine-mosaic components are too small for extinction contour movement to
be visible but it seems reasonable to suppose that they are similarly
constituted.
From X-ray data on 1000'C laboratory cokes prepared from widely
differing coals [135), it has been shown that the La and Lc dimensions of
the constituent graphitic crystallites remain almost constant, at
approximately 3 and 1nm respectively, over the whole range of coal rank.
On the basis of these values, the carbon structure of metallurgical coke
can be interpreted in two ways. Coke carbon can be regarded as being
composed of small, near-perfect graphitic crystallites separated by
disordered carbon ( Fig. 57 ). An alternative view would be that more
extensive graphitic layers are present but that these are distorted and
defected and possibly contain foreign atoms. The mutual alignment of
these layers is such that only small volumes are sufficiently well-
. ordered to give X-ray difraction patterns ( Fig. 58 ); hence the low La
and Lc dimensions. It seems unlikely however that these layers exceed the
27nm value obtained by the author for the Lc dimension of metallurgical
coke heated to 3000·C. Even this crystallite dimension is an order of
magnitude smaller than the dimensions of textural components in
metallurgical coke measured during this work. One way in which X-ray and
microscopic measurements can be reconciled is to consider that all
textural units are composed of small crystallites, whether· perfect or not,
-101-
and that the variation in the textural unit size, represented by
isochromatic areas evident under polarized light, reflects the extent to
which these individual crystal lites adopt a common orientation. The effect
is illustrated. in simplified form in Figs 59 and 60 for a mosaic and a
flow structure respectively, the crystallites being represented by small
dashes. On this basis, it is considered that the boundaries between
" isochromatic areas are positions w?re the alignment of the crystal lites
changes. Thus, on the smallest scale the carbon in metallurgical coke is
considered to be composed of individual crystallites separated by less
well-ordered carbon but on a scale relevant to considerations of
microscopical study, all structures can be regarded as continous.
Fracture surfaces of lamellar/flow components in coke are clearly the
result of translamellar breakage but intergranular failure can no longer
be supported as a mechanism of breakage of granular/mosaic components.
However, rather than invoking translamellar failure for these components,
it is suggested that the appearance of the fracture surface and the
tortuosity of the fracture cracks in granular/mosaic material can best be
explained in terms of cleavage parallel to the aligned constituent
crystallites.
In Figs 59 and 60, within each isochromatic area, the dotted lines
indicate the general orientation of the component crystallites, the
gradual change in orientation considered to occur at an interface between
isochromatic areas being shown in the inset to Fig. 59. The colours on
both figures are derived from the variation of colour with basal layer
orientation relative to the vibration direction of the incident light,
observed when using crossed polars and a A-retarder plate, as described
earlier, and as indicated in the circular diagram included in Fig. 59. On
this circular diagram, the orientation of the layers which give rise to
extinction contours are shown by the black lines at the centres of the
purple segments.
On the illustration of a mosaic structure in Fig. 59, extinction contours
are shown by black lines at the interface between certain isochromatic
areas. Only at those interfaces where the orientation of the crystallites
becomes parallel or perpendicular to the vibration direction is an
-102-
extinction contour developed. Thus, the interface between blue and yellow
areas, eg along A-A, will always be a position of an extinction contour.
For other interfaces, extinction contours may, or may not, occur depending
on the particular orientation of the crystallites at the two sides of the
interface. It is also evident, eg along A-B, that extinction contours may
occur within a purple area, although a difference in shading, not shown
on the diagram would be detectable.
An illustration of a flow structure is shown in Fig. 60. The upper
diagram again shows that changes in crystal lite orientation do not
necessarily result in changes of colour or induce extinction contours.
Thus the change in orientation along A-A- induces neither colour change
nor extinction contour, while that along A-B induces a colour change but
no extinction contour. The lower figure illustrates the result of rotating
the polarizer, analyser and tint plate through 30' in a aRticlockwise
direction. ( This is equivalent, in terms of colour changes, to rotating a,,,,{,;
the microscope stage 30· in a~clockwise direction, except that the
orientation of the field of view is unchanged. ) Colour changes take place
and an additional extinction contour ( along C-C ) is formed. As pointed
out earlier, for this type of material, it is clearly much easier to
interpret the structure from an electron micrograph of an etched surface
similar to that shown in Fig. 29.
-103-
4.2 The influence of coal blending on coke texture
Before considering the effects of coal blending on coke texture, some
further comments on the mode of formation of textural components in
metallurgical coke are worthwhile. Re-.examination of cokes from high rank
coals suggested that the component originally termed 'flow' contained
disparate textural types. Accordingly, for this work, this category was
subdivided into patterned-anthracitic, broad-flow and striated-flow in
the hope that a simpler explanation of their mode of formation would
ensue (133]. The anisotropy exhibited by patterned-anthracitic material is
not now regarded as being associated with flow during carbonization, no
evidence of softening, ego pore formation, being observed in such
material. However, it still appears that both broad- and striated-flow
components can be formed directly or via the intermediate formation of a
form of fine mosaic component from vitrinite exhibiting basic anisotropy.
Thus, despite the restricted definition of such components used during
the present work, their mode of formation clearly does not easily fit
into the general framework for the formation of graphitizing carbons
described earlier. The view that these structures are formed as a result
of relatively minor changes to the structures present in vitrains from
high-rank coals (84] seems the only tenable explanation. Despite their
different mode of formation, these flow structures and the structures,
similar in appearance, formed during pitch carbonization by the·
coalescence of large mesophase spheres are considered to be composed of
similar extensive lamelliform units. The similarity of the etched surfaces
of metallurgical coke in Fig. 25a and pitch coke in Fig. 29 confirms this
view. In fact the metallurgical coke appears to possess a higher degree
of order.
As explained in Section 2.4,· it is here considered that the larger mosaic
and granular-flow structures on the one hand and the larger flow
components on the other are formed by different routes. Nevertheless, the
textural components from fine mosaic to broad-flow are now considered, on
a microscopic scale, to be continous structures differing only in the size
of the regions consisting of commonly-aligned crystallites. These
variations in textural unit size are considered to arise from variations
-104-
in the nature of the parent vitrinite influencing both the molecular
species formed on pyrolysis and the environment within which they occur.
Thus, the increase in textural size from fine to granular-flow is regarded
as resulting primarily from changes in the environment, the accompanying
increases in fluiditY'of the carbonizing system favouring the alignment A.
of lamellar molecules. Fu~her increases in coal rank are associated with
decreasing fluidity, but the size of the anisotropic structures continues
to rise. This is explicable if the influence of the parent vitrinite on
the nature of the molecular species formed during carbonization becomes
increasingly more important. Such views are helpful in explaining changes
in textural composition resulting from variations in carbonization
conditions or from additions of tars or pitches to coals before
carbonization.
The studies [11,83,84], described earlier, of the development of
anisotropic structures during coal carbonization were carried out, by
heating in nitrogen at atmospheric. pressure, thin ( <10mm ) layers of
crushed coal contained in open boats. Relative to the size of textural
units in cokes produced in this way, the effect of heating in vacuo was
to downgrade the size of the textural units [1361. Conversely, carbonizing
coals in a small sealed tube upgraded the textural unit size, flow units
being unaffected [1361. These effects appear to be associated
predominantly with the environmemtal factor, changes in the .extent of
retention of the low molecular weight compounds influencing the fluidity
of the system and hence the alignment of planqr molecules. Co
carbonization of coals or vitrains with tars [137] or pitches [102,103]
resulted in the upgrading of mosaic units to larger size or to granular
flow units, but downgrading of flow-components was observed. Tar and
pitch additions will clearly enhance the fluidity of the carbonizing
system and this is considered to be the major factor involved. Some
effect on the chemistry of the system cannot be excluded however. The
effects were emphasized when a tar was mixed with a coal differing
widely in rank from that used to produce the tar. Thus, for example, a
tar from a 204 class coal enhanced mosaic sizes in coke from a 502 class
coal to a greater extent than the tar from a 301 class coal.
-105-
In open-boat carbonization, diffusion paths for released volatile matter
are short. In contrast, in the small pilot oven, as in a full scale oven,
diffusion paths are long so that escape of volatile matter from the
carbonizing system is retarded. Thus, 'these conditions would' appear to
approximate more closely to coal carbonization carried out in sealed
tubes or in presence of some tar. Comparison of textural data of pilot
oven cokes from Single-coal charges ( Table 6 ) with corresponding data
obtained from open-boat carbonization of coals of similar N.C.B. class
[lll, tends to comfirm this view. Pilot-oven cokes tend to contain more
granular-flow material and less flow and smaller mosaic components than
expected on the basis of open-boat carbonization.
When 1:1 mixtures of vitrains were carbonized in open boats, the
departures from additivity of textural data, as inferred from histograms
of measured and calculated textural data, did not fit an obviously
consistent pattern [136]. General effects appeared to be limited to those
blends containing vitrains from highly fluid coals in N.C.B. classes 301b
and 401. The effects were more marked when the 301b was used, especially
when blended with vitrains from coals of lower rank. Downgrading of the
granular-flow material from the 301b vitrain was accompanied by
upgrading of the mosaic components from the lower rank vitrain. Mosaics
in coke from the 401 vitrain suffered some downgrading when carbonized
with vitrains from coals of higher or lower rank. The influence on coke
textural composition of blending coals in the presence of added tar does
not appear to have been studied.
In the present studies, many of the departures of textural data from
additivity are small so that, were textural data not being used in
mathematical expressions, they would hardly be noteworthy. Nevertheless,
as the hatched bars in 'Figs 54-56 indicate, consistency in the direction'
of departures from additivity was common. Small consistent changes are
probably associated with interactions taking place over short distances
at the interface between particles of different rank. Generally, the
differences between measured and calculated textural data were not
significantly greater than those observed for open-boat carbonizations.
The effect is again considered to be associated with the enhancement of
-106-
the fluidity of the carbonizing system due to retained volatiles. If, for
comparison purposes, the broad- and striated-flow components are
regarded as belonging to a single textural class, then, for seven out of
eight types of the two-component blend shown in Table 36, blending
res'ults' in consistent d~wngrading of flow components. Since mosaics are
generally upgraded, the direction of all changes are in accord with those
observed on co-carbonization of coals with tars (137] or pitches
(101,138] in open boats. This pattern of changes results in cokes from
blended coal charges having textural compositions nearer to those of
cokes from 301b and 401 coals than would be expectd on the basis of
addi ti vity.
Increases in the size of mosaics are consistent with increases in the
fluidity of the system, although some effect of the mixed volatiles on the
molecular species present cannot be excluded. Flow-type structures are
considered to be formed when the nature of the coal results in the
formation of planar molecules able to align in low fluidity situations.
The formation of a type of fine mosaic component as an intermediary in
the formation of flow structures has been regarded as a manifestation of
some disruption of the original structural order of the vitrinite (91]. If
this process were enhanced, in the presence of retained volatiles, to such
an extent that this process became more dominant than the subsequent
alignment, then downgrading of flow components could well ensue. When
broad- and striated-flow components are considered as different entities,
then the behaviour of broad-flow material depends on whether it
originated from a 204 or a 301 coal, that in a 204 being consistently
downgraded on blending while that in a 301 being upgraded. No
explanation for this effect can be offered at the present time.
-107-
4.3 The influence of coke texture on coke strength
It is clear from Sections 3.3 and 3.4 that the tensile strength of cokes
from blended coal charges can be related to textural composition data
calculated from the blend composition and either SEM or PLM textural data
for the corresponding single-coal cokes. Such relationships are useful in
coke strength prediction and hence blend formulation. However, before
considering strength prediction in more detail, it is worthwhile
investigating the extent to which the strength of cokes can be regarded
as being causally dependent on their textural composition.
The SEM study of fractured coke surfaces found variations in the mode of
failure of different textural components. Because of crack propagation
from pore to pore, circumferentially-aligned lamellar components appeared
to fail by a translamellar mechanism to produce a very rough fracture
surface. The appearance of granular components in fracture surfaces was
suggestive of intergranular failure, hence the name chosen to describe
them. Flat and inert components produced very smooth fracture surfaces
bearing brittle fracture river patterns. By examining etched surfaces cut
perpendicular to fracture cracks, corresponding variations in the
tortuosity of the cracks through textural components were observed. These
differences in surface roughness and tortuosity of failure path indicated
differences in surface energy which, according to the Griffiths view and
unless compensated by changes in Young's modulus or flaw size, would lead
to differences in the contribution of the various textural components to
. coke strength. Subjecting tensile strength and measured SEM textural data
to multi-linear regression analysis [132] showed that it was possible to
relate strength and textural data but such a purely statistical
relationship was of little scientific value. It gave neither insight into
the breakage of coke nor ready indication of the type of coke texture
necessary for high strength. Hence the TRANS(31) and INTER(33) equations
were derived from· very simple views of transgranular and intergranular
failure in an attempt to represent the two modes of failure evident from
SEM studies.
-108-
The derivation of these equations is given in Appendix Ill. The approach
adopted was based on the probability of occurrence of textural components
in any plane in coke. Intergranular failure is then simply regarded as
the pulling apart of two such planes alo'ng interfaces between textural
components while in transgranular breakage the fracture path passes
through the components constituting the layer. The simplicity of this
approach is acknowledged. Since no account is taken of any flaws present,
the approach can be reconciled with a Griffiths view of brittle fracture
only if it is assumed that the flaws are a constant factor.
For intergranular fracture, the coke strength is considered to be
dependent upon the probability of contact betw,een components across the
interface and the strength of the bond between them. Then, for a two
component coke the tensile strength S is given by:
<43)
where T,and T2 are the fractional concentrations of components 1 and 2
and S, ." S'.2 and S2,2 are the strength of the bond between two units of
the type indicated by the suffixes. S,., and S2,2 can be regarded as
being the strength of components 1 and 2 while S'.2 can be regarded as
an intertextural strength. The derivation of this equation is based on
that of equation <23) [124].
In transgranular failure, the coke tensile strength is considered to be
dependent upon the probability of occurrence of textural components in
any plane and their strength. The tensile strength of a two component
coke is then given by the equation :
<44)
where T,and T" are again the fractional concentrations of components 1
and 2 and S, and 82 are their' strengths:
The INERT<35) equation represents an attempt to incorporate the
reactives/inerts concept, so often used by coal petrologists in assessing
the quality of coking coals, into a transgranular fracture mechanism.
Inerts are assumed to be associated with individual non-inerts
-109-
( reactives ) into units whose strength is dependent upon the proportion
of the inerts present. The assumed nature of this dependency is explained
in Appendix Ill. For a two-component coke the equation reduces to:
where I is the inert content and the other symbols have the same meaning
as before.
These equations in the forms of the INTER(31), TRANS(33) and INERT(35)
equations were regarded as purely empirical when used in conjunction with
calculated SEM or PLM textural data for cokes from blended-coal charges.
Because of the artificiality of using calculated data, they are still so
regarded. Nevertheless, it is convenient to describe the coefficients
obtained by fitting the equations to calculated data in strength terms.
For example, the coefficient in Table 19 associated with the INTER(31)
equation cross terms T,T j can be regarded as intertextural strength
values, those associated with the T,2 being textural strengths. On the
same basis, the TRANS(33) equation coefficients in Tables 21 and 33 can
also be regarded as textural strengths.
It is now believed that the structures of all the optically-anisotropic
textural components are continuous so that intergranular breakage is not
expected. Nevertheless, the INTER(31) and TRANS(33) equations were both
used in an attempt to seek relationships between the tensile strengths
and the measured textural composition data for the forty-four cokes from
blended-coal, pilot-oven charges. Data for the single-coal cokes were not
included thus ensuring that all the standard errors of estimating tensile
strengths quoted in this thesis were calculated for the same number of
cokes. The INERT(35) was not considered to reflect a realistic variation
of coke strength with textural composition and was not used for this
purpose.
As explained in Section 3.3, for nine textural components there are forty
five textural and intertextural strengths. In order to reduce to
manageable proportions the number of sucn terms used when fitting the
INTER(31) equation to the data, the following assumptions were made:
-110-
1. that inter textural strengths between large inert components and
any other reactive component were equal.
2. that inter textural strengths between small inert components and
any other reactive component were equal.
3. that, in terms of their textural and intertextural strengths,
isotropic and broad-flow components behaved as a single
component.
4. that intertextural strengths between isotropic or broad-flow
components and any other reactive component were equal.
5. that the strengths between inert components, whether large or
small were zero.
This resulted in the number of strength terms requiring calculation being
reduced to nineteen. These are identified in the listing below at the
intersections of the rows and columns. Terms numbered 1 to 4 respectively
are those influenced by the above assu~ptions 1 to 4.
1nl Ins Fb Fs Fg Mc Mm Mf I
1nl 0 0 1 1 1 1 1 1 1
Ins 0 2 2 2 2 2 2 2
Fb 3 4 4 4 4 4 3
Fs 5 6 7 8 9 4
Fg 10 11 12 13 4
Mc 14 15 16 4
Mm 17 18 4
Mf 19 4
I 3
Similar assumptions were made when fitting the INTER(31) equation to
calculated SEM textural data in Section 3.3. Flat and very fine granular
components were then considered to behave as if belonging to a single
textural class.
The values of the nineteen strength terms, obtained by application of the
algorithm in Appendix I to the INTER(33) equation and the measured PLM
textural data, which gave the lowest observed standard error of
estimation, 0.55MPa, are listed in Table 41. Values of strength terms in
the TRANS(33) equation which gave a standard error of estimation of
-111-
O.47MPa are listed in Table 42, while Table 43 compares the measured coke
strengths with those calculated from the two equations. It is evident
that, using the TRANS<33) equation, the tensile strength of cokes from
blended-coal charges can more closely be related to calculated textural
data, a standard error of estimation of O.38MPa having been obtained in
Section 3.4, than to measured data.
As with all brittle materials ( Section 2.3.2 ), porosity has an important
influence on coke strength. The equation derived by Knudsen for ceramics:
S= SoG-o .c. exp ( -bp ) <8)
where G is the grain size, b is a pore shape factor and p the volume
porosity, has been adapted to explain the variation of coke tensile
strength with porosity [10], the form of the equation used being:
S= 450x Fmax-O .5 exp-[ 2( Fmax/Fmin )O.5 x p ] <22) .
In this, the pore shape factor b was replaced by an expression based on
the stress concentration at the tip of a crack whose length greatly
exceeded its breadth (equation <2) ). Fmax and Fmin are the maximum
and minimum Feret's diameters of the larger pores.
In developing this equation, coke breakage was considered in terms of
classical brittle failure. The pores were regarded as the strength
regulating, inherent flaws, the larger pores being the Griffiths critical
flaws. The study described in Section 3.2 demonstrates that coke breakage
can no longer be considered to be an immediate consequence of the f
prqagation of flaws present in unstressed coke. Instead, critical flaws
result from the joining together of subcritical microcracks, initiated at
the larger pores at lower stress levels. Growth of these subcritical
microcracks, many of which are not involved in the formation of the
critical flaw, will result in a high specific surface energy. Thus coke
breakage does not conform to ideal brittle fracture behaviour. High
surface energies can also result when local plasticity blunts the crack
tip. However, for coke, and also graphites, where departures from ideal
brittle behaviour have been characterized in terms of variations in
surface energy [139], it is not suggested that failure, at least at room
-112-
temperature, involves other than brittle fracture processes. These
considerations do not conflict with the view that, as work with equation
<22) showed, pores in coke play an important strength controlling role.
No detailed pore structural data are available for the cokes made in this
study, although, apparent densities, p_, but not specific gravities, p.-,were
measured for all the cokes. However, since specific gravities of crushed
coke do not vary widely [140], a single value of 1900kg/m8 was used to
calculate the fractional volume porosities of the coke from blended-coal
charges according to :
<46).
The values obtained, listed in Table 44, were used in an attempt to
modify the TRANS<33) equation to reflect variations in the porosity of
the cokes. The equation used: s
S=l:. Sit; exp( -bp) , -,
is here termed the POROSITY<47> equation and is based on the
Ryshkewitch-Duckworth equation:
S= So exp ( - bp )
<47)
<6>
where b is a constant unless pore shape varies. In equation <6>, So is
considered to be the strength of a non-porous body, the implication being
that the inherent flaws, at which fracture is initiated, are unaffected by
porosity changes. On this basis, the summation term in the POROSITY<47>
equation could be regarded as indicating the varying contribution of the
textural components to the strength of pore-free cokes. However, this
would not be realistic since, for cokes, it is the pores themselves which
provide ,the stress concentration for crack initiation.
Studies of the porous structure of cokes', using automated, image analysis
[10], showed that, for most blast-furnace cokes, the aspect ratio
Fmax/Fmin lay within the range 1.72 to 1.90. Using the expression for b
in equation <22>, these values correspond to values for b of 2.62 to 2.76.
An explicit derivation of the Knudsen relationship has since been
published' [1411. This included, from consideration of far field
displacement effects, the derivation of an alternative expression for the
-113-
variation of b with pore shape. On this basis, for aspect ratios of 1.0 to
2.0, b remains almost constant at 2.78. Otherwise, for aspect ratios up to
10, the two expressions for calculating b give values in reasonable
agreement. Therefore,in attempting to relate coke strength and textural
data by evaluating the strength terms in the POROS1TY(47) equation, only
b values of 2.6 and 2.8 were used. The values of the strength terms,
obtained using the algorithm in Appendix I, are given in Table 44. For b
values of 2.6 and 2.8, the coke tensile strengths can be calculated with
standard errors of estimation of 0.46 and 0.47MPa respectively. As could
be anticipated from the form of the POROS1TY(47) equation, strength terms
are higher for the larger value of b but the difference in the precision
of coke strength calculation is negligible. Measured and calculated
strengths of cokes from blended-coal charges are compared in Table 45.
These standard errors of estimation show no improvement over that,
0.47MPa ( Table 42 ), obtained without taking account of the porosity of
the cokes. They are, however, lower than the value, 0.54MPa, obtained by
applying equation (22) to tensile strength and pore structural data for
forty-two blast-furnace and test-oven cokes. These standard errors of
estimation imply that the equations account for approximately 70% of the
variation of tensile strength. All the standard errors of estimation
obtained using measured textural data are higher than the value, 0.38MPa,
obtained when the TRANS(33) equation was applied to textural data
calculated from the blend composition and the textural data of the
Single-coal cokes. Whether the tensile strength of coke is causally
dependent upon coke textural composition to the extent implied by the low
standard errors of estimation is therefore open to question. It is
recognised that for coal blends carbonized under a single set of
carbonizing conditions, changes in blend composition lead simultaneously
to changes in both pore· structural characteristics and textural
composition. Thus, it is dificult to obtain a relationship between coke
tensile strength and either pore structural parameters or textural data
individually which excludes an effect due to the other factor. Hence
individual approaches can both account for a high proportion of the
variation in tensile strength .. Thus, although on· theoretical grounds, coke
strength is expected to be dependent upon both the porous structure and
-114- .
the nature of the coke matrix, only after further careful work will the
relative contributions of the porous structure and textural composition be
separately identified.
Since textural data is not a strictly additive property of the coals in a
coal blend, the fact that coke tensile strengths can be related more
closely to calculated textural data than to measured data implies that
the textural data of single-coal cokes may be reflecting some other coke
property, related to strength, which is additive" or indeed that the
tensile strength itself is additively dependent upon blend composition
and the strengths of the cokes obtained from individual blend components.
To investigate this view, attempts were made to relate the tensile
strengths of cokes from blended-coal charges with the blend composition
and the tensile strengths of the single-coal cokes using the equation:
" S= L SiFiC, <48) ~~l
where Si, Fi and Ci are the tensile strength of the coke from the ith
coal, the fractional concentration of the ith coal in the blend and the
yield correction factor ( see Table 7 ) for the ith coal respectively.
This equation is referred to as the ADDTS<48) equation.
By analogy with the derivation of the TRANS<33) equation, this equation
implies that coke tensile strength is dependent upon the probability of
finding coke from an individual blend component in a particular layer and
the strength of the coke obtained from that coal. This equation has been
applied in two slightly different ways. In the first, the strengths of the
single-coal cokes were assumed to be accurate and were used directly to
calculate the strengths of cokes from blended-coal charges. This
permitted the coke strengths to be calculated with standard errors of
estimation of 0.44, 0.45 and 0.45MPa for the three methods ( C, Y and V )
of calculating the contribution of the individual blend components to the
coke from the blended-coal charge. Measured and calculated strengths of
cokes from blended-coal charges are compared in Table 46.
-115-
In the second approach, notional values for the strengths of the cokes
from the single-coal cokes were obtained from the tensile strengths of
cokes from blended-coal charges and the blend composition. Values of
these notional strengths, obtained using the three methods ( C, Y and V )
of calculation are listed in Table 47. Using these values in the
ADDTS(48) equation permitted the strengths of cokes from blended-coal
charges to be calculated with standard errors of estimation of O.39MPa
for all three sets of notional strengths. The values are almost identical
with those obtained using the TRANS<33) equation and calculated textural
data for cokes from blended-coal charges. Measured strengths of cokes
from blended-coal charges are compared with calculated values in Table
48.
Thus, the tensile strengths of cokes from blended-coal charges can be
calculated from the blend composition and the strengths of the cokes
from the individual blend components with sufficient accuracy for this to
provide an alternative method of coke strength prediction. It follows that
any coke properties which are related to the tensile strengths of a
single-coal cokes can be used for predictive purposes. Further work is
necessary to obtain a detailed description of coke tensile strength in
terms of pore structural parameters and textural composition data, but it
is evident that if such a relationship is obtained the data used can also
be used for strength prediction and hence blend formulation purposes.
-116-
4.4 The prediction of the tensile strength of coke
Present coke stength prediction methods, outlined in Section 2.5, involve
the accumulation of sufficient data·lo· der'lve a strength/property
relationship and the subsequent application of this relationship to
calculate the strength of coke to be expected from other blends of coals.
Following the same basic approach, in various sections of this thesis,
four methods of predicting the strength of cokes produced from blended
coal charges have been developed:-
Method I
Data required:
1. The tensile strengths of cokes from a number of blended-coal
charges.
2. Textural composition data for single-coal cokes.
3. Blend compositions.
From this data, textural data for cokes from blended-coal charges may be
calculated. This may then be used in conjunction with the tensile strength
values to obtain a strength/texture relationship. This, in turn, can be
used to predict the strength of cokes from any other blend of the coals
considered. Textural data for single-coal cokes may be obtained by pOint
counting during the examination of etched coke surfaces in a scanning
electron microscope or of polished surfaces under polarized light.
Textural data of cokes from blended-coal charges may be calculated using
blend composition data or, if the yields of single-coal cokes or
analytical data for the coals are known, such data may be corrected to
reflect more accurately the contribution of each coal to the coke. Purely
empirical strength/texture relationships may be obtained using the
MLR<29> or TRANS<33> equations, the latter having the advantage that it
permits the ready identification, from the textural composition of their
cokes, those coals likely to produce high strength cokes.
Method 11
Data required:
1. The tensile strengths of cokes from a number of blended-coal
-117-
charges.
2. Textural composition data for a number of cokes from blended
coal charges.
3. Blend compositions.
This method is essentially similar to method I except that notional
textural data for single-coal cokes are obtained from the textural data
measured for a number of cokes from blended-coal charges. All the coals
of interest should be included in a number of blends. The method has only
been demonstrated using PLM textural data, the notional textural data for
single-coal cokes being calculated from blend compositions without yield correction.
Method III
Data required:
1. Tensile strengths of single-coal cokes.
2. Blend compositions.
From this small amount of data, tensile strengths of cokes from blended
coal charges can be computed readily, using the ADDTS(47) equation, from
the tensile strengths of the single-coal cokes and the blend composition
with or without yield correction.
Method IV
Data required:
1. Tensile strengths of cokes from a number of blended-coal
charges containing all the relevant single coals.
2. Blend compositions.
This is similar to method III except' that notional strengths of single
coal cokes are obtained from those of cokes from blended-coal charges
and the blend composition. The latter may be used directly or after yield
correction. These notional strengths may then be used to'calculate the
strengths of cokes from blended-coal charges using the ADDTS(47)
equation.
Before considering the suitability of the four methods of coke strength
prediction in different situations, some further points should be 'made.
Since it is the simpler, technique, it is recommended that polarized light
-118-
microscopy of polished coke surfaces be used to measure textural data.
The TRANS(33) equation can then be used. Also, since in general there is
no improvement in precision of strength prediction if the blend
composition is corrected to reflect the contribution of each coal blend
component to the coke from a blended-coal charge, blend composition data
should be used directly. It is expected that normally the tensile strength
and textural composition of coke from any coal or blend will vary
depending on the carbonization conditions used. However, it is considered
that operating conditions for a pilot oven, of, for example, 250kg size,
could be so chosen that both the tensile strength and textural
composition of coke from any coal or blend would be identical to those
obtained from the same charge carbonized in a commercial oven. ·For small
scales of carbonization, it is deemed possible only to simulate conditions
of larger ovens to the extent that the textural compositions of cokes
produced on the different scales would be identical. All equations used in
strength prediction in this thesis are regarded as empirical. Thus,
although the form of the equations is expected to be applicable in other
situations, it will be necessary to re-evaluate the coefficients in them
for other carbonization conditions. Further work is required to enable the
prediction methods developed to be adapted to oven charges which include
any type of pitch, non-fusing coals, petroleum coke or coke breeze.
The strength/texture relationships developed in this work were obtained
using coke textural compositions covering the range likely to be
encountered in blast-furnace cokes produced in the U.K. However, the
extent to which the relationships can be regarded as applicable to cokes
from other coals carbonized under similar conditions has not been
investigated. The usefulness of strength prediction methods I and 11
depends on these relationships being applicable to cokes of similar
textural composition made from other coals, otherwise there is no merit
in using them in preference to methods III or IV. Clearly, any extension
of this work should involve the examination of this pOint. For present
purposes, however, it is assumed that the relationships do have general
applicability.
-119-
In applying the strength prediction methods in any of the situations
described below, it will first be necessary to accumulate, for each
method, the data listed above and to build up a data base containing
relevant information. For methods III and IV this should contain
respectively the tensile strengths or notional tensile strengths of
single-coal cokes. For methods I and 11 the data base should contain the
coefficients in the relevant strength/texture relationships and either the
textural data or notional textural data for all coals of interest.
Depending on the assumptions made, especially regarding the comparability
of cokes produced under different scales of carbonization, all four
methods of coke strength prediction developed are considered to have
some merit. To explain this further, three sitatuions are considered. In
each case, it is assumed that the necessary data bases have been
established.
A. Only a commercial oven is available and no single-coal charges are
carbonized.
In this situation, since no direct measurement of data from single-coal
cokes can be made, options available are limited to methods 11 and IV. On
introducing a new coal, to be able to calculate the composition of blends
of this coal with other coals of known notional strength or textural
composition, it. would be necessary to carbonize a number of blends
containing the new coal and then to calculate its notional data. The
tensile strengths of cokes from blends of the new coal with any
combination of other coals could then be obtained using the TRANS(33) or
ADDTS(47) equations. Since ·the use of method IV and the ADDTS(47)
equation involves accumulating fewer data, this method is preferred in
this situation.
B. Commercial and 250kg ovens are available. Operating conditions of the
250kg oven have been carefully chosen so that a coal or blend
carbonized in both ovens produces coke having the same tensile
strength and textural composition. Single-coal carbonizations are
possible in the 250kg oven.
-120-
It is possible to use any of the four strength prediction methods in this
situation. For methods I and Ill, to introduce a new coal necessitates the
carbonization of one or two Single-coal charges in the 250kg oven and to
measure the coke textural composition or tens fIe strength. For methods Il
and IV however, several carbonizations would be necessary to permit
notional data to be calculated. Hence methods I and III are preferred to
methods 11 and IV. Because less data and computation is involved, method
III is recommended over method 1. The strength of coke from any blend of
coals on the data base can be obtained by computation from the measured
tensile strength of cokes from single coals carbonized in the 250kg oven
using the ADDTS(47) equation.
C. Commercial, 250kg and small-scale ovens are available. Coals
carbonized in the two larger ovens produce cokes having the same
tensile strength. Cokes with similar textural compositions are
produced in all three ovens.
In this situation, it is again possible to apply any of the four strength
prediction methods. However, once the data bases have been built up using
data obtained using the 250kg oven, because textural data for cokes from
any further coals can be obtained using the small-scale oven, it becomes
advantageous to use either methods I or method Il. Of these, method I is
preferred. This situation w'ould be particularly appropriate for the
evaluation of small samples of coal, for example from bore holes.
Thus, it is possible to envisage situations where all the methods of coke
tensile strength prediction developed during this work can be usefully
employed.
-121-
4.5 Recommendations for future work
S~me of the work described in this thesis may be considered as
fundamental~s-,,-ientific,_the remainder technica.l in appr:o.",ch. The work
may·be usefully extended following both approaches. --.-
On a scientific level, further work aimed ~t obtai ing, a more cqmplete c - F~~"t RC understanding of the factors contrcJl~i~.Lthe ~~te stre~~ .~~~e- is __
r;commended. The present work has shown t~high proportion of the
variation of tensile strength of cokes produced under a single set of
carbonizing conditions can be accounted for by consideration of the
textural composition of the cokes; A similar· proportion of the variation
in tensile strength is explicable in terms of parameters characteristic
of the porous structure of cokes. Clearly, it would be scientifically
interesting to seek a single relationship to explain coke tensile strength
in terms of both texture and porous structure. However, since changes in
blend composition can result in simultaneous changes in both texture and
pore structure, much careful work will .be necessary before this single
strength/texture-pore structure relationship can be established.
On a technical level, the obvious way in which this work could be (.
extended would be to first confirm the findings using larger ovens. If aJ
variety of ovens were available, from the several-gramme laboratory size
up to 20-40t commercial ovens, it would be possible to investigate the
four methods of strength prediction. Initial studies could be carried out
quite cheaply using the 7kg oven, an oven often used in the past for
investigative work. Establishment of the general applicability, or
otherwise, of strength/texture relationships should have priority.
Depending on the results obtained, the study could then be extended
gradually to ovens of increasing size.
• j " j\
,_.j
-122-
5. CONCLUSIONS
1. The carbon in metallurgical cokes is composed of textural
components whose size and shape vary with the rank of the coal
carbonized. These components can be studied by scanning electron
microscopy (SEM) of fractured or etched coke surfaces, or by
polarized-light microscopy (PLM) of polished surfaces.
2. Textural components can be classified into categories, a different
nomenclature being used depending on the type of microscope used in
their examination. Components revealed by SEM study of surfaces
etched in atomic oxygen can be identified with those evident under
polarized light, although not with one-to-one correspondence.
3. The principal textural components, derived from reactive coal
macerals, classified by SEM study are termed lamellar, intermediate
and granular. Corresponding terms applied when PLM is used are flow,
granular-flow and mosaics. Some subdivision of these categories is
necessary to differentiate adequately between cokes. A subgroup
termed very-fine granular on SEM examination corresponds to
isotropic material revealed by PLM. Carbonaceous inerts are
classified in both systems.
4. SEM examination of fractured coke surfaces reveals variations in
the mode of fracture of the various textural components. Lamellar
components fail by a translamellar mechanism while the appearance of
granular components gives the impression of intergranular fracture.
Differences in the roughness of the fracture surfaces of the various
components imply variations in their surface energy and, hence,. in
their contribution to the tensile strength of coke.
5. Lamellar components in etched coke surfaces are evident as
parallel arrangements of ridges and channels, while granular
components give uniformly pitted surfaces.
-123-
6. The tortuosity of microcracks and fracture cracks visible in
etched coke surfaces cut pe-rpendicular to the applied tensile stress
corresponds to the shape of textural components deduced from study
of fracture surfaces and confirms the differences in their mode of
failure. There· is no tendency, however, for the fracture crack to be
diverted through any particular component or along the interface
between two components.
7. Although metallurgical coke is aCknowle~dged to be a brittle
material, its fracture does not entirely conform to the classical
Griffiths view in which failure is an inevitable consequence of the
propagation of a critically-sized flaw present in unstressed material.
Instead, failure occurs when a critical flaw is formed by the joining
together of stable microcracks initiated at the larger pores at lower
stress levels.
8. Optically-anisotropic textural components in polished coke
surfaces are characterized under polarized light, with crossed polars
and a full-wave retarder plate, by isochromatic areas varying in size
and shape. The colours present reflect variations in the orientation
of the component graphitic crystallites relative to the vibration
direction of the incident polarized light. Black lines visible on the
surface at high magnification mark the loci of points where the
graphitic layers lie parallel or perpendicular to the vibration
direction. They are termed extinction contours.
9. When the surface is rotated, the change in shape of the
isochromatic areas and the movement of extinction contours implies
that, on a.scale relevant to microscopic observation, the structures
can be regarded as continuous. On this basis, the size of the
isochromatic areas reflects variations in the extent of common
alignment of the constituent crystallites.
10. The PLM textural compositions, ie. the proportions of the various
components present, of cokes from blended-coal charges are not
.. additively dependent upon the corresponding data for the single-coal
-124-
cokes and the blend composition, even when the latter is corrected to
reflect accurately the contribution of the single coals to the coke
from the blended-coal charge. Departures from additivity are often
small but consistent. They reflect a downgrading of larger flow
components and an upgrading of small mosaic material probably as a
consequence of the interaction of trapped low molecular weight'
materials with the plastic mass modifying the growth of anisotropic
structures.
11. The t~nsile strength of cokes, produced in a small pilot oven,
from blended coal charges can be related to their PLM textural
compositions using equations, 'the form of which were derived from
consideration of a very simple model of coke failing by an
intergranular or transgranular mechanism. Only a small improvement
in the precision of calculating the coke tensile strength results
from including a porosity term in the transgranular failure equation.
12. There is some doubt as to the extent to which the low standard
errors of estimating the coke tensile strengths from the equations
represent a causal dependence of tensile strength on coke texture.
Coke tensile strengths are also related to parameters characteristic
of the porous structure of cokes and both pore structure and texture
change as a result of alterations in the composition of the blend
carbonized.
13. Tensile strengths of cokes from blended-coal charges can be
related to textural composition data calculated from the textural
composition of the constituent single-coal cokes and the blend
composition. The latter may be used directly or when corrected to
take into account either the measured yield of single-coal cokes or a
yield calculated from analytical data. This assumes that textural
data is additive. Both SEM and PLM textural data can be used.
Relationships can be obtained using multi-linear regression analysis
or by fitting to the data the equations de~ived from consideration of
intergranular and transgranular modes of coke breakage. The equation
based on transgranular breakage, although not necessarily the most
-125-
accurate, has the merit of simplicity and the ability to identify
readily, from the textural composition of their cokes those coals
able to produce high strength cokes.
14. These equations are useful as bases for predicting the tensile
strength of cokes from· blended-coal charges. For this purpose, it is
feasible to calculate the textural composition of cokes from blended
coal charges from notional textural data for single-coal cokes.
Notional textural data can be computed from measured data for a
number of blended-coal charges and the blend composition.
15. The standard errors of estimating the tensile strengths of cokes
from blended-coal charges from equations derived using calculated
textural data are lower than corresponding values obtained when
equations derived using measured data are used. The implication that
the tensile strength itself can be considered an additive property of
cokes is correct. The tensile strengths of cokes from blended-coal
charges can be calculated from the blend composition and either the
tensile strengths of single-coal cokes or notional values, themselves
calculated from the tensile strengths of cokes from other blended
coal charges.
16. The usefulness of the various methods· of predicting the tensile
strengths of cokes from blended-coal charges depends on the number
and size of the coke ovens available and the extent to which
conditions in the smaller ovens have been adjusted to achieve cokes
identical in tensile strength and textural composition to those
obtained on the commercial scale. Where identicality has been
achieved, it is simpler to calculate the tensile strengths of cokes
from blended-coal charges directly from either measured or notional
strengths of single-coal cokes. It is likely, however, that·
identicality in textural composition can be achieved down to a scale
of carbonization so small that the coke produced would be
insufficient for tensile strength measurement. In this situation,
there is considerable advantage, in terms of cost, in using a
prediction ~eth~d ba~ed ·on a: strengthitextural composition equation.
-126-
Thus, the various tensile strength prediction methods have merit in
different situations.
-127-
REFERENCES
1. '1983 Energy Statistics Yearbook', United Nations, New York, 1985
2. 'Digest of U.K. Energy Statistics 1985', H.M. Stationery Office, London, 1985
3. Gibson, J. and Gregory, D. H. 'The Carbonisation of Coal', Mills and Boon, London, 1971
4. Graham, J. P. Yearbook of the Coke Oven Managers Assoc. 1977, 328
5. Mackowsky, H.-Th. J. Microscopy 1977, 109, 119
6. Honson, J. R. Yearbook of the Coke Oven Managers Assoc. 1979, 267
7. Thibault, G. in 'Chemistry of Coal Utilisation', (H. H. Lowry ed.), Supplementary Volume, John WHey, New York, 1963, p494
8. Patrick, J. W. and Stacey, A. E. in 'Proc. Fourth London Int. Conf. on \ Carbon and Graphite', 1974, S.C.!., London, 1976, p634
9. inter alia, Patrick, J. W. and Stacey, A. E. Fuel 1978, 57, 258 Jeulin, D. and Steiler, J. M. Mem. Sc. Rev. Met. 1980, 2, 107 Klose, W. and Suginobe, H. in 'Proc. 1985 Conf. on Coal Science', Sydney, New South Wales, 1985, p977
10. Patrick, J. W. and Stacey, A. E. I.S.S. Transactions 1983, 3, 1
11. Patrick, J. '0., Reynolds, M. J. and Shaw, F. H. Fuel 1973, 52, 198
12. Mott, R. A. 'History of Cokemaking', Heffer, Cambridge, 1936
13. Steer, F. and Antrobus, K. Yearbook of the Coke Oven Managers Assoc. 1976, 219
14. Soule, J. L. Fuel 1955, 34, 68
15. Gregory, D. H. Yearbook of the Coke Oven Managers Assoc. 1971, 174
16. Wall, M. K. Yearbook of the Coke Oven Managers Assoc. 1974, 265
17. Graham, J. P. and Wilkinson, H. C. Yearbook of the Coke Oven Managers Assoc. 1980, 161
18. Beimann, '0., Auvil, H. S. and Coleman, D. C. in 'Chemistry of Coal
,
-128-
Utilisation', (H. H. Lowry ed. ), Supplementary Volume, John Wiley, New York, 1963, p461
19. Mott, R. A. and Wheeler, R.V. 'Coke for Blast Furnaces', The Colliery Guardian, London, 1930
20. Miura, Y. Yearbook of the Coke Oven Managers Assoc. 1978, 292
21. Graham, J. P., Pater, W. J. and Lee, G. W. in 'Coke in Ironmaking', I.S.I. Publication 127, London, p54
22. Staniforth, W. Yearbook of the Coke Oven Managers Assoc. 1974, 208
23. Sorensen, S. M. jr., Arsenault, A. A. and Rupp, P. A. in 'Proc. 41st Ironmaking Conf.', Pittsburgh, 1982, A. I.M.E. Iron and Steel Soc., 1982, p351
24. Peacey, J. G. and Davenport, W. G. Pergamon, Oxford, 1979
'The Iron Blast Furnace',
25. Nakamura, N., Togino, Y. and Tateoka, M. Ironmaking and Steelmaking 1978, 5(1), 1
26. Dartnel1, J. D. Ironmaking and Steelmaking 1978, 1, 18
27. McKee, D. W. in 'Physics and Chemistry of Carbon', (P. L. Walker Jr. and P. A. Thrower eds), Dekker, New York, 1981, Vol 16, p1
28. Araham, K. P. and Staffansson, L. I. Scandinavian J. Metallurgy, 1975, 4, 193
29. Berger, D., Carton, B., Metrot, A. and Herold, A. in 'Physics and Chemistry of Carbon', (P. L. Walker Jr. and P. A. Thrower eds.), Dekker, llew York, 1975, Vol 12. p1
30. Miyazu, T., Okuyama, Y., Fukushima, T., Yamaguchi, M. and Izawa, T. in 'Int. Conference on Coal, Coke and the Blast Furnace', Middlesborough, 1977, The Metals Society, London, 1978
31. Walker, P. L. jr., Rusinko, F. and Austin, L. G. Advances in Catalysis 1959, 11, 133
32. Gibson, J. and Gregory, D. H. Yearbook of the Coke Oven Managers Assoc. 1975, 137
33. Chow, C. K. and Lu, W-K. in 'Proceedings of the 38th Ironmaking Conf', Detroit, 1979, A. I.M.E. Iron and Steel Soc., 1979, p38
-129-
34. Goleczka, J. and Tucker, J. in 'Information Symoposium : Coke Oven Techniques', Luxembourg, 1981, Graham and Trotman, London, 1982, p109
35. Marshal I , J. R. Yearbook of the Coke Oven Managers Assoc. 1979, 135
36. Hill, J. D. Yearbook of the Coke Oven Managers Assoc. 1980, 211
37. Hyslop, W. Yearbook of the Coke Oven Managers Assoc. 1981, 131
38. Van Krevelen, D. W. 'Coal', Elsevier, Amsterdam, 1961
39. Lowry, H. H. (ed.) 'Chemistry of Coal Utilization', Supplementary Volume, Wiley, New York, 1963
40. British Standard 1016, Part 12, 1980
41. Patrick, J.W. Scientific Progress, Oxford 61, 375
42. Speight, J. G. 'The Chemistry and Technology of Coal', Dekker, New York, 1983
43. Mackowsky, M.-Th. and Wolff, E. M. in 'Coal Science', American Chemical Society, Advances in Chemistry Series 55, Washington D.C., 1966, p327
44. Hays, D., Patrick, J. W. and Walker, A. Fuel 1976, 55, 297
45. Brown, F. Yearbook of the Coke Oven Managers Assoc. 1980, 87
46. Wilmers, R. R. in 'Round Table Coke', Commission of the European Communities, 1982
47. Patrick, J. W. and Stacey, A.E. Fuel 1972, 51, 81
48. Blayden, H. E., Noble, W. and Riley, H. L. J. Iron and Steel Inst. 1937, No. II, 47
49. Blayden, H. E. J. Chimie Physique 1969, 15
50. Davidge, R. W. 'Mechanical Behaviour of Ceramics', Cambridge University Press, Cambridge, 1979
51. Inglis, C. E. Trans. Naval 'Archit. 1913, 55, 219
52. Griffiths, A. A. Phil. Trans. R. Soc. London 1920, A221, 163
53. Griffiths, A. A. in 'Proceedings of the First International Congress on Applied Mechanics', (C. B. Biezeno and J. M. Burgers
-130-
eds. >, Waltham, Delft, 1924, p55
54. Duckworth, W. J. Am. Ceramic Soc. 1953, 36, 68
55. Knudsen, F. P. J. Am. Ceramic Soc. 1959, 42, 376
56. Orowan, E. Progress in Physics 1949, 12, 185
57. Carbonization Research Report 5, British Carbonization Research Association, Chesterfield, 1974
58. Pierce, T. J. and Horton, A. E. in 'E.C.S.C. Round Table Meeting on Coke Ovens and Coke Research', Commission of the European Communities, 1980, p277
59. Kesler, C. E. in 'Tests and Properties of- Concrete', American Society for Testing and Materials; 1966
60. Hendren, A. J. in 'Rock Mechanics in Engineering Practise', (K. G. Stagg and K. G. Zieckiewicz eds.>, Wiley, London, 1968, p21
61. Rudnick, A., Hunter, A. R. and Holden, F. C. Materials Research and Standards 1963, 3, 283
62. Carbonization Research Report 43, British Carbonization Research Association, Chesterfield, 1977
63. Hartshorne, N. H. and Stuart, A. 'Crystals and the Polarizing Microscope', Arnold, London, 1950
64. Bloss, F. D. 'An Introduction to the Methods of Optical Crystallography', Holt, Reinhart and Winston, New York, 1961
65. White, J. L., Guthrie, G. L. and Gardner, J. O. Carbon 5, 517, 1967
@ Taylor, G. H. Fuel 1961, 40, 465 cJC;,-,~'!I:;~'; U'.J "'-'""-
67. Marshal I , C. E. Fuel 194, 24, 120
68.
([§.) Dahme, A. and Mackowsky, M-Th. Brennstoff-Chemie 1949, 30, 141
Brooks, J. D. and Taylor, G. H. in 'Chemistry and Physics of Carbon', (P. L. Walker Jr. ed.), Arnold, London, 1968, Vol 4, P 243
White, J. L. Progress in Solid State Chemistry 1974, 9, 59
White, J. L. American Chemical Society Symposium, Series 21, 'Petroleum Derived Carbons', 1976, 282
,/
-131-
Marsh, H. in 'Proceedings of the Fourth London International Carbon and Graphite Conference', S. C. I., London, 1976, p 2
Marsh, H. and Walker, P. L. Jr. in 'Chemistry and Physics of Carbon', (P. L. Walker Jr. and P. A. Thrower eds.), Dekker, New York, 1979, Vol 15, P 237
rf!J5.~ Dubois, J., Agache, C. and White, J. 1. Metallography 1970, 3, 337
~ Hoover, D. S., Davies, A., Perrotta, A. J. and Spackman, A. in 'Extended Abstracts, 14th Biennial Conference on Carbon', American Carbon Society, 1979, p 393
76. Fitzer, E. ,'Mueller, K. and Schaefer, W. in 'Chemistry and Physics of Carbon', (P. L. Walker Jr. ed.), Dekker, New York, 1971, Vol 7, p237
~ Brooks, J. D. and Taylor, G. H. Nature 1965, 206, 697
78. Patrick, J. W., Reynolds, M. J. and Walker, A. Fuel 62, 129
79. Rester, D. O. and Row, C. R. Carbon 1974, 12, 218
80. Scott, C. B. and Folkins, H. O. J. Metals 1972, 24, 25
Fuel 1983,
81. Fitzer, E. in 'Proceedings of the 9th World Petroleum Congress', Applied Science, London, 1975, Vol 5, P 307
82. Ihnatowicz, H., Chiche, P., Deduit, J., Pregermain, S. and Tournant, R. Carbon. 1966, 4, 41
83. Patrick, J. W., Reynolds, M. J. and Shaw, F. S. Carbon 1975, 13, 509
84. Patrick, J. W., Reynolds, M. J. and Shaw, F. S. Fuel 1979, 58, 501
85. Marsh, H. and Smith, J. in 'Analytical methods for coal and coal products', (C. Karr Jr. ed.), Academic Press, New York, 1978, Vol 2, P 371
86. Marsh, H. in 'Proceedings of the 41st Ironmaking Conference', Pittsburgh, 1982, A. I.M.E. Iron and· Steel Soc., 1982, P 2
87. Marsh, H., Forrest, M. and Pacheco, L. A. Fuel 1981, 60, 423
88. Oya, A., Qian, Z. and Marsh, H. Fuel 1983, 62, 274
-132-
89. Marsh, H. and Cornford, D. W. American Chemical Society Symposium, Series 21, 'Petroleum Derived Carbons', 1976, 266
90. Sanada, Y., Furuta, T., Kimura, H. and Honda, H. Fuel 1973, 52, 143
91. Patrick, J. W. , Reynolds, M. J. and Walker, A. Fuel 1983, 62, 131
92. Gray, R. J. Proceedings of the Illinois Mining Institute 1976, 20
93. Goldring, D. ( Private Communication )
94. Koba, K. and Ida, S. Fuel 1980, 59, 59
95. Grint, A. and Marsh, H. Fuel 1981, 60, 513
96. Vogt, D. (Private Communication' )'
97. Mitchell, G. D. and Benedict, L. D. in 'Proceedings of the 42nd Ironmaking Conference', Atlanta, 1983, A.I.M.E. Iron and Steel Soc., 1983, P 347
98. Coin, C. D. A. (Private Communication)
99. Coin, C. D. A. Fuel 1987, 66, 702
100. Ragan, S., Marsh, H. and French, M. in 'Proceedings of the Fifth London International Conference on Carbon and Graphite', Society of Chemical Industry, London, 1978, Vol 11, P 72
101. Grint, A. and Marsh, H. Fuel 1981, 60, 513
102. Ragan. S. and Marsh, H. Fuel 1981, 60, 523
103. Grint, A. and Marsh, H. Fuel 1981, 60, 519
104. Marsh, H. , Wilkinson, A. and Machnikowski, J. Fuel 1982, 61, 835
105. Forrest, M. and Marsh, H. Fuel 1981, 60, 369
106. Fukuyama, T., Funabiki, Y and Itagaki, S. J. Fuel Soc. Japan 1981, 60, 174
107. Koba, K., Sukata, K. and Ida, S. Fuel "1981, 60, 499'
108. Fujita, H., Hijiriyama, H. and Nishida, S. Fuel 1983, 62, 875
-133-
109. Patrick, J. W., Reynolds, M. J. and Shaw, F. H. in 'Proceedings of the 64th C. I.C. Coal Symposium', (A. M. Al Taweed ed.), Canadian Society of Chemical Engineering, Ottawa, 1982
110. Wilkinson, H. C. Yearbook of the Coke Oven Managers Assoc. 1971, 131
111. Ammosov, I. I., Eremin, I. V., Sukhenko, S. I. and Oshurkoval, S Koks i Khimiya 1957, 12, 9
112. Shapiro, N., Gray, R. J. and Eusner, C. R. Blast Furnace, Coke, and Raw Materials Proc. 1961, 20, 89
113. Shapiro, N. and Gray, R. J. Journal Institute of Fuel 1964, 37, 234
114. Harrison, J. A., Jackman, H. W. and Simon, J. A. Illinois State Geol. Survey Circ. 366, 1964
115. Thompson, R. R. and Benedict, L. G. Journal of Metals 1968, 20, 79
116. Steyn, J. G. D. and Smith, W. H. Coals, Gold, Base Minerals. S. A. 1977, 25, 107
117. Miyazu, T., Okuyama, Y., Fukuyama, T. and Sugimura, H. Nippon Kokan Technical Report - Overseas 1971, Dec., 15
118. Simonis, W., Gnuschke, G. and Beck, K-G. GluckaufForschungshefte 1966, 27, 105
119. Simonis, W., Gnuschke, G. and Beck, K-G. GluckaufForschungshefte 1966, 27, 181
120. Simonis, W. and Gnuschke, G. Gluckauf-Forschungshefte 1968, 29, 205
121. Simonis, W. and Mackowsky, M-Th. Gluckauf-Forschungshefte 1969, 30" 25
122. Gibson, J. Yearbook of the Coke Oven Managers Assoc. 1972, 182
123., Patrick, J. W. and Stacey, A.E. Yearbook of the Coke Oven Managers Assoc. 1982, 151,
124. Merrick, D. Fuel 1983, 62, 567
125. Moelwyn-Hughes, E. A. 'Physical Chemistry', Pergamon Press, London, 1957, p 753
-134-
126. Mochida, I., Marsh, H. and Grint, A. in 'Proceedings of the Fifth London International Conference on Carbon and Graphite', Society of Chemical Industry, London, 1978, Vol. I, P 245
127. Patrick, J. W., Shaw, F. S., and Willmers, R. R. Fuel 1981, 60, 81
128. Barnet, F. R. and Knorr, M. K. Carbon 1973, 11, 281
129. Walker, A. Fuel 1985, 64, 1327
130. Colback, P. S. B. in 'Proceedings of the First Congress of the International Society for Rock Mechanics', 1966, Vol. I, P 385
131. Smith, M. C. in 'Proceedings of the Conference on Continuum Aspects of Graphite Design', Gatlinburg 1970, ( W. I. Greenstreet and G. C. Battle Jr. eds. ), U.S.A.E.C. Conference Report 70115, 1972, P 475
132. Patrick, J. W. and Walker, A. Fuel 1985, 64, 136
133. Moreland, A., Patrick, J. W. and Walker, A. Paper accepted for publication in Fuel.
134. Patrick, J. W. and Walker, A. Fuel 1983, 62, 1079
135. Blayden, H. E., Gibson, J. and Riley, H. L. in 'Proceedings of the Conference on the Ultrafine Structure of Coals', British Coal Utilisation Research Association, London, 1944, p 176.
136. British Carbonization Research Association, Carbonization Research Report 35, . Chesterfield, 1976.
137. British Carbonization Research Association, Carbonization Research Report 60, Chesterfield 1978.
138. Patrick, J. W. and Walker, A. Fuel 1987, 66, 1523
139. Pickup, 1. M. , McEnaney, B and Cooke, R. G. Carbon 1986,
140. . inter alia Patrick, J. W . and Stacey, A. E. Fuel 1978,
24, 535
57, 258
141. . Buch, J. D. in 'Extended Abstracts, 16th Biennial Conference on Carbon' I American Carbon Society, 1983, p 400
~135-
APPENDIX 1
Fitting of INTER<31>, TRAHS<33> and lHERT<35> equations to
data
Description of algorithm
One method of finding the lowest point of a basin would be to select an
arbitary starting point on the surface having coordinates x and y in
arbitary directions X and Y, increment x and y both positively and
negatively, singly and in combination, acertain which of the eight
positions is at the lowest level, move the starting coordinates to this
position and repeat. Eventually, the lowest point of the basin would be
approached. It would only be found accurately if the increment used were
infinitely small. Essentially, this is the algorithm used to fit the
INTER(31), TRANS(33) and INERT(35) equations to the tensile strength and
textural data. Of course, many more variables were involved and the
objective was to find the lowest standard error of estimating the
measured coke tensile stengths from the equations. The essential features
of the Pascal program written for this task are listed below, the
variable declarations and the procedures necessary for data input, file
handling and result screening and printing being omitted.
On running the section of the program listed, the procedure SETSTART is
called. This fixes the starting values of the coefficients and sets the
stored standard error of estimation ( ssee ) to a high value.
The program then reiterates the following cycle of operations:
I.Procedure NEXTC is called. For an equation with n coefficients, this
increments the nth coefficient by an amount inc. However, if the
value of this coefficient then exceeds a predefined maximum value, the
value is reduced to the predefined minimum level and the n-Ith
-136-
coefficient is incremented. This value is compared with its predefined
maximum value etc.
2.After each successful increment of a coefficient, CALCULATE is called
and this uses the existing values of the coefficients to calculate the
coke tensile strengths and the corresponding standard error of
estimation ( see ). If see is less than ssee, then see becomes ssee
and the values of the coefficients are stored.
3.The end condition is reached when the value of coefficient <1) exceeds
its predefined maximum value.
Because of the high number of reiterations, program run times are long.
If ten coefficients are invorved and, for each, the difference beween the
minimum and maximum predefined values is three times the chosen
increment, the 59049 required reiterations takes some twenty hours. To
achieve the best possible fit between measured and calculated tensile
strengths requires the use of infinitely small increments and infinite
computer time. To achieve even a good fit, the program needs to be run a
number of times with different starting coefficients and increments. The
coefficients, quoted in this thesis, obtained using this algorithm should
not therefore be regarded as giving the best possible fit between
calculated strengths and measured values.
Variables used in program
cno:coke number
cnomax:total number of cokes
tno: textural component number
tnomax:total number of textural components
Utno,cnol: array containing textural composition data
ts[cno,2l: array containing measured and calculated coke tensile
strengths
lMtno,2l: array containing minimum and maximum values of
coefficients
-137-
inc: value of increment
coltnol:array containing current values of coefficients
SS:sum of squares
see: standard error of estimation
ssee:stored lowest value of see
scoltnol:array containing coefficients corresponding to ssee
Program listing
PROCEDURE SETSTART; BEGIN ssee:= 10; coltnomaxl:=. lhltnomax,11- inc;
FOR tno:= 1 TO tnomax-1 DO BEGIN coltnol:= lhltno,ll; END END;
PROCEDURE STORE; BEGIN ssee:= seej
FOR tno:= 1 TO tnomax DO BEG IN scol tnol:= col tnol; END END;
PROCEDURE CALCULATE; BEGIN ss:= 0.0; see:= 0.0;
FOR cno:= 1 TO cnomax DO BEGIN tslcno,21:= 0.0; END;
FOR cno:= 1 TO cnomax DO BEGIN FOR tno:=l TO tnomax DO BEGIN tsltno,21:= tsltno,2l+ coltnoH tlcno,tnol; END; END;
FOR cno:= 1 TO cnomax DO BEGIN ss:= ss+ SQR ( tslcno,ll- tslcno,21 ); END;
see:= SQRT( ssl cnomax )
IF see < ssee THEN STORE END;
PROCEDURE NEXTC; BEGIN co[tnomaxl:= co[tnomaxl+ inc;
FOR tno:= tnomax DOWN TO 2 DO BEGIN
-138-
IF co[tnol > Ih[tno,2J. THEN co[tno-ll:= c[tno-ll+ inc; IF co[tnol > IMtno,21 THEN co[tnol:=lMtno,ll ;END;
IF co[l] < Ih[1,21 THEN CALCULATE END;
(* Main program t)
BEGIN SETSTART; WHILE ca[l] < IM1,21 DO NEXTC END.
-139-
APPENDIX II
Triangular diagrams iso-strength line calculation
If the fractional contents of coals 1,2 and 3 in a coal blend are F" F~
and F~, the corresponding correction factors used to calculate the
textural composition are C, , C" and C~ and the textural composition of
the cokes from the three coals are T, " to T, ,'" T"" to T",,9 and T"", to
T8,9, then the fractional content, T" of textural component 1 in the coke
from the blend is given by :
F, Cl T, ,1 + F2 C2T2:,l + F::.C::;,T:::.. ,1
T,= -------------------------------r
q ( F,C,T, ,it F"C"T",it F~C~T""i ) .. ~\
the lower term being necessary to correct to a total fractioal content of
unity. But F",= 1- F,- F2 and therefore
F1C,T,,1+ C:::.T3 • 1 - F1C:::tT:;..,+ F:;;J C2T2 ,,- C3T3 ,1 ]
T,= ---------------------------------------------------r, t F"L.,
The content of the other eight textural components can be described by
similar functions.
If the coke tensile stengths are given by an MLR equation of the form , S= Kt ri.,A,T,
then substituting for T"
-140-
Thus F2= ([ K-S Jr, + b) / ([ S-K Jr2- r4).
This equation gives the variation of Fz with F, and S. Thus, at fixed
values of S, the equation gives fractional concentration terms which"lie
along an iso-strength line on the triangular diagram for the three coals
1,2and3.
-141-
APPEND IX I I I
Derivation of INTER<31>, TRANS<33> and INERT<35> equations
Multi-linear regression analysis gave strength/textural composition
relationships which failed readily to identify those textural components
contributing most strongly to coke strength. Alternative equations ,(31)
to (35) on page 82, were therefore sought and their use examined. This
appendix explains how they were derived using a very simple model of
coke structure and assuming intergranular or transgranular failure.
Model of coke structure
It is assumed that coke consists of a regular array of close-packed,
equi-sized cubic grains, and that the i textural components present are
randomly distributed. Thus, in a layer, aligned parallel to the grain
edges, 1m2 in size and consisting of N grains, there will be NT" NT2,
NT, of type 1, 2, ... i, where T" T2 etc. are the fractional contents of
components 1, 2, etc.
Equations derived assuming intergranular failure
Intergranular failure occurs when this layer is pulled away from an
adjacent layer. The coke tensile strength is assumed to be dependent upon
the probability of contact between grains of different type and the
strength of the interface between them. The probability of a position
immediately below a grain in the original layer being occupied by a grain
of type 1, 2, .. i is T, , T"" .. T,. Thus the interface is composed of
HTd T,+ T,,+ ... ·T, grains below a type 1 grain plus
HT2[ T,+ T2+'" T, grains below a type 2 grain plus
HTd T,+ T2+'" T, ) grains below a type i grain.
-142-
Then, the coke strength is given by:
S= Td T,S, ,,+ T",S, ,,,,+ T ,S, " 1+
Td T,S"",+ T"S""",+ T iS2 ,1 1+
Td T,8 1 ,,+ T",S, ,,,,+ T ,S", 1 < lILl>
where 81,2, etc. are the strengths of cokes consisting of alternate layers
of textural components 1 and 2, etc. when failing by an intergranular
failure mechanism under a tensile force applied perpendicular to the
layers. Since S, ,,,,=S""" for nine textural components, the equation reduces
to:
. f 1 S= E ET, T jS, .j
i:o, j~, < III.2>
where i may equal j. This is equivalent in form to the the INTER<31>
equation.
If, rather than being randomly distributed, all textural components were
grouped together into variously sized cubic units having, for example, 1,
10 or 100 individual units along each edge, then, if each component
present were distributed evenly among the three size ranges, the effect
would be to reduce severely the probability of grains being in contact
with grains of different type. Equation <III.2> would then reduce to:
< lII.3>
this having the form of equation <32>.
Equations derived assuming transgranular failure
In transgranular failure, the fracture path· is considered to pass through
the grains in a layer so that the tensile strength is dependent upon the
number of grains of each type, NT, etc, in the 1m'" layer and the strength
of each grain.
The tensile strength is then:
S= T,S,+ T",S",+ .... T,S, < III.4>
-143-
where S, is the tensile strength of a coke consisting of type i textural
units failing under tension by a transgranular failure mechanism. If b
inerts are considered to ~elong to a single textural class, the number of
classes reduces to eight and the equation becomes identical in form to
the TRANS(33) equation:
, s= r T,S,
L" <IIl.5)
Equations (34) and INERT(35) represent attempts to incorporate the
reactives/inerts concept used by coal petrologists into a transgranular
failure mechanism. For this purpose all textural components other than
large and small inerts are deemed reactive.
It is assumed that the strength of coke consisting of a mixture of inerts
and a single textural component varies according to
S= kI (case I) or S= k+ kI (case Il ).
It is further assumed that inerts in a coke are associated with the other
textural components present in proportion to their concentration, ie. the
ammount of inerts associated with the ith textural component is T,I/R,
where T" I and R are the fractional contents of the ith component, the
total inerts and the total reactives respectively.
For case I, and replacing the proportionality constant with S, etc, the
coke tensile strength for transgranular failure becomes :-
S= T,"'S,I/R+ T","'S"I1R+ .... T.",Sd/R (IlL6)
which reduces to:
, S= i. T ,"S.I/R
'" This has the form of equation (34).
( IlL7)
Repeating for case 11, if S. is the tensile strength of an inert-free
coke composed of the component i only, and the proportionality constant
kis numerically equal to S" the coke tensile strength becomes
S= Td S,+ S,T,I/R J+ Td S,,+ S"T"I/R J+ .. T.[ S,+ S.T.I/R 1 (IlL8)
This reduces to:
-144-
1 S= E Si[ Ti+ Ti2 1/R 1 < III.9)
'''I
which is identical in form to the INERT<35) equation.
TABLE 1. Specifications for metallurgical cokes.
Blast-furnace coke Foundry coke A B C
Hearth diameter (m) 9.5 12 14
Coke properties:-
Mean size (mm) 55 50 >100 Size range (mm) 25-80 25-75 30-75
Xicum 1(40 index (min) 75 80 78 I(icum 1(10 index (max) 8.5 7.5 7.0 2" shatter index (min) 90
Sulphur (wt%, max) 1.2 1.15 1.1 0.85 Ash (wt%, max) 8.5 • 8.5 8.0 9.0
Reactivity (max) 30 Post reaction strength (min) 53
.1
TABLE 2. Characteristics of drum tests.
A.S.T.X )licum J. 1. S. Drum dimensions : Diameter (mm) 914 1000 1500 Length (mm) 457 1000 1500
Lifting flights: Humber 2 4 6 Depth (mm) 51 100 250
Rotation Speed (rev/min) 24 25 15 Revolutions 1400 100 30/150
Test sample ){ass (kg) 10 50 10 Size (mm) 50-75 >60 >50
Strength indices : Stability= )140= 0130/01150*= wtt>25mm wtt>40mm wtt>15mm
Hardness= )110= wtt>6mm wtt<10mm
*Originally 30 but recently indices for both 30 and 150 drum revolutions have been quoted.
TABLE 3. Comparison of classification schemes for anisotropic
components in cokes from low-rank coals.
Reference Components recognised
Isotropic Isotropic Kosaics ( sizes in jllll)
plastic Very fine Fine Xedium Coarse
11 Yes Ba 0.3 0.'1 1.2
92 Yes Ba 0-0.5 0.5-1. 0 1. 0-1. 5 1.5-2.0
93 Yes Ba <0.1 0.1-----1. 0 >1. 0
94 Yes Ba <1. 0 >1.0
95 Yes Ba <1.5 1. 5-5.0 5.0-10.0
96 Yes Yes <1. 0 1. 0-5. 0 5.0-10.0
9'1 Yes Ba <1 1.0-3.0 3.0-12.0 >12.0
98 Yes Ba <0.5 0.5-2.5 2.5-5.0 )5.0
TABLE 4. Comparison of classification schemes for anisotropic components in cokes from high-rank coals.
Reference
11
92
93
94
95
96
97
98
Components recognised
Granular-flow Flow Basic
Lenticular, widths: 2-5, 5-10, 10-15 Ribbon, widths: 15-20, 20-25, >25
Sinous Lamellar
Fibrous, flow-type with l:b>3 and 1>5 Leaflet, flat and featureless >20
Flow-type, Domains,
<30t<5, 30-60t5-10, >60t>10 >60t60
Fibrous, lob >2, fine b<5, medium b=5-10, coarse b>10
Anthraci tic
Flow
Foliate Basic
TABLE 5. licum 140 indices calculated by German and British methods.
Charge details German method Bri tish method
VX wt% G value A B C 140 140
20 1 88.3 -0.12 -0.13 83.1 81. 3
25 1 101. 0 -0.78 -0.15 79.7 81. 7
30 1 131. 8 -2.55 -0.09 69.7 80.2
35 1 185.1 -5.72 -0.31 43.7 69.9
TABLE 6. Analytical data for coals used.
Coal N.C.B. Air-dried-basis Dry-ash-free-basis class ){oisture Ash C H Volatile matter
wt% wt% wt% wt% wt%
A 204 0.8 7.5 91.5 4.3 19.7
B 30la 0.6 4.9 90.6 4.6 20.2
C 30lh 0.8 1.3 89.5 4.9 26.4
D 401 0.9 1.5 87.2 6.4 36.4
E 501 2.4 4.2 86.0 5.3 35.0
F 602 2.4 5.4 83.4 5.3 36.9
TABLE 7. Strengths and yields of single-coal cokes.
Coal Coke tensile strength Fractional coke yields (w/w) lIPa Standard error Jleasured (Y) Calculated (Z)
A 4.92 0.28 0.824 0.811 B 6.12 0.23 0.828 0.803 C 6.26 0.29 0.775 0.734 D 6.61 0.21 0.719 0.636 E 5.86 0.26 0.709 0.649 F 4.42 0.21 0.679 0.636
TABLE 6. SEX textural component classification.
Component Initial
Flat F
Lamellar L
Intermediate I
Granular: coarse medium fine very
Inerts:
large small
fine
Gc Gm Gf Gvf
In
1nl Ins
Appearance of etched surface
Generally rather flat, sometimes with a fine granularity. Some regions contain scattered circular pits or short, narrow channels.
Surface consists of parallel ridges and channels >5~m long. Channels are usually about 0.5~m wide, ridges vary up to 3~m.
Intermediate in appearance between lamellar and granular forms with short ( <4~m ) channels, often branched.
Uniform, pitted surface. Pit size approximately 0.2 -0.35~m. Pit size approximately 0.15-0.2~m. Pit size approximately 0.1 -0.15~m. Pit size approximately <O.l~m.
Carbonaceous inerts are identifiable by their woody structure, or if small by their unfused sharp edges. Particles often appear darker and more deeply etched than the reactive matrix. Mineral matter is also included in this class. >50~m.
<50~m.
TABLE 9. SEX textural composition of single-coal cokes.
Coal Fractional textural composition (v/v)
1nl Ins F L I Gc Gm Gf Gvf A .170 .084 .094 .372 .194 .074 .010 .002 0 B .078 .082 .074 .431 .303 .025 .006 0 0 C .165 .117 .032 .285 .348 .048 .002 .003 0 D .064 .092 .006 .042 .502 .270 .016 .004 0 E .099 .139 .002 .004 .014 .058 .596 .070 .018 F .160 .136 0 .002 0 .026 .498 .150 .028
TABLE 10. Comparison of the frequency of observation of coke textural components within the coke matrix and at cracks.
Posi tion Fractional frequency of observation of components
In F L I Gc Gm Gf Gvf
Within coke matrix .12 0 .17 .31 .03 .17 .15 .05
At microcracks .02 0 .20 .36 .02 .20 .15 .04
At fracture cracks .06 0 .19 .38 .03 .16 .16 .03
•
TABLE 11. Experimental data for blended-coal carbonizations.
Blend Fractional blend composition Tensile Fractional coke number Coal Coal Coal Coal Coal Coal strength yields (w/w)
A B C D E F (JlPa) Measured Calculated (Yb) (Zb)
1 .205 .000 .000 .000 .000 .795 4.43 .710 .709 .672 2 .103 .000 .181 .000 .000 .716 4.42 .701 .711 .672 3 .000 .000 .364 .000 .000 .636 5.27 .708 .714 .672 4 .365 .000 .000 .000 .000 .635 4.52 .726 .732 .700 5 .182 .000 .326 .000 .000 .492 5.20 .736 .737 .700 6 .000 .000 .645 .000 .000 .355 5.87 .736 .741 .699 7 .526 .000 .000 .000 .000 .474 4.65 .754 .755 .728 8 .264 .000 .469 .000 .000 .267 5.59 .770 .762 .728 9 .000 .000 .940 .000 .000 .060 5.77 .752 .769 .728
10 .688 .000 .000 .000 .000 .312 4.90 .770 .779 .756 11 .491 .000 .353 .000 .000 .156 5.69 .788 .784 .757 12 .292 .000 .708 .000 .000 .000 5.76 .786 .789 .756 13 .169 .000 .000 .000 .435 .396 5.09 .697 .717 .671 14 .129 .000 .000 .000 .871 .000 5.51 .709 .724 .670 15 .338 .000 .000 .000 .346 .316 5.41 .723 .738 .700 16 .307 .000 .000 .000 .693 .000 5.11 .736 .744 .699 17 .523 .000 .000 .000 .242 .235 5.32 .752 .762 .731 18 .484 .000 .000 .000 .516 .000 5.43 .759 .765 .727 19 .676 .000 .000 .000 .170 .154 5.69 .775 .782 .757 20 .661 .000 .000 .000 .339 .000 5.40 .784 .785 .756 21 .326 .336 .000 .000 .338 .000 4.52 .790 .786 .754 22 .000 .672 .000 .000 .328 .000 5.79 .801 .789 .752 23 .238 .243 .000 .000 .519 .000 5.31 .765 .765 .725 24 .000 .488 .000 .000 .512 .000 4.91 .769 .767 .724 25 .147 .152 .000 .000 .701 .000 5.94 .751 .744 .696 26 .000 .304 .000 .000 .696 .000 5.70 .744 .745 .696 27 .059 .060 .000 .000 .881 .000 6.10 .722 .723 .668 28 .000 .122 .000 .000 .878 .000 5.35 .731 .724 .668 29 .433 .000 .435 .000 .132 .000 5.30 .781 .788 .756 30 .293 .000 .707 .000 .000 .000 6.16 .785 .789 .757 31 .315 .000 .317 .000 .368 .000 6.06 .763 .766 .727 32 .000 .000 .928 .000 .072 .000 6.64 .771 .770 .728 33 .198 .000 .198 .000 .604 .000 6.15 .735 .745 .698 34 .000 .000 .581 .000 .419 .000 6.28 .733 .747 .698 35 .079 .000 .080 .000 .841 .000 6.29 .737 .723 .669 36 .000 .000 .233 .000 .767 .000 6.05 .710 .724 .669 37 .671 .000 .000 .162 .167 .000 5.17 .778 .788 .755 38 .684 .000 .000 .316 .000 .000 5.25 .794 .791 .755 39 .500 .000 .000 .248 .252 .000 6.27 .765 .769 .727 40 .520 .000 .000 .480 .000 .000 5.02 .769 .774 .727 41 .328 .000 .000 .334 .338 .000 6.64 .740 .750 .697 42 .355 .000 .000 .645 .000 .000 6.96 .747 .756 .697 43 .156 .000 .000 .421 .423 .000 5.69 .728 .731 .668 44 .190 .000 .000 .810 .000 .000 6.78 .737 .739 .668
I: .
TABLB 12 SEX textural composition of cokes from blended coal charges calculated using method C.
Coke Tensile Fractional textural composition number strength 1nl Ins F L I Gc Gm Gf Gvf
(XPa) 1 4.43 .162 .125 .019 .078 .040 .036 .398 .120 .022 2 4.42 .162 .127 .015 .091 .082 .035 .358 .108 .020 3 5.27 .162 .129 .012 .105 .125 .034 .317 .096 .018 4 4.52 .164 .117 .034 .137 .071 .044 .320 .096 .018 5 5.20 .163 .120 .028 .162 .147 .042 .247 .075 .014 6 5.87 .163 .124 .021 .185 .221 .040 .178 .055 .010 7 4.65 .165 .109 .049 .197 .102 .051 .241 .072 .013 8 5.59 .165 .113 .040 .232 .212 .049 .137 .042 .007 9 5.77 .165 .118 .030 .268 .322 .047 .032 .012 .002
10 4.90 .167 .100 .065 .257 .133 .059 .162 .048 .009 11 5.69 .167 .104 .057 .284 .216 .057 .083 ".025 .004 12 5.76 .166 .107 .050 .310 .299 .056 .004 .003 .000 13 5.09 .135 .129 .017 .065 .039 .048 .458 .090 .019 14 5.51 .108 .132 .014 .051 .037 .060 .520 .061 .016 15 5.41 .142 .119 .032 .128 .070 .053 .367 .072 .015 16 5.11 .121 .122 .030 .117 .069 .063 .416 .049 .012 17 5.32 .150 .110 .050 .196 .105 .059 .266 .053 .011 18 5.43 .133 .112 .047 .182 .101 .066 .312 .037 .009 19 5.69 .156 .101 .064 .252 .134 .064 .185 .036 .007 20 5.40 .146 .103 .063 .247 .133 .069 .209 .025 .006 21 4.52 .115 .102 .056 .267 .170 .052 .207 .024 .006 22 5.79 .085 .101 .050 .291 .208 .036 .200 .023 .006 23 5.31 .111 .112 .041 .195 .127 .054 .313 .037 .009 24 4.91 .089 .111 .037 .212 .155 .042 .308 .036 .009 25 5.94 " .106 .122 .026 .123 .084 .055 .420 .049 .013 26 5.70 .093 .122 .024 .134 .102 .048 .417 .049 .013 27 6.10 .102 .132 .012 .051 .042 .057 .526 .062 .016 28 5.35 .096 .132 .011 .056 .049 .054 .524 .061 .016 29 5.30 .158 .106 .055 .286 .235 .061 .084 .011 .002 30 6.16 .166 .107 .050 .310 .299 .056 .004 .003 .000 31 6.06 .142 .115 .040 .209 .175 .060 .223 .027 .007 32 6.64 .160 .119 .030 .265 .319 .049 .045 .008 .001 33 6.15 .126 .124 .026 .133 .115 .059 .362 .043 .011 34 6.28 .137 .126 .019 .167 .205 .052 .251 .031 .008 35 6.29 .110 .133 .012 .056 .055 .058 .502 .059 .015 36 6.05 .114 .134 .009 .069 .091 .056 .458 .054 .014 37 5.17 .141 .094 .064 .257 .214 .103 .109 .014 .004 38 5.25 .137 .087 .066 .268 .291 .136 .012 .003 .001 39 6.27 .126 .100 .049 .197 .225 .119 .159 .020 .006 40 5.02 .119 .088 .052 .214 .342 .168 .013 .003 .002 41 6.64 .111 .105 .034 .137 .236 .134 .210 .026 .007 42 6.96 .102 .089 .037 .159 .393 .200 .014 .003 .003 43 5.69 .095 .111 .018 .077 .248 .150 .260 .032 .009 44 6.78 .084 .090 .023 .105 .443 .233 .015 .004 .003
TABLE 13 SEK textur~l composition of cokes from blended coal charges calculated using method Y.
Coke Tensile Fractional textural composition number strength 1nl Ins F L I Gc Gm Gf Gvf
OlPa) 1 4.43 .162 .124 .022 .090 .046 .037 .382 .115 .021 2 4.42 .162 .126 .018 .102 .091 .036 .342 .103 .019 3 5.27 .162 .128 .013 .114 .136 .035 .302 .092 .017 4 4.52 .164 .115 .039 .154 .080 .046 .297 .089 .016 5 5.20 .164 .119 .030 .174 .157 .043 .229 .069 .013 6 5.87 .163 .123 .022 .193 .231 .041 .163 .051 .009 7 4.65 .166 .106 .054 .214 .111 .054 .218 .065 .012 8 5.59 .165 .112 .042 .243 .219 .050 .122 .038 .007 9 5.77 .165 .118 .030 .270 .325 .047 .028 .011 .001
10 4.90 .167 .098 .068 .271 .141 .061 .143 .042 .008 11 5.69 .167 .103 .060 .292 .220 .058 .073 .022 .004 12 5.76 .167 .107 .051 .312 .298 .056 .004 .003 .000 13 5.09 .136 .127 .019 .075 .044 .049 .445 .087 .018 14 5.51 .109 .131 .016 .058 .040 .060 .510 .060 .015 15 5.41 .144 .117 .036 .142 .078 .055 .346 .068 .014 16 5.11 .123 .120 .033 .129 .075 .063 .397 .047 .012 17 5.32 .152 .107 .054 .212 .113 .060 .244 .048 .010 18 5.43 .136 .110 .050 .196 .108 .066 .290 .035 .009 19 5.69 .158 .099 .067 .266 .140 .065 .166 .032 .007 20 5.40 .148 .101 .066 .259 .139 .069 .189 .023 .006 21 4.52 .116 .100 .059 .281 .178 .052 .187 .022 .005 22 5.79 .084 .099 .053 .305 .218 .035 .180 .021 .005 23 5.31 .112 .110 .045 .211 .136 .053 .291 .034 .009 24 4.91 .088 .109 .040 .229 .166 .041 .285 .033 .009 25 5.94 .107 .120 .029 .136 .092 .055 .401 .047 .012 26 5.70 .092 .120 .026 .148 .112 .047 .397 .046 .012 27 6.10 .102 .131 .013 .058 .046 .057 .516 .061 .016 28 5.35 .096 .131 .012 .064 .054 .053 .514 .060 .015 29 5.30 .159 .105 .057 .291 .236 .061 .076 .011 .002 30 6.16 .167 .107 .051 .312 .297 .056 .004 .003 .000 31 6.06 .144 .113 .043 .219 .180 .060 .207 .025 .006 32 6.64 .161 .118 .030 .266 .321 .049 .041 .007 .001 33 6.15 .128 .122 .028 .142 .121 .059 .345 .041 .010 34 6.28 .139 .126 .020 .173 .212 .052 .238 .030 .007 35 6.29 .111 .132 .013 .061 .058 .059 .492 .058 .015 36 6.05 .115 .134 .009 .074 .096 .056 .448 .053 .014 37 5.17 .144 .093 .067 .268 .212 .101 .099 .013 .003 38 5.25 .140 .086 .069 .277 .282 .130 .012 .003 .001 39 6.27 .129 .099 .052 .210 .224 .116 .148 .018 .005 40 5.02 .123 .088 .055 .225 .331 .161 .013 .003 .002 41 6.64 .113 .104 .036 .149 .235 .132 .199 .024 .007 42 6.96 .105 .089 .040 .170 .383 .194 .014 .003 .002 43 5.69 .097 .110 .020 .084 .248 .149 .253 .031 .009 44 6.78 .086 .090 .025 .112 .437 .228 .015 .004 .003
---- - -----
TABLE 14 SIlK textural composition of cokes from blended coal charges calculated using method V.
Coke Tensile Fractional textural composition number strength 1nl Ins F L I Gc Gm Gf Gvf
OlPa) 1 4.43 .162 .123 .023 .094 .048 .038 .377 .113 .021 2 4.42 .162 .126 .018 .104 .092 .036 .339 .103 .019 3 5.27 .162 .128 .013 .115 .136 .035 .301 .092 .017 4 4.52 .164 .114 .040 .158 .082 .046 .292 .087 .016 5 5.20 .164 .119 .031 .177 .158 .044 .225 .069 .013 6 5.87 .163 .123 .022 .194 .232 .041 .162 .050 .009 7 4.65 .166 .106 .055 .219 .114 .054 .212 .063 .012 8 5.59 .165 .112 .043 .245 .219 .051 .120 .037 .007 9 5.77 .165 .118 .030 .270 .325 .047 .028 .011 .001
10 4.90 .167 .098 .069 .275 .143 .061 .138 .041 .007 11 5.69 .167 .102 .060 .294 .220 .059 .071 .022 .004 12 5.76 .167 .107 .051 .312 .296 .056 .005 .003 .000 13 5.09 .136 .127 .020 .078 .046 .049 .440 .086 .018 14 5.51 .110 .130 .016 .061 .042 .060 .504 .059 .015 15 5.41 .144 .117 .037 .148 .081 .055 .338 .066 .014 16 5.11 .124 .119 .035 .135 .078 .064 .387 .046 .012 17 5.32 .153 .106 .055 .217 .116 .061 .236 .047 .010 18 5.43 .137 .109 .052 .203 .111 .067 .280 .033 .008 19 5.69 .158 .099 .068 .270 .143 .065 .159 .031 .006 20 5.40 .149 .100 .067 .265 .142 .069 .181 .022 .005 21 4.52 .116 .099 .060 .286 .181 .052 .179 .021 .005 22 5.79 .084 .098 .054 .310 .221 .034 .173 .020 .005 23 5.31 .112 .109 .046 .217 .140 .053 .281 .033 .008 24 4.91 .088 .108 .041 .235 .170 .040 .277 .032 .008 25 5.94 .107 .120 .030 .142 .095 .055 .392 .046 .012 26 5.70 .092 .119 .027 .154 .115 .046 .389 .045 .012 27 6.10 .103 .131 .014 .061 .048 .057 .511 .060 .015 28 5.35 .096 .131 .013 .067 .056 .053 .509 .060 .015 29 5.30 .160 .104 .057 .294 .237 .061 .073 .010 .002 30 6.16 .167 .107 .051 .312 .296 .056 .005 .003 .000 31 6.06 .145 .113 .044 .223 .183 .060 .200 .025 .006 32 6.64 .161 .118 .030 .267 .322 .049 .040 .007 .001 33 6.15 .129 .122 .029 .147 .124 .060 .337 .040 .010 34 6.28 .139 .126 .020 .176 .215 .052 .233 .029 .007 35 6.29 .112 .132 .013 .064 .060 .059 .488 .058 .015 36 6.05 .116 .133 .010 .076 .098 .055 .444 .053 .013 37 5.17 .145 .093 .069 .274 .210 .098 .095 .012 .003 38 5.25 .142 .086 .071 .284 .276 .126 .012 .003 .001 39 6.27 .131 .098 .054 .218 .220 .113 .143 .018 .005 40 5.02 .126 .087 .057 .234 .323 .156 .013 .003 .002 41 6.64 .115 .104 .038 .156 .231 .129 .196 .024 .007 42 6.96 .108 .089 .042 .178 .375 .189 .014 .003 .002 43 5.69 .098 .110 .021 .089 .243 .146 .253 .031 .009 44 6.78 .088 .090 .026 .118 .431 .225 .015 .004 .003
TABLB 15. Coefficients obtained by applying the XLR<29> equation to SBX textural data calculated using methods C, Y and V.
Textural component
Constant
Flat
Lamellar
Intermediate
Granular:
coarse
medium
fine
very fine
Inerts:
large
small
Standard error of estimation:-
Initial Coefficients in XLR<29> equation for :-C data Y data V data
K -82.84 -10.22 -7.00
F -737.16 136.27 -51.53
L 357.91 -24.97 25.21
I -907.34 40.74 -12.54
Gc 669.24 -25.13 36.42
Gm -932.27 20.78 -7.98
Gf -2362.93 4.00 -59.21
Gvf 17933.22 -95.74 373.57
1nl -1034.79 17.39 -4.41
Ins 4053.24 30.42 98.10
0.445 0.443 0.442
TABLE 16. Comparison of measured coke tensile strengths with strengths calculated using the XLR<29) equations.
Coke number
1 2 3 4 5 6 7 8 9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44
Coke tensile strengths, IPa. Xeasured
4.43 4.42 5.27 4.52 5.20 5.87 4.65 5.59 5.77 4.90 5.69 5.76 5.09 5.51 5.41 5.11 5.32 5.43 5.69 5.40 4.52 5.79 5.31 4.91 5.94 5.70 6.10 5.35 5.30 6.16 6.06 6.64 6.15 6.28 6.29 6.05 5.17 5.28 6.27 5.02 6.64 6.96 5.69 6.78
Textural data calculation method C y V 4.50 4.45 4.45 4.82 4.79 4.79 5.34 5.13 5.13 4.61 4.59 4.60 5.18 5.18 5.18 5.73 5.75 5.75 4.71 4.72 4.74 5.53 5.54 5.54 6.36 6.36 6.36 4.82 4.85 4.86 5.43 5.49 5.45 6.06 6.04 6.04 5.08 5.14 5.13 5.83 5.82 5.82 5.14 5.13 5.12 5~6 5~5 5M 5.09 5.09 5.10 5.50 5.49 5.49 5.08 5.10 5.10 5.33 5.34 5.34 5.24 5.24 5.24 5.13 5.14 5.14 5A3 5.U 5.42 5.36 5.34 5.34 5~3 5~1 5~1 5.58 5.56 5.56 5.82 5.81 5.81 5.80 599 599 5.78 5.77 5.77 6.05 6.04 6.03 5.82 5.82 5.82 6.44 6.44 6.44 5.87 5.86 5.86 626 6.~ 627 5.92 5.91 5.91 6.07 6.08 6.09 5~0 5~0 5A9 5.65 5.65 5.64 595 5.74 593 5.98 5.97 5.96 6.00 6.00 5.99 6.31 6.30 6.30 6.25 6.27 6.27 6.64 6.66 6.68
TABLE 17. Coefficients obtained by applying the NOKXLR(30) equation to SEX textural data calculated using methods C, Y and V.
Textural Initial Coefficients in NOKXLR(30) equation for component C data Y data V data
Flat F 1.74
Lamellar L 1.58 1.25
Intermediate I 11.08 -11.11 0.96
Granular:
coarse Gc -0.19 21.71 7.78
medium Gm 7.62 -6.61 -0.71
fine Gf -35.74 58.71 -24.42
very fine Gvf 118.80 -445.80 60.5
Inerts:
large 1nl 16.19 -25.81
small Ins 109.79 46.83
Standard error of estimation:- 0.391 0.387 0.389
indicates that the equation was derived without including this component.
TABLE 18. . Comparison of measured coke tensile strengths with strengths calculated using NOKKLR<30> equations.
Coke tensile strengths, MPa. Coke Measured Textural data calculation method number C Y V
1 4.43 4.45 4.46 4.44 2 4.42 4.80 4.85 4.79 3 5.27 5.17 5.06 5.11 4 4.52 4.64 4.62 4.60 5 5.20 5.20 5.01 5.22 6 5.87 5.73 5.68 5.73 7 4.65 4.69 4.52 4.74 8 5.59 5.47 5.75 5.50 9 5.77 6.39 6.49 6.31
10 4.90 4.87 4.97 4.84 11 5.69 5.40 5.63 5.45 12 5.76 6.04 5.93 6.05 13 5.09 5.17 5.18 5.17 14 5.51 5.84 5.77 5.79 15 5.41 5.12 5.07 5.16 16 5.11 5.58 5.63 5.67 17 5.32 5.11 5.20 5.05 18 5.43 5.46 5.46 5.44 19 5.69 5.03 5.12 5.08 20 5.40 5.34 5.32 5.33 21 4.52 5.26 5.23 5.24 22 5.79 5.17 5.11 5.16 23 5.31 5.40 5.50 5.41 24 4.91 5.34 5.41 5.35 25 5.94 5.63 5.38 5.60 26 5.70 5.52 5.78 5.60 27 6.10 5.82 5.94 5.83 28 5.35 5.83 5.30 5.76 29 5.30 5.75 5.63 5.79 30 6.16 6.05 5.96 6.08 31 6.06 5.87 6.12 5.82 32 6.64 6.40 6.33 6.46 33 6.15 5.89 5.84 5.83 34 6.28 6.30 6.23 6.25 35 6.29 5.90 6.20 5.92 36 6.05 6.09 6.17 6.13 37 5.17 5.55 5.43 5.50 38 5.28 5.64 5.48 5.66 39 6.27 5.80 6.05 5.78 40 5.02 6.00 5.67 6.01 41 6.64 5.93 6.29 6.01 42 6.96 6.39 6.39 6.30 43 5.69 6.20 6.27 6.26 44 6.78 6.59 6.66 6.63
TABLE 19. Coefficients obtained by applying the IHTER(31) equation to textural data calculated using methods C, Y, and V.
Coefficients obtained using data calculated by method C. 1nl Ins F L I Gc Gm Gf Gvf
1nl 0 0 3.0 3.0 3.0 3.0 3.0 3.0 3.0 Ins 0 0 4.5 4.5 4.5 4.5 4.5 4.5 4.5 F 3.0 4.5 3.0 2.5 2.5 2.5 2.5 2.5 3.0 L 3.0 4.5 2.5 3.5 112.4 7.0 7.0 1 5.5 2.5 I 3.0 4.5 2.5 12.4 8.2 9.5 9.0 B.7 2.5. Gc 3.0 4.5 2.5 7.0 9.5 5.0 9.3 5.5 2.5 Gm 3.0 4.5 2.5 7.0 9.0 9.3 B.l 5.0 2.5 Gf 3.0 4.5 2.5 5.5 B.7 5.5 5.0 3.5 2.5 Gvf 3.0 4.5 2.5 2.5 2.5 2.5 2.5 2.5 3.0
Standard error of estimation = 0.47B KPa.
Coefficients obtained using data calculated by method Y. 1nl Ins F L I Gc Gm Gf Gvf
1nl 0 0 3.0 3.0 3.0 3.0 3.0 3.0 3.0 Ins 0 0 4.5 4.5 4.5 4.5 4.5 4.5 4.5 F 3.0 4.5 3.0 2.5 2.5 2.5 2.5 2.5 3.0 L 3.0 4.5 2.5 3.5 12.4 7.0 7.0 5.5 2.5 I 3.0 4.5 2.5 12.4 B.4 9.5 9.0 B.5 2.5 Gc 3.0 4.5 2.5 7.0 9.5 5.0 9.4 5.5 2.5 Gm 3.0 4.5 2.5 7.0 9.0 9.4 B.l 5.0 2.5 Gf 3.0 4.5 2.5 5.5 8.5 5.5 5.0 3.5 2.5 Gvf 3.0 4.5 2.5 2.5 2.5 2.5 2.5 2.5 3.0
Standard error of estimation = 0.477 KPa.
Coefficients obtained using data calculated by method Y. 1nl Ins F L I Gc Gm Gf Gvf
1nl 0 0 3.0 3.0 3.0 3.0 3.0 3.0 3.0 Ins 0 0 4.5 4.5 4.5 4.5 4.5 4.5 4.5 F 3.0 4.5 3.0 2.5 2.5 2.5 2.5 2.5 3.0 L 3.0 4.5 2.5 3.5 \12.4 7.0 7.0 5.5 2.5 I 3.0 4.5 2.5 12.4 B.7 9.5 9.0 B.5 2.5 Gc 3.0 4.5 2.5 7.0 9.5 5:0 9.4 5.5 2.5 Gm 3.0 4.5 2.5 7.0 9.0 9.4 B.l 5.0 2.5 Gf 3.0 4.5 2.5 5.5 B.5 5.5 5.0 3.5. 2.5 Gvf 3.0 4.5 2.5 2.5 2.5 2.5 2.5 2.5 3.0
Standard error of estimation = 0.475 KPa.
TABLE 20. Comparison of measured coke tensile strengths with strengths calculated using the IHTER(31) equation.
Coke tensile strengths, XPa. Coke Measured Textural data calculation method number C Y V
1 4.43 4.85 4.86 4.87 2 4.42 5.00 5.03 5.03 3 5.27 5.16 5.19 5.20 4 4.52 4.90 4.92 4.92 5 5.20 5.19 5.22 5.23 6 5.87 5.47 5.51 5.53 7 4.65 4.95 4.96 4.97 8 5.59 5.39 5.41 5.42 9 5.77 5.82 5.85 5.88
10 4.90 4.99 5.00 5.01 11 5.69 5.33 5.35 5.36 12 5.76 5.68 5.68 5.70 13 5.09 5.21 5.23 5.22 14 5.51 5.62 5.61 5.61 15 5.41 5.19 5.19 5.19 16 5.11 5.49 5.47 5.46 17 5.32 5.14 5.14 5.14 18 5.43 5.37 5.35 5.34 19 5.69 5.12 5.12 5.12 20 5.40 5.25 5.23 5.23 21 4.52 5.65 5.65 5.66
. 22 5.79 6.05 6.08 6.11 23 5.31 5.65 5.65 5.65 24 4.91 5.93 5.96 5.98 25 5.94 5.67 5.67 5.67 26 5.70 5.83 5.85 5.86 27 6.10 5.70 5.70 5.70 28 5.35 5.76 5.77 5.77 29 5.30 5.51 5.51 5.52 30 6.16 5.68 5.68 5.70 31 6.06 5.53 5.53 5.53 32 6.64 5.86 5.88 5.91 33 6.15 5.59 5.57 5.57 34 6.28 5.75 5.76 5.78 35 6.29 5.66 5.65 5.65 36 6.05 5.71 5.71 5.72 37 5.17 5.60 5.56 5.54 38 5.28 5.89 5.84 5.81 39 6.27 5.86 5.82 5.80 40 5.02 6.25 6.21 6.18 41 6.64 6.11 6.08 6.06 42 6.96 6.56 6.54 6.53 43 5.69 6.35 6.34 6.33 44 6.78 6.82 6.83 6.85
TABLE 21. Coefficients obtained by applying the TRABS(33) equation to textural data calculated using methods C, Y and V.
Textural Initial Coefficients in TRABS(33) equation for component C data Y data V data
Flat F 0.8 0.8 0.8
Lamellar L 5.5 5.5 5.6
Intermediate I 11.0 11.0 11.1
Granular:
coarse Gc 4.5 4.5 4.7
medium Gm 7.7 7.7 7.7
fine Gf 2.2 2.2 2.2
very fine Gvf 1.0 0.9 1.0
Inerts In 1.4 1.4 1.3
Standard error of estimation:- 0.461 0.462 0.460
TABLE 22. Comparison of measured coke tensile strengths .. i th strengths calculated using the TRAHS(33) equations.
Coke tensile strengths, XPa. Coke Measured Textural data calculation method number C Y V
1 4.43 4.80 4.80 4.80 2 4.42 4.99 5.02 5.02 3 5.27 5.19 5.24 5.24 4 4.52 4.85 4.85 4.86 5 5.20 5.20 5.22 5.24 6 5.87 5.55 5.58 5.61 7 4.65 4.88 4.90 4.93 8 5.59 5.40 5.42 5.45 9 5.77 5.92 5.93 5.97
10 4.90 4.93 4.94 4.97 11 5.69 5.32 5.33 5.36 12 5.76 5.71 5.71 5.74 13 5.09 5.13 5.13 5.13 14 5.51 5.46 5.45 5.44 15 5.41 5.10 5.11 5.11 16 5.11 5.37 5.36 5.36 17 5.32 5.08 5.08 5.10 18 5.43 5.28 5.27 5.28 19 5.69 5.07 5.06 5.09 20 5.40 5.20 5.18 5.21 21 4.52 5.57 5.58 5.61 22 5.79 5.95 5.97 6.01 23 5.31 5.56 5.57 5.59 24 4.91 5.83 5.85 5.89 25 5.94 5.54 5.55 5.56 26 5.70 5.73 5.74 5.76 27 6.10 5.54 5.54 5.54 28 5.35 5.60 5.60 5.60 29 5.30 5.52 5.50 5.54 30 6.16 5.71 5.70 5.74 31 6.06 5.52 5.50 5.53 32 6.64 5.97 5.96 6.01 33 6.15 5.53 5.51 5.52 34 6.28 5.80 5.81 5.84 35 6.29 5.53 5.52 5.53 36 6.05 5.65 5.65 5.65 37 5.17 5.49 5.44 5.45 38 5.28 5.75 5.68 5.70 39 6.27 5.72 5.69 5.68 40 5.02 6.14 6.05 6.06 41 6.64 5.96 5.92 5.93 42 6.96 6.51 6.44 6.45 43 5.69 6.21 6.19 6.19 44 6.78 6.89 6.84 6.88 ,
TABLB 23. Coefficients obtained by applying the IHBRT(35) equation to textural data calculated using methods C, Y and V.
Textural Initial Coefficients in IHBRT(35) equation for component C data Y data V data
Flat F 0.7 0.7 0.7
Lamellar L 5.9 5.9 5.9
Intermediate I 10.8 10.8 10.8
Granular:
coarse Gc 3.6 3.7 3.7
medium Gm 7.8 7.8 7.8
fine Gf 0.9 1.0 0.9
very fine Gvf 0.9 0.9 0.9
Standard error of estimation:- 0.453 0.452 0.453
TABLE 24. Comparison of measured coke tensile strengths with strengths calculated using the IHERT(35) equation.
Coke tensile strengths. IIPa. Coke Measured Textural data calculation method number C y V
1 4.43 4.79 4.78 4.78 2 4.42 4.94 4.95 4.96 3 5.27 5.11 5.16 5.17 4 4.52 4.73 4.74 4.74 5 5.20 5.08 5.13 5.13 6 5.87 5.50 5.58 5.59 7 4.65 4.75 4.77 4.78 8 5.59 5.36 5.41 5.42 9 5.77 6.16 6.22 6.22
10 4.90 4.82 4.86 4.87 11 5.69 5.33 5.37 5.35 12 5.76 5.95 5.96 5.95 13 5.09 5.27 5.25 5.24 14 ·5.51 5.75 5.72 5.71 15 5.41 5.10 5.09 5.10 16 5.11 5.47 5.43 5.41 17 5.32 4.99 4.99 4.99 18 5.43 5.27 5.24 5.23 19 5.69 4.98 5.00 5.00 20 5.40 5.14 5.14 5.13 21 4.52 5.57 5.59 5.60 22 5.79 5.99 6.04 6.05 23 5.31 5.57 5.57 5.57 24 4.91 5.86 5.89 5.90 25 5.94 5.65 5.63 5.63 26 5.70 5.81 5.82 5.83 27 6.10 5.82 5.80 5.80 28 5.35 5.88 5.87 5.87 29 5.30 5.57 5.59 5.58 30 6.16 5.94 5.96 5.95 31 6.06 5.49 5.50 5.49 32 6.64 6.19 6.24 6.24 33 6.15 5.56 5.55 5.54 34 6.28 5.82 5.85 5.86 35 6.29 5.78 5.76 5.75 36 6.05 5.82 5.82 5.82 37 5.17 5.41 5.40 5.38 38 5.28 5.75 5.71 5.67 39 6.27 5.61 5.59 5.56 40 5.02 6.11 6.07 6.01 41 6.64 5.84 5.82 5.78 42 6.96 6.49 6.46 6.39 43 5.69 6.08 6.08 6.05 44 6.78 6.87 6.86 6.82
TABLE 25. PLM textural component classification.
Component Ini tial
Anthraci tic:
plain A patterned Ap
Flow:
broad striated granular
Mosaic:
coarse medium fine
Isotropic:
Inerts:
large small
Fb Fs Fg
Mc Km Xi
I
In
1nl Ins
Appearance of polished surface
A non-porous anisotropic component of cokes made from high-rank coals which does not merge with mosaic components. Single-coloured particles. Particles with layered structure of contrasting colour.
Composed of elongated isochromatic areas often curved round pores. Size >20~m x >10~m. Size >20~ x >2~m. Size >3 x >l~m.
Composed of small rounded isochromatic areas. Mean size O.91~m. Mean size O.63~m. Xean size O.50~m.
An optically-isotropic component of cokes from low-rank coals. Fuses well to mosaic components.
Carbonaceous inerts are isotropic components identifiable by their woody structure, or if small by their unfused sharp edges. Mineral matter is included in this class. Size >50~m. Size <50~m.
TABLE 26. Measured PLK textural compositions of single-coal cokes.
Coal Fractional PLK textural composi ti on ( v/v )
1nl Ins Fb Fs Fg Xc Km Xi Is A .154 .084 .175 .312 .238 .028 .008 .002 0 B .177 .033 .243 .227 .227 .040 .047 .003 .003
C .108 .104 .008 .205 .482 .084 .004 . q06 0 D .070 .028 0 .015 .245 .480 .135 .010 .017 E .085 .053 0 .010 .036 .078 .564 .134 .040 F .098 .096 0 0 .012 .020 .349 .339 .086
TABLE 27. Measured FLM textural compositions of cokes prepared from blended coal charges.
Coke Fractional FLM textural composition number Inl Ins Fb Fs Fg Mc Mm Mf I
1 0.082 0.132 0.032 0.042 0.082 0.028 0.316 0.240 0.036 2 0.062 0.114 0.016 0.052 0.134 0.040 0.334 0.1960.052 3 0.106 0.084 0.008 0.018 0.206 0.032 0.348 0.1460.052 4 0.146 0.114 0.030 0.1120.176 0.022 0.156 0.1620.082 5 0.137 0.071 0.014 0.070 0.293 0.082 0.207 0.123 0.026 6 0.160 0.066 0.020 0.048 0.334 0.062 0.148 0.112 0.050 7 0.144 0.052 0.064 0.134 0.224 0.026 0.226 0.094 0.036 8 0.130 0.074 0.048 0.126 0.426 0.026 0.120 0.0300.020 9 0.196 0.090 0.026 0.186 0.456 0.024 0.012 0.0060.004
10 0.116 0.084 0.104 0.174 0.246 0.020 0.134 0.0920.030 11 0.144 0.078 0.046 0.162 0.400 0.040 0.094 0.0300.006 12 0.142 0.062 0.062 0.228 0.456 0.034 0.012 0.000 0.004 13 0.098 0:070 0.034 0.038 0.092 0.074 0.416 0.174 0.004 14 0.122 0.050 0.018 0.039 0.077 0.080 0.590 0.024 0.000 15 0.080 0.085 0.057 0.060 0.150 0.073 0.360 0.1300.005 16 0.1~4 0.060 0.010 0.058 0.180 0.159 0.359 0.0200.010 17 0.082 0.069 0.050 0.077 0.256 0.082 0.246 0.1200.018 18 0.110 0.068 0.032 0.114 0.194 0.130 0.316 0.0320.004 19 0.148 0.052 0.106 0.142 0.226 0.062 0.154 0.1020.008 20 0.136 0.078 0.066 0.114 0.222 0.150 0.204 0.030 0.000 21 0.097 0.050 0.176 0.133 0.187 0.060 0.253 0.037 0.007 22 0.157 0.087 0.200 0.123 0.180 0.023 0.200 0.027 0.003 23 0.093 0.073 0.148 0.089 0.126 0.070 0.315 0.0460.040 24 0.084 0.057 O. 138 0.084 O. 157 0.084 0.336 0.0440.016 25 0.136 0.097 0.071 0.070 0.087 0.070 0.416 0.0360.017 26 0.117 0.087 0.113 0.083 0'.100 0.043 0.404 0.0400.013 27 0.060 0.083 0.030 0.037 0.100 0.073 0.557 0.040 0.020 28 0.123 0.103 0.050 0.043 0.067 0.057 0.460 0.067 0.030 29 O. 130 0.067 0.087 0.090 0.437 0.087 0.093 0.006 0.003 30 0.160 0.086 0.102 0.126 0.442 0.060 0.015 0.0000.010 31 0.095 0.085 0.080 0.057 0.327 0.104 0.212 0.0300.010 32 0.176 0.100 0.084 0.110 0.403 0.084 0.043 0.0000.000 33 0.097 0.043 0.033 0.043 0.227 0.103 0.384 0.0570.013 34 O. 130 0.093 0.040 0.040 0.411 0.096 0.167 0.0160.007 35 0.097 0.106 0.017 0.0330.103 0.060 0.497 0.0600.027 36 0.097 0.083 0.000 0.0060.167 0.117 0.487 0.0400.003 37 0.126 0.054 0.074 0.152 0.320 0.170 0.096 0.0080.000 38 0.123 0.049 0.043 0.144 0.413 0.180 0.036 0.008 0.004 39 O. 154 0.074 0.059 0.060 0.298 0.208 0.143 0.004 0.000 40 0.094 0.048 0.054 0.130 0.396 0.230 0.044 0.004 0.000 41 0.061 0.054 0.032 0.106 0.252 0.292 0.190 0.011 0.002 42 0.102 0.068 0.034 0.070 0.396 0.248 0.080 0.0020.000 43 0.104 0.073 0.021 0.090 0.218 0.318 0.172 0.0040.000 44 O. 120 0.056 0.012 0.080 0.422 0.272 0.036 0.0040.002
TABLE.28 PLM textural compositions of cokes from blended coal charges calculated using method C.
Coke Fractional PLM textural composition number 1nl Ins Fb Fs Fg Mc Mm Mf I
1 0.109 0.094 0.036 0.064 0.058 0.022 0.279 0.270 0.068 2 0.106 0.096 0.019 0.069 0.120 0.032 0.251 0.244 0.062 3 0.102 0.099 0.003 0.075 0.183 0.043 0.223 0.218 0.055 4 0.118 0.092 0.064 0.114 0.094 0.023 0.225 0.216 0.055 5 0.111 0.096 0.034 0.124 0.206 0.042 0.174 0.169 0.042 6 0.104 0.101 0.005 0.132 0.315 0.061 0.126 0.124 0.031 7 0.127 0.090 0.092 0.164 0.131 0.024 0.170 0.162 0.041 8 0.117 0.097 0.050 0.179 0.292 0.052 0.097 0.094 0.023 9 0.107 0.104 0.008 0.193 0.454 0.080 0.025 0.026 0.005
10 0.137 0.088 0.120 0.215 0.167 0.026 0.114 0.107 0.027 11 0.129 0.093 0.088 0.226 0.289 0.047 0.060 0.056 0.013 12 0.121 0.098 0.056 0.236 0.411 0.068 0.005 0.005 0.000 13 0.102 0.075 0.029 0.057 0.060 0.047 0.385 0.193 0.051 14 0.094 0.057 0.022 0.049 0.062 0.072 0.492 0.117 0.035 15 0.112 0.077 0.059 0.109 0.097 0.043 0.308 0.154 0.041 16 0.106 0.063 0.053 0.103 0.098 0.063 0.393 0.093 0.028 17 0.124 0.079 0.091 O. 166 0.136 0.038 0.223 0.113 0.030 18 0.118 0.068 0.084 0.156 0.134 0.054 0.295 0.070 0.021 19 0.134 0.081 0.118 0.213 0.169 0.035 0.155 0.076 0.020 20 0.131 0.073 0.115 0.210 0.170 0.045 0.196 0.047 0.014 21 0.138 0.056 0.138 0.181 0.166 0.049 0.209 0.047 0.015 22 0.147 0.040 0.163 0.156 0.164 0.052 0.217 0.046 0.015 23 0.124 0.056 0.100 0.135 0.130 0.057 0.306 0.071 0.021 24 0.130 0.043 0.119 0.116 0.129 0.059 0.312 0.070 0.022 25 0.109 0.055 0.063 0.087 0.095 0.065 0.404 0.095 0.028 26 0.113 0.047 0.074 0.076 0.094 0.066 0.407 0.094 0.029 27 0.095 0.054 0.025 0.041 0.059 0.073 0.500 0.118 0.035 28 0.096 0.051 0.030 0.036 0.059 0.073 0.501 0.118 0.035 29 0.125 0.089 0.079 0.226 0.317 0.059 0.080 0.021 0.005 30 0.121 0.098 0.057 0.236 0.411 0.068 0.005 0.005 0.000 31 0.114 0.079 0.057 0.167 0.241 0.064 0.211 0.052 0.015 32 0.106 0.100 0.007 0.191 0.450 0.084 0.044 0.015 0.003 33 0.103 0.069 0.036 0.108 0.164 0.069 0.343 0.083 0.024 34 0.098 0.083 0.005 0.123 0.295 0.081 0.239 0.060 0.017 35 0.092 0.060 0.014 0.049 0.088 0.075 0.475 0.113 0.034 36 0.090 0.065 0.002 0.055 0.140 0.079 0.434 0.104 0.031 37 0.129 0.070 0.117 0.213 0.205 0.110 0.121 0.025 0.009 38 0.127 0.066 0.119 0.218 0.240 0.171 0.048 0.005 0.005 39 0.116 0.062 0.087 0.162 0.189 0.153 0.180 0.037 0.014 40 0.114 0.057 0.090 0.169 0.241 0.245 0.069 0.006 0.008 41 0.103 0.055 0.057 0.111 0.172 O. 196 0.238 0.049 0.019 42 0.100 0.048 0.062 0.120 0.243 0.320 0.090 0.007 0.011 43 0.089 0.047 0.027 0.059 0.156 0.239 0.297 0.061 0.024 44 0.086 0.039 0.033 0.071 0.244 0.394 0.111 0.008 0.014
TABLE.29 PLM textural compositions of cokes from blended coal charges calculated using method Y.
Coke Fractional PLM textural composition number 1nl Ins Fb Fs Fg Mc Mm Mf I
1 0.111 0.093 0.041 0.074 0.066 0.022 0.268 0.259 0.066 2 O. 107 0.096 0.022 0.078 0.132 0.034 0.240 0.233 0.059 3 0.102 0.099 0.003 0.081 0.198 0.045 0.213 0.207 0.052 4 0.121 0.091 0.071 0.128 0.105 0.023 0.209 0.201 0.051 5 0.113 0.096 0.038 0.134 0.219 0.044 0.161 0.156 0.039 6 0.105 0.101 0.005 0.138 0.329 0.063 0.116 0.114 0.028 7 0.130 0.089 0.100 0.179 0.142 0.025 0.153 0.146 0.037 8 0.119 0.096 0.053 0.187 0.301 0.053 0.087 0.084 0.020 9 0.107 0.104 0.008 0.194 0.457 0.081 0.022 0.024 0.005
10 0.139 0.087 0.127 0.227 0.177 0.026 0.101 0.094 0.023 11 O. 130 0.093 0.093 0.233 0.293 0.046 0.053 0.049 0.012 12 0.122 0.098 0;059 0.238 0.408 0.067 0.005 0.005 0.000 13 0.103 0.075 0.034 0.065 0.066 0.047 0.376 0.185 0.050 14 0.095 0.058 0.026 0.054 0.066 0.071 0.482 0.115 0.034 15 0.115 0.077 0.066 0.121 0.105 0.042 0.292 0.144 0.038 16 0.108 0.064 0.059 0.113 0.105 0.061 0.375 0.089 0.026 17 0.127 0.080. 0.098 0.179 0.145 0.038 0.205 0.102 0.027 18 0.121 0.069 0.091 0.168 0.141 0.052 0.274 0.065 0.019 19 0.136 0.081 0.124 0.224 0.177 0.035 0.139 0.067 0.018 20 0.133 0.075 0.121 0.220 0.176 0.043 0.178 0.042 0.012 21 O. 141 0.057 0.145 0.190 0.173 0.047 0.191 0.043 0.013 22 0.150 0.039 0.171 0.163 0.171 0.051 0.199 0.042 0.014 23 0.127 0.056 0.108 0.144 0.138 0.055 0.286 0.066 0.020 24 0.133 0.042 0.128 0.124 0.137 0.058 0.292 0.065 0.021 25 0.112 0.055 0.069 0.096 0.101 0.063 0.386 0.090 0.027 26 0.116 0.046 0.082 0.083 0.101 0.065 0.389 0.090 0.028 27 0.096 0.054 0.028 0.045 0.063 0.072 0.491 0.116 0.035 28 0.098 0.050 0.034 0.040 0.063 0.073 0.492 0.116 0.035 29 0.126 0.089 0.082 0.230 0.318 0.058 0.072 0.019 0.005 30 0.122 0.098 0.059 0.238 0.407 0.067 0.005 0.005 0.000 31 0.116 0.080 0.062 0.175 0.247 0.063 0.196 0.048 0.014 32 0.106 0.101 0.007 0.192 0.452 0.084 0.041 0.014 0.003 33 0.105 0.070 0.040 0.116 0.172 0.068 0.327 0.079 0.023 34 0.099 0.084 0.005 0.127 0.305 0.082 0.227 0.057 0.016 35 0.093 0.060 0.016 0.054 0.092 0.074 0 .. 466 0.111 0.033 36 0.091 0.066 0.002 0.059 0.147 0.079 0.424 0.102 0.030 37 0.131 0.071 0.122 0.223 0.209 0.102 0.110 0.023 0.009 38 0.130 0.068 0.124 0.227 0.240 0.158 0.044 0.004 0.005 39 0.118 0.064 0.093 0.173 0.193 0.144 0.167 0.035 0.013 40 0.117 0.059 0.096 0.180 0.241 0.230 0.065 0.006 0.008 41 0.105 0.056 0.063 0.120 0.176 0.189 0.226 0.047 0.018 42 0.102 0.050 0.067 0.130 0.242 0.305 0.086 0.007 0.010 43 0.091 0.048 0.031 0.065 0.158 0.236 0.289 0.059 0.023 44 0.088 0.040 0.037 0.078 0.244 0.384 0.108 0.008 0.013
TABLE 30. PLM textural compositions of cokes from blended coal charges calculated using method V.
Coke Fractional PLM textural composition number 1nl Ins Fb Fs Fg Mc Mm Mf I
1 0.112 0.093 0.043 0.077 0.068 0.022 0.265 0.256 0.065 2 0.107 0.096 0.023 0.079 0.133 0.034 0.238 0.231 0.058 3 0.102 0.099 0.003 0.082 0.199 0.045 0.212 0.207 0.052 4 0.122 0.091 0.074 0.132 0.108 0.023 0.205 0.196 0.050 5 0.113 0.096 0.039 0.136 0.220 0.044 0.159 0.154 0.038 6 0.105 0.101 0.005 0.139 0.330 0.063 0.115 0.114 0.028 7 0.131 0.089 0.1020.183 0.144 0.025 0.149 0.142 0.036 8 0.119 0.096 0.0550.189 0.301 0.053 0.086 0.082 0.020 9 0.107 0.104 0.008 0.194 0.457 0.081 0.022 0.023 0.005
10 0.139 0.087 0.1280.230 0.179 0.026 0.097 0.090 0.023 11 0.131 0.092 0.094 0.234 0.292 0.046 0.051 0.048 0.011 12 0.122 0.098 0.060 0.238 0.406, 0.p66 0.005 0.005 0.000 13 0.104 0.075 0.035 0.068 0.068 0 .. 046 0.370 0.184 0.049 14 0.096 0.058 0.027 0.057 0.068 0.070 0.477 0.113 0.034 15 0.116 0.077 0.068 0.125 0.108 0.042 0.284 0.141 0.038 16 0.110 0.064 0.062 0.118 0.108 0.060 0.366 0.087 0.026 17 0.128 0.080 0.101 0.183 0.148 0.037 0.197 0.099 0.026 18 0.122 0.070 0.094 0.173 0.145 0.051 0.264 0.063 0.018 19 0.137 0.081 0.126 0.228 0.179 0.034 0.133 0.065 0.017 20 0.134 0.075 0.123 0.224 0.179 0.043 0.170 0.040 0.012 21 0.142 0.057 0.148 0.194 0.175 0.047 O. 184 0.041 0.013 22 0.151 0.039 0.174 0.166 0.1730.051 0.193 0.040 0.013 23 0.128 0.056 0.112 0.149 0.141 0.054 0.277 0.064 0.019 24 0.135 0.042 0.131 0.127 0.139 0.057 0.284 0.063 0.020 25 0.113 0.055 0.072 0.100 0.'104 0.063 0.378 0.088 0.027 26 0.117 0.046 0.085 0.086 0.103 0.065 0.383 0.088 0.027 27 0.097 0.054 0.030 0.047 0.064 0.072 0.487 0.115 0.034 28 0.098 0.050 0.036 0.042 0.064 0.072 0.488 0.115 0.035 29 0.127 0.089 0.084 0.233 0.318 0.057 0.069 0.019 0.005 30 0.122 0.098 0.060 0.239 0.405 0.066 0.005 0.005 0.000 31 ' 0.117 0.080 0.064 0.179 0.250 0.062 0.189 0.047 0.013 32 0.107 0.101 0.007 0.192 0.453 0.084 0.040 0.014 0.003 33 0.106 0.071 0.0420.120 0.175 0.068 0.319 0.077 0.022 34 0.099 0.084 0.005 0.129 0.308 0.082 0.222 0.056 0.016 35 0.094 0.060 0.017 0.056 0.095 0.074 0.462 0.110 0.033 36 0.091 0.066 0.002 0.060 0.150 0.080 0.421 0.101 0.030 37 0.133 0.072 0.125 0.228 0.210 0.097 O. 105 0.022 0.008 38 0.132 0.069 0.128 0.233 0.240 0.148 0.042 0.004 0.005 39 0.120 0.065 0.097 0.180 0.194 0.137 0.161 0.033 0.013 40 0.119 0.060 0.101 0.187 0.241 0.218 0.061 0.005 0.007 41 0.107 0.057 0.066 0.127 0.177 0.181 0.221 0.046. 0.018 42 O. 105 0.051 0.072 0.137 0.242 0.294 0.083 0.007 0.010 43 0.092 0.049 0.033 0.069 0.158 0.229 0.287 0.059 0.023 44 0.089 0.041 0.040 0.083 0.243 0.376 0.106 0.008 0.013
TABLE 31. Coefficients obtained by applying the MLR<29) equation to
PLM textural data calculated using methods C, Y and V.
Textural Initial Coefficients in MLR<29) equation for :-
component C data Y data V data
Constant K -156.21 127.75 73.33
Flow:
broad Fb 181.08 -60.38 -221.71
striated Fs 116.13· -205.91 62.32
granular Fg 153.93 -131.36 -62.82
)[osaic:
coarse Mc 164.25 -117.44 -59.94
medium Mm 166.72 -123.23 -65.18
fine Mf 123.60 -147.12 54.13
Isotropic I 55.08 -226.29 -273.39
Inerts:
large 1nl 178.40 -130.67 36.05
small Ins 276.63 95.89 -447.26
Standard error of estimation:- 0.424 0.423 0.414
TABLE 32. Comparison of measured coke tensile strengths with strengths calculated from calculated PLM textural data using the MLR(29) equation.
Coke tensile strengths, MPa. Coke Measured Calculated using: number C data Y data V data
1 4.43 4.58 4.49 4.26 2 4.42 4.78 4.69 4.82 3 5.27 4.95 5.21 5.09 4 4.52 4.65 4.62 4.26 5 5.20 5.08 5.19 5.42 6 5.87 5.77 5.86 5.96 7 4.65 4.78 4.63 4.62 8 5.59 5.57 5.66 5.55 9 5.77 6.61 6.10 6.32
10 4.90 4.80 4.81 4.84 11 5.69 5.51 5.47 5.82 12 5.76 5.92 6.16 5.98 13 5.09 5.07 5.00 5.43 14 5.51 5.78 5.83 5.68 15 5.41 5.23 5.19 5.22 16 5.11 5.68 5.76 5.65 17 5.32 5.06 5.08 5.07 18 5.43 5.47 5.46 5.48 19 5.69 5.09 5.00 5.24 20 5.40 5.38 5.48 5.29 21 4.52 4.99 5.36 5.10 22 5.79 5.19 5.13 5.21 23 5.31 5.51 5.49 5.48 24 4.91 5.32 5.32 5.34 25 5.94 5.90 5.80 5.51 26 5.70 5.56 5.44 5.61 27 6.10 5.91 5.81 5.81 28 5.35 5.75 5.59 5.69 29 5.30 5.95 5.93 5.77 30 6.16 6.10 5.96 5.98 31 6.06 5.74 5.76 5.92 32 6.64 6.23 6.47 6.23 33 6.15 5.66 5.82 5.86 34 6.28 6.33 6.23 6.19 35 6.29 5.92 5.95 6.09 36 6.05 6.02 6.09 6.00 37 5.17 5.47 5.39 5.57 38 5.28 5.50 5.63 5.54 39 6.27 5.78 5.78 5.64 40 5.02 5.85 5.57 5.78 41 6.64 6.04 6.02 6.17 42 6.96 6.51 6.50 6.42 43 5.69 6.06 6.32 6.26 44 6.78 6.68 6.70 6.68
TABLE 33. Coefficients obtained by applying the TRANS<33> equation
to PLM textural data calculated using methods C, Y, and V.
Textural Initial Coefficients in Trans<33> equation for
component C data Y data V data
Flow:
broad Fb 2.1 2.4 2.5
striated Fs 5.9 5.6 5.9
granular Fg 8.5 8.5 8.3
Xosaic:
coarse Mc 7.6 7.7 7.9
medium Mm 7.5 7.6 7.6
fine Mf 2.9 2.4 2.3
Isotropic I 2.3 2.2 2.1
Inerts In 1.9 2.2 2.1
Standard error of estimation:- 0.380 0.377 0.377
TABLE 34. Comparison of measured coke tensile strengths with strengths calculated from calculated PLM textural data using the TRANS(33) equation.
Coke tensile strengths, MPa. Coke Measured Calculated using: number C data Y data V data
1 4.43 4.53 4.49 4.47
2 4.42 4.84 4.83 4.79
3 5.27 5.15 5.16 5.12
4 4.52 4.62 4.61 4.60
5 5.20 5.17 5.19 5.16
6 5.87 5.71 5.74 5.70
7 4.65 4.70 4.71 4.72
8 5.59 5.50 5.52 5.50
9 5.77 6.29 6.31 6.27
10 4.90 4.79 4.81 4.84
11 5.69 5.39 5.40 5.40
12 5.76 5.99 5.99 5.96
13 5.09 5.16 5.16 5.13
14 5.51 5.81 5.82 5.80
15 5.41 5.13 5.12 5.11
16 5.11 5.63 5.63 5.62
17 5.32 5.07 5.07 5.08
18 5.43 5.46 5.46 5.45
19 5.69 5.04 5.05 5.07
20 5.40 5.29 5.29 5.30
21 4.52 5.25 5.26 5.27
22 5.79 5:20 5.23 5.24
23 5.31 5.44 5.44 5.44
24 4.91 5.40 5.42 5.41
25 5.94 5.62 5.63 5.61
26 5.70 5.60 5.61 5.60
27 6.10 5.81 5.83 5.81
28 5.35 5.80 5.82 5.80
29 5.30 5.72 5.72 5.71
30 6.16 5.99 5.98 5.96
31 6.06 5.78 5.78 5.77
32 6.64 6.38 6.39 6.35
33 6.15 5.83 5.84 5.83
34 6.28 6.21 6.24 6.21
35 6.29 5.89 5.91 5.89
36 6.05 6.04 6.08 6.06
37 5.17 5.47 5.46 5.46
38 5.28 5.63 5.62 5.61
39 6.27 5.73 5.72 5.71
40 5.02 5.98 5.96 5.95
41 6.64 6.00 6.00 5.98
42 6.96 6.33 6.33 6.31
43 5.69 6.27 6.29 6.28
44 6.78 6.68 6.71 6.71
TABLE 35. Differences between measured PLH textural contents of cokes from blended-coal charges and those calculated using method Y.
Coke Measured minus calculated fractional textural contents number 1nl Ins Fb Fs Fg Mc Km Ht I
1 -.021 .011 .009 -.032 .016 .006 .058 -.019 -.030 2 -.007 .034 -.006 -.026 .002 .006 .094 -.037 -.007 3 -.004 .015 .005 -.063 .008 -.013 .135 -.061 .000 4 .025 .023 -.041 -.016 .071 -.001 -.053 -.039 .031 5 -.024 .025 .024 -.064 .074 .015 .046 -.033 -.013 6 .055 -.035 .015 -.090 .005 -.001 .032 -.002 .022 7 .014 -.037 -.036 -.045 .082 .001 .073 -.052 -.001 8 .011 -.022 -.005 -.061 .126 -.027 .033 -.054 .000 9 .089 -.013 .018 -.008 -.001 -.057 -.010 -.018 -.001
10 -.023 -.003 .023 .053 .069 -.006 .033 -.002 .007 11 .014 -.015 -.047 -.070 .107 -.006 .041 -.019 -.006 12 .020 -.036 .003 -.009 .049 -.033 .007 -.005 .004 13 -.005 -.005 .000 -.027 .026 .027 .040 -.011 -.046 14 .027 -.008 -.008 -.015 .011 .009 .108 -.091 -.034 15 -.035 .008 -.009 -.061 .045 .031 .068 -.014 -.033 16 .036 -.004 -.049 -.055 .075 .098 -.016 -.069 -.016 17 -.027 -.040 .022 -.159 .095 .002 .085 .048 -.027 18 -.011 -.001 -.059 -.054 .053 .078 .042 -.033 -.015 19 .012 -.029 -.018 -.082 .049 .027 .015 .035 -.010 20 .003 .003 -.055 -.106 .046 .107 .026 -.012 -.012 21 -.044 -.007 .031 -.057 .015 .013 .061 -.006 -.006 22 .007 .048 .029 -.040 .010 -.028 .000 -.015 -.011 23 -.034 .017 .040 -.055 -.012 .015 .029 -.020 .020 24 -.049 .015 .010 -.040 .021 .026 .044 -.021 -.005 25 .024 .042 .002 -.026 -.014 .007 .030 -.054 -.010 26 .001 .041 .031 .000 .000 -.022 .014 -.050 -.015 27 -.036 .029 .002 -.008 .037 .001 .066 -.076 -.015 28 .025 .053 .016 .003 .004 -.016 -.032 -.049 -.005 29 .004 -.022 .005 -.140 .119 .029 .021 .013 -.002 30 .038 -.013 .043 -.112 ;035 ~. 007 .010 -.005' .010 31 -.021 .005 .019 -.118 .080 .041 .016 -.018 -.004 32 .070 -.001 .077 -.082 -.049 .000 .002 -.014 -.003 33 -,008 -.027 -,007 -.073 .055 .035 .057 -.022 -.010 34 .031 .009 .035 -.087 .106 .014 -.059 -.041 -.009 35 .004 .046 .001 -.021 .011 -.014 .031 -.051 -.006 36 .006 .017 -.002 -.053 .020 .038 .063 -.062 -.027 37 ,005 -.017 -.048 -.071 .111 .068 -.014 -.015 -.009 38 -.007 -.019 -.081 -.083 .173 .022 -.008 .004 -.001 39 .036 .010 -.034 -.113 .105 .064 -.024 -,031 -.013 40 -.023 -.011 -.042 -.050 .155 .000 -.021 -.002 -.008 41 -.044 -.002 -.031 -.014 .076 .103 -.036 -.036 -.016 42 .000 .018 -.033 -.060 .154 -.057 -.006 -.005 -.010 43 .013 .025 -.010 .025 .060 .082 -.117 -.055 -.023 44 .031 .016 -.025 .002 .176 -.113 -.072 -.004 -.011
TABLE 36. Differences between measured PLM textural contents of cokes from blended-coal charges and those calculated using method Y, averaged for cokes made using the same coals.
Coals Average measured minus calculated textural contents Ins 1nl Fb Fs Fg Mc Mm )If I
Two-component blends:
A-C .029 -.024 .023 -.060 .042 -.020 .009 -.005 .007
A-D .000 -.001 -.045 -.048 .165 -.037 -.027 -.002 -.007
A-E .014 -.002 -.043 -.057 .046 .073 .040 -.051 -.019
A-F -.007 .009 -.027 -.036 .060 .000 .028 -.028 .002
B-E -.004 .039 .022 -.019 .009 -.010 .007 -.034 -.009
C-E .036 .008 .037 -.074 .026 .017 .002 -.039 -.013
C-F .049 -.021 .013 -.054 .004 -.024 .052 -.027 .007
Three-component blends:
A-B-E -.022 .020 .019 -.036 .007 .009 .047 -.039 -.003
A-C-E -.005 .001 .005 -.088 .066 .023 .031 -.026 -.005
A-C-F .014 -.024 -.020 -.055 .077 -.003 .054 -.036 -.006
A-D-E .000 .004 -.031 -.043 .088 .079 .048 -.034 -.015
A-E-F -.014 -.016 -.001 -.082 .054 .022 .052 .015 -.029
TABLE 37. Notional PLM textural compositions of single-coal cokes calculated from measured data for cokes from ten three-component coal blends.
Coal Notional PLM textural composition
1nl Ins Fb Fs Fg Mc Km Mf Is
A .131 .077 .120 .147 .454 .060 .011 0 0
B .112 .043 .353 .212 .086 .004 .156 .017 .017
C .151 .053 .027 .111 .520 .076 .049 0 .013
D .127 .064 0 .049 .265 .496 0 0 0
E .094 .074 0 .016 .032 .111 .587 .075 .011
F .093 .075 0 .028 .042 .065 .384 .285 .028
TABLE 38. PLM textural compositions of cokes from blended-coal charges calculated, using method C, from notional textural data for single-coal cokes.
Coke Fractional PLM textural composition number 1nl Ins Fb Fs Fg Mc Mm Mf I
1 0.i01 0.075 0.025 0.052 0.1260.064 0.308 0.227 0.022 2 0.107 0.071 0.017 0.055 0.171 0.066 0.285 0.204 0.022 3 0.114 0.067 0.010 0.058 0.216 0.069 0.262 0.181 0.023 4 0.107 0.076 0.044 0.071 0.192 0.063 0.248 0.181 0.018 5 0.119 0.068 0.031 0.077 0.273 0.068 0.207 0.140 0.018 6 0.130 0.061 0.017 0.082 0.350 0.072 0.168 0.101 0.018 7 0.113 0.076 0.063 0.091 0,259 0.062 0.188 0.135 0.013 8 0.130 0.065 0.044 0.098 0.375 0.069 0.128 0.076 0.014 9 0.148 0.054 0.025 0.106 0.491 0.075 0.069 0.017 0.014
10 0.119 0.076 0.083 0.110 0.325 0.062 0.127 0.089 0.009 11 0.132 0.068 0.068 0.116 0.413 0.066 0.083 0.044 0.009 12 O. 145 0.060 0.054 0.122 0.501 0.071 0.038 0.000 0.009 13 O. 100 0.075 0.020 0.043 0.107 0.084 0.409 0.145 0.016 14 0.099 0.074 0.015 0.033 0.086 0.104 0.513 0.065 0.010 15 0.106 0.075 0.041 0.064 0.178 0.079 0.328 0.116 0.013 16 0.105 0.075 0.037 0.056 0.162 0.095 0.410 0.052 0.008 17 0.113 0.076 0.063 0.087 0.255 0.074 0.238 0.085 0.009 18 0.112 0.075 0.058 0.079 0.236 0.086 0.308 0.039 0.006 19 0.119 0.076 0.081 0.106 0.319 0.069 0.166 0.057 0.006 20 0.1180.076 0.079 0.103 0.311 0.077 0.206 0.025 0.004 21 0.1120.065 0.158 0.125 0.188 0.058 0.254 0.031 0.009 22 0.1060.053 0.237 0.148 0.068 0.039 0.297 0.036 0.015 23 0.107 0.067 0.114 0.095 0.146 0.073 0.345 0.043 0.010 24 0.1030.059 0.172 0.112 0.058 0.059 0.377 0.047 0.014 25 0.102 0.070 0.071 0.065 0.102 0.087 0.437 0.055 0.010 26 0.099 0.065 0.107 0.076 0.048 0.078 0.456 0.057 0.013 27 0.097 0.072 0.028 0.035 0.060 0.102 0.527 0.067 0.011 28 0.096 0.070 0.043 0.040 0.039 0.098 0.534 0.068 0.012 29 ". O. 135 0.066 0.064 0.114 0.427 0.074 0.104 0.010 0.007 30 0.145 0.060 0.054 0.122 0.501 0.071 0.038 0.000 0.009 31 0.124 0.068 0.046 0.087 0.320 O. 084 0.235 0.028 0.008 32 O. 147 0.055 0.025 0.104 0.485 0.079 0.088 0.005 0.013 33 0.113 0.070 0.029 0.061 0.212 0.094 0.366 0.045 0.009 34 O. 127 0.062 0.016 0.071 0.316 0.091 0.274 0.031 0.012 35 0.101 0.073 0.012 0.034 0.104 0.104 0.498 0.063 0.010 36 0.107 0.069 0.006 0.038 0.146 0.103 0.462 0.058 0.011 37 0.124 0.074 0.081 0.109 0.353 0.139 0.105 0.013 0.002 38 0.130 0.073 0.082 0.116 0.394 0.198 0.008 0.000 0.000 39 0.121 0.073 0.060 0.090 0.301 0.181 0.153 0.019 0.003 40 0.129 0.071 0.062 0.100 0.363 0.269 0.006 0.000 0.000 41 0.117 0.072 0.039 0.070 0.248 0.223 0.202 0.025 0.004 42 0.128 0.069 0.043 0.084 0.332 0.341 0.004 0.000 0.000 43 0.114 0.070 0.019 0.050 0.196 0.265 0.250 0.032 0.005 44 0.128 0.066 0.023 0.068 0.301 0.413 0.002 0.000 0.000
------~
TABLE 39. Coefficients obtained by applying the TRANS(33) equation to PLM textural data calculated from national textural data far single-coal cakes.
Textural Ini tial Coefficients in TRANS(33) equation component far C data
Flaw:
broad Fb 2.7
striated Fs 2.8
granular Fg 5.3
l!osaic:
coarse Mc 7.2
medium Mm 8.1
fine Kf 6.9
Isotropic I 1.7
Inerts In 1.7
Standard error of estimation:- 0.435
TABLE 40. Comparison of measured coke tensile strengths with those obtained, using the TRANS(33) equation, from PLM textural data calculated from notional single-coal coke data using method Y.
Coke tensile strengths, MPa. Coke Measured Calculated number
1 2 3 4 5 6 7 8 9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44
4.43 4.42 5.27 4.52 5.20 5.87 4.65 5.59 5.77 4.90 5.69 5.76 5.09 5.51 5.41 5.11 5.32 5.43 5.69 5.40 4.52 5.79 ·5.31 4.91 5.94 5.70 6.10 5.35 5.30 6.16 6.06 6.64 6.15 6.28 6.29 6.05 5.17 5.28 6.27 5.02 6.64 6.96 5.69 6.78
4.80 4.95 5.10 4.94 5.21 5.47 5.08 5.47 6.86 5.23 5.53 5.81 5.30 5.82 5.35 5.76 5.38 5.69 5.43 5.63 5.23 4.83 5.40 5.11 5.58 5.40 5.75 5.68 5.74 5.81 5.77 5.93 5.81 5.91 5.84 5.88 5.76 5.88 5.89 6.02 6.03 6.28 6.16 6.47
TABLE 41. Coefficients obtained by applying the IHTER(31) equation to measured PLM textural data.
1nl Ins Fb Fs Fg Mc Mm )If I
1nl 0 0 3.5 3.5 3.5 3.5 3.5 3.5 3.5
Ins 0 0 5.0 5.0 5.0 5.0 5.0 5.0 5.0
Fb 3.5 5.0 4.5 3.5 3.5 3.5 3.5 3.5 4.5
Fs 3.5 5.0 3.5 3.0 9.5 4.5 7.5 3.5 3.5
Fg 3.5 5.9 3.5 9.5 8.5 8.5 8.0 6.5 3.5
)lc 3.5 5.0 3.5 4.5 8.5 5.0 6.5 4.0 3.5
Km 3.5 5.0 3.5 7.5 8.0 6.5 8.0 3.5 3.5
)If 3.5 5.0 3.5 3.5 6.5 4.0 3.5 3.5 3.5
I 3.5 5.0 4.5 3.5 3.5 3.5 3.5 3.5 4.5
Standard error of estimation = 0.548 MPa.
TABLE 42. Values of textural strength terms obtained by applying the TRANS<33> equation to measured PLM textural data.
Textural component
Flow:
broad
striated
granular
Mosaic:
coa.rse
medium
fine
Isotropic
Inerts
Standard error
Ini tial
Fb
Fs
Fg
Mc
Mm
Kt
I
In
of estimation:-
Coefficients in TRANS<33> equation.
3.4
4.1
8.0
6.2
6.7
1.9
1.9
3.4
0.474
TABLE 43. Comparison of measured coke tensile strengths with those calculated, from measured PLM textural data, using the INTER(31) and TRANS(33) equations.
Coke tensile strengths, KPa. Coke Measured Calculated using: number INTER(3D TRANS(33)
1 4.43 4.65 4.48 2 4.42 5.20 4.89 3 5.27 5.40 5.30 4 4.52 4.65 4.50 5 5.20 5.80 5.56 6 5.87 5.33 5.39 7 4.65 5.26 5.15 8 5.59 5.94 5.84 9 5.77 5.65 5.72
10 4.90 5.16 4.97 11 5.69 5.85 5.72 12 5.76 6.01 5.79 13 5.09 5.29 5.16 14 5.51 6.12 5.92 15 5.41 5.49 5.32 16 5.11 5.82 5.85 17 5.32 5.70 5.47 18 5.43 5.89 5.72 19 5.69 5.09 5.06 20 5.40 5.48 5.55 21 4.52 5.30 5.29 22 5.79 4.76 4.99 23 5.31 5.14 5.15 24 4.91 5.46 5.43 25 5.94 5.26 5.34 26 5.70 5.27 5 .. 29 27 6.10 6.16 5.84 28 5.35 5.25 5.27 29 5.30 5.90 6.01 30 6.16 5.51 5.73 31 6.06 5.85 5.87 32 6.64 5.39 5.71 33 6.15 6.08 5.92 34 6.28 6.10 5.94 35 6.29 5.70 5.57 36 6.05 6.17 6.04 37 5.17 5.80 5.76 38 5.28 6.16 6.01 39 6.27 5.59 5.86 40 5.02 6.28 6.10 41 6.64 6.26 6.06 42 6.96 6.28 6.23 43 5.69 5.87 5.92 44 6.78 6.39 6.28
Table 44. Values of textural strength terms obtained by applying the POROSITY(47) equation to measured textural data.
Component Ini tial Strength terms for: b=2.6 b=2.8
Flow:
broad Fb 12.0 15.0
striated Fs 12.0 15.0
granular Fg 33.5 37.5
Mosaic:
coarse Mc 23.5 26.5
medium Mm 31. 5 35.5
fine JoIf 7.5 8.5
Isotropic I 6.5 8.5
Inerts In 18.0 18.0
Standard error
of estimation: .463 .476
Table 45. Comparison of measured coke tensile strengths with those calculated using the POROSITY<47> equation.
Coke Apparent Fractional Tensile strength, JoIPa number density porosity Keasured Calculated
kg/nf' h=2.6 h=2.8 1 780 .589 4.43 4.35 4.29 2 800 .579 4.42 4.83 4.79 3 792 .583 5.27 5.22 5.16 4 867 .544 4.52 4.73 4.71 5 823 .567 5.20 5.52 5.48 6 835 .561 5.87 5.44 5.39 7 845 .555 4.65 5.21 5.22 8 853 .551 5.59 5.91 5.91 9 827 .565 5.77 5.52 5.47
10 868 .543 4.90 5.06 5.10 11 861 .547 5.69 5.80 5.80 12 830 .563 5.76 5.46 5.46 13 760 .600 5.09 4.86 4.81 14 774 .593. 5.51 5.78 5.73 15 810 .574 5.41 5.28 5.26 16 791 .584 5.11 5.65 5.59 17 849 .553 5.32 5.58 5.59 18 856 .550 5.43 5.93 5.93 19 897 .528 5.69 5.39 5.44 20 863 .546 5.40 5.72 5.72 21 838 .559 4.52 5.22 5.28 22 897 .528 5.79 5.41 5.46 23 857 .549 5.31 5.33 5.38 24 837 .560 4.91 5.48 5.52 25 817 .570 5.94 5.44 5.40 26 831 .563 5.70 5.44 5.44 27 805 .576 6.10 5.90 5.88 28 793 .583 5.35 5.27 5.21 29 867 .544 5.30 6.17 6.17 30 830 .563 6.16 5.53 5.50 31 846 .555 6.06 5.97 5.96 32 837 .560 6.64 5.63 5.58 33 806 .576 6.15 5.83 5.81 34 813 .572 6.28 5.97 5.90 35 800 .579 6.29 5.64 5.58 36 807 .575 6.05 6.12 6.06 37 878 .538 5.17 5.87 5.90 38 863 .546 5.28 5.96 5.97 39 884 .535 6.27 6.21 6.19 40 873 .541 5.02 6.10 .6.13 41 859 .548 6.64 6.04 6.08 42 863 .546 6.96 6.28 6.28 43 842 .557 5.69 5.82 5.80 44 854 .551 6.78 6.21 6.19
Table 46. Comparison of measured tensile strengths with those calculated using the ADDTS(48) equation from measured tensile strengths of single-coal cokes.
Coke Tensile strengths, MPa. l!easured Calculated using method:
C y V 1 4.43 4.52 4.54 4.54 2 4.42 4.80 4.84 4.85 3 5.27 5.09 5.15 5.15 4 4.52 4.60 4.63 4.63 5 5.20 5.11 5.15 5.15 6 5.87 5.61 5.66 5.67 7 4.64 4.68 4.71 4.71 8 5.59 5.41 5.44 5.44 9 5.77 6.15 6.16 6.16
10 4.90 4.76 4.78 4.79 11 5.69 5.32 5.32 5.31 12 5.76 5.87 5.85 5.84 13 5.09 5.13 5.14 5.13 14 5.51 5.74 5.72 5.71 15 5.41 5.09 5.09 5.08 16 5.11 5.57 5.54 5.53 17 5.32 5.03 5.03 5.02 18 5.43 5.41 5.37 5.35 19 5.69 5.00 5.00 4.99 20 5.40 5.24 5.21 5.19 21 4.52 5.64 5.63 5.62 22 5.79 6.03 6.04 6.05 23 5.31 '5.70 5.69 5.68 24 4.91 5.99 6.00 6.00 25 5.94 5.76 5.75 5.74 26 5.70 5.94 5.95 5.95 27 6.10 5.82 5.81 5.81 28 5.35 5.89 5.90 5.90 29 5.30 5.63 5.61 5.59 30 6.16 5.87 5.85 5.84 31 6.06 5.69 5.67 5.66 32 6.64 6.23 6.23 6.23 33 6.15 5.75 5.74 5.73 34 6.28 6.09 6.10 6.10 35 6.29 5.82 5.81 5.81 36 6.05 5.95 5.96 5.96 37 5.17 5.35 5.31 5.29 38 5.28 5.45 5.41 5.37 39 6.27 5.58 5.53 5.50 40 5.02 5.73 5.67 5.63 41 6.64 5.80 5.76 5.76 42 6.96 6.01 5.96 5.96 43 5.69 6.03 6.01 5.98 44 6.78 6.29 6.25 6.22
Table 47. Comparison of measured and notional tensile strengths of single-coal cokes.
Coal Coke tensile strengths, KPa. Measured Calculated using method:
C y V
A 4.92 5.03 5.07 5.09
B 6.12 5.45 5.41 5.42
C 6.26 6.42 6.39 6.41
D 6.66 7.03 7.09 7.11
E 5.86 5.83 5.85 5.86
F 4.43 4.38 4.29 4.26
Table 48. Comparison of measured tensile strengths with those calculated using the ADDTS(48) equation from notional tensile strengths of single-coal cokes.
Coke Tensile strengths, HPa. Heasured Calculated using method:
C y V 1 4.43 4.51 4.48 4.47 2 4.42 4.82 4.80 4.79 3 5.27 5.12 5.13 5.12 4 4.52 4.62 4.61 4.61 5 5.20 5.16 5.18 5.17 6 5.87 5.70 5.72 5.72 7 4.65 4.72 4.74 4.75 8 5.59 5.51 5.52 5.52 9 5.77 6.30 6.30 6.30
10 4.90 4.82 4.86 4.87 11 5.69 5.42 5.45 5.43 12 5.76 6.01 6.00 6.00 13 5.09 5.12 5.11 5.10 14 5.51 5.73 5.74 5.74 15 5.41 5.10 5.10 5.10 16 5.11 5.58 5.58 5.59 17 5.32 5.07 5.08 5.09 18 5.43 5.44 5.44 5.44 19 5.69 5.07 5.09 5.09 20 5.40 5.30 5.31 5.31 21 4.52 5.44 5.43 5.43 22 5.79 5.57 5.54 5.54 23 5.31 '5.55 5.53 5.54 24 4.91 5.64 5.62 5.62 25 5.94 5.65 5.65 5.65 26 5.70 5.71 5.70 5.71 27 6.10 5.76 5.77 5.77 28 5.35 5.78 5.79 5.80 29 5.30 5.74 5.74 5.73 30 6.16 6.01 6.00 6.00 31 6.06 5.77 5.77 5.77 32 6.64 6.38 6.37 6.37 33 6.15 5.79 5.79 5.80 34 6.28 6.17 6.19 6.20 35 6.29 5.81 5.83 5.83 36 6.05 5.97 5.99 6.00 37 5.17 5.49 5.49 5.48 38 5.28 5.66 5.65 5.63 39 6.27 5.73 5.72 5.70 40 5.02 5.99 5'.97 5.94 41 6.64 5.97 5.97 5.95 42 6.96 6.32 6.31 6.28 43 5.69 6.21 6.23 6.21 44 6.78 6.65 6.66 6.64
CONSTANT TEMP.
SLAG 1600
METAL 1500
TEMPERATURE. OC
~~~i!==~ ~ COKE' SLITS'
METAL + COKE
FUSED SLAG + Fe LAYERS
FIG 1. Diagrams of a blast-furnace showing temperature
distribution and various zones.
6 8 6 4
I
, ......... ' .... / - / .-:., ;-b ~.---:> ,
'H , ~ ..... , a •... a 5 .--c
~ :.:
+'
" <JJ +'
4
" 0 0
" <JJ bO 0
3 ... -c >. ~
90 80
Carbon content, wtZ dmmf
FIG 2. Simplified version of Seyler's coal chart.
401 402
G8 r-
301a 301b 501 502
G6
2 G4 - 0
QJ P. 4 >. ..., 601 602 QJ
G2 .!4 0' 0
bel
'" .... G 0>4
I >.
2 I-0
;-3
701 702 (1j I..,
'" E 2 -0 2 801 802
c '2
~ 0_ 1
0 ~. 901 902
101 102 ,1 a
A
1 0 20 30
Volatile matter content, wt7. dmmf
FIG 3. N.C.B. coal rank classification system.
400 40
• I I /-r • •
t 0 • I I
I I 300 30 ~
E' ~ ~ ,....
0- , ~
E > 0
3- ~ • ~ .. N
,!:! ';; 200
E ~ - 20 u .. 0 .........
~ ~ • • ~ 0 • ~
0.. .. 0. ° ~ . ' 0..
" ° '0 ..(L. Ji> 0 0.. - --1:)'--U-- '0 .. ::i: " 100 10 ~
0 .. .. .0
~ E ~
J z
0 0
I 150 I 1500
-I r. , c -10 E
;.. "'" 100 -~ 1000 B ~ r-, ,.. c '--' ~
" 'C E 0 '5 \
6 '" ';; 0 - ......... ~ '" . .2 ~
\ E
6 50 .!! 500 4 '--' .. ~ ~ ~ ~
\ £
\!) 2 ~
"" \ '" ';;;
0 0 0 ~ " ,
\. \ .J
300 400 500 600 700
Temporaturo (DC)
FIG 4. Relationship between para structure development and cokine
characteristics.
., ... cd ... IOU ::> •
!l 550 x '" cd 0 IO~
..... cd 500 o.!ll .,:;:! ...... 450 ::> cd ... .-< cd 0 ... > ., ., 400 p.,,,,, 10 0) .....
!-< 0
500 ... '" o ..... ... cd ... .s 300 .....
""
100
10 p.,
"" to 0
.-<
>. 5 ... .....
"" ..... 4 ::> .... ....
... 3 ., .-< 0) (J) 0) ..... 2
(!)
10 ::>
1 10 ..... X cd :.:
a
b
c
94
, ,
Carbon
, , , , • • ,
'~-----; , ,
86 82
content, wt%.
25
20
15
10
5
80
'" o ..... ... cd (J) ....
.-< .... "'u .s. 0 0 >0 0).-< "" ...... ... 10 .... ::> :t 10 _
..... 0) x .... cd cd :.: ...
FIG 5. Devolatilisation of coals varies smoothly with rank (a) but
the dilatation (b) and fluidity (c) exhibited by coals
attain maximum values in the middle of the rank range.
Movczmcznt of •
Plastic layczr
/~/
Tczmpcroturcz •
Comprczssion
o < .. ;,.
~
FIG 6. Variation of stresses in coke and semicoke between oven
wall and plastic layer.
w
1
a D
, ~~~ b , ,
I , , I
O'TI O'T O'cl O'T
I , , , , , , -. 6 4 2 0 2 4 6
c
FIG 7. Diametral compressive load, W, (a) generates tensile (O'T)
and compressive (O'c) stresses along the loaded
diameter (b) which result in tensile fracture (c).
y
\ \ \
\ \
Z 1\ , ,
FIG 8. Representation of the uniaxial graphite crystal, the optical
axis lying in the Z-direction.
a
Vibration <:~---------------------------------->
direction
b
, . , ~ ~ , ~ , , , , , , , ~ , , • , , , ,
~ , , , , , , , , , , • , ,
• • • • A A
B
~ • \. • • , \, , ... \, ... , , , ... ... , , , , ... , , \ , , ... , • • • , • , , , , , ,- I , , ~ ,-,
~ ,- ,- ~
c
B
, ,- ,- ,-, , ,- ,
FIG 9. Shading effects obtained when polished carbon or graphite
surfaces are viewed under polarized light with crossed
polars, (a) variation of shading depending on orientation
of basal layer edges to vibration direction, (b) mosaic
effect observed with randomly aligned small crystallites,
and (c) origin of 'extinction contours in folded structure.
a
Vibration
direction
" " /.
b
A A c
FIG 10. Corresponding tinted effects when a A-retarder plate is inserted into the reflected light beam.
Mesophase sphere
Trace of lamellae direction
Section through a . mesophase sphere
Pole
----
-----'--------_.. ....- ....
Pole
Pole
c - axis
Edge of disk of sphere
FIG 11. Alignment of lamellar molecules in· a mesophase sphere.
.. ..... 0 :>
I/) .., <l Q)
<l 0 0.. a 0 0
'H 0
<l 0 .... .., .... 0 0.. 0 ....
p.,
80
Isotropic Fi ne
60
Flow
40 flow
20
0·7 o·g
Mean maximum vitr!nite reflectance, 7.
FIG 12. Variation of composition of vitrain cokes with mean
maximum vitrinite reflectance.
.. ..... o > 75
50
25
I
\/ I .\ I , I
-I \
\/ 400
\ . \ . , ,,'\
I ,
: .
.. ,: ;\
... \ \
450
." '.
\ , ,
'.
500
Temperature, 'C
.... " ........ . ......
550
FIG 13. Variation of the proportion of anisotropic components
with heat-treatment temperature during the carbonization
of a medium volatile vitrain (- -isotropic,-·- fine
mosaic, - - _ medium mosaic, ....... coarse mosaic,
granular-flow ).
, .
a 10000 -0..
"0
>. - -+' ..... 'd
1000 ..... ;:l .....
'H r-
~ ,....
'" ..... '" Ul
'" 100 ..... I-(!>
a 0 a ..... x Ol 10 =-: I-
r-
204 301 301 401 501 602
N .C.B. class of coal
FIG 14. Variation of Gieseler fluidity with N.C.B. class of coal.
9
15-
6.0 _-20
;--- - __ -21
- 13 >< -W· ;:;.--"d I'l 4.0
....- H-__ .... ;:;.--.<:I 1::, I'l W ... 7--"" 00 2.0 5 .......
3"""'----.0
Vitrinite class
0 60 40 20 0
Inert content, wtt
FIG 15. Variation" of strength index with inert content for
vitrinites in reflectance classes 3 to 21.
0 .... +' Id .. (/) +' .. w <l .... "-(/) W
" .... +' 0 Id w
c>:
25 96.2"
20 95.2
15 \ 93.8
\ 10 \ - 90.9
\ \
83.3
0 3 5 7 9 11 13 15 IT 19 21
" Vitrinite classes
FIG 16. Optimum reactives/inerts ratio of vitrinites in
reflectance classes 3 to 21.
.. +' :.
~ " .... +' 0 Id w
c>:
Stability factor ..
\ \ \ ,
\ \ \ 65 7.0 \ \ / \ / \ \ \ / \ \ \ /
/ , \ \ \ .- 60 \ \ \ .-6.0 \ .- .-
\ \ \ / / \ \ \ \ / \ \ \ \ \ \ /50 :< \ \ \
Q) 5.0 \ \ "d \
/ <l \ /40 .... \ \ /
.<I
~ 4.0 .30' <l
" ;' /
Q) " ,20 I-< " " ...,
" ,," 0/ Cl) 3.0
" "
2.0 L--'-----::::::~===::::;;~----__Jo' I 10.0 5.0 1.0 0.5 .
Composition-balance index
FIG 17. Curves showing the relationship between strength index,
composition balance index and coke stability factor.
"/.,R,..lmod
1·25
80 I-OS
70 0-95
65
0-65 -60
M Q) 55 '1j ~ .....
0 <l' 50 0-75 -:.: I'l :J 45 0 ..... :.:
40 -
35
30 -
25 10 15 20 25 30 35
Coal inert content, wt:l.
FIG 18. Dependence of coke Kicum K40 index on reflectance and
inert content of coal carbonized.
>: 15 QI 'd
'" ..... 0 10
"" :E:
0 ..... 5
'" 0 ..... ..... 0 0 QI ... ... 0
CJ -5 0 10 20 30 40 50
Total inerts <_O.12mm, wt%
2
>: QI 'd 0
'" ..... 0 ... :E: -2
0 ..... i:: -4 -0 ..... ..... 0 QI
" -6
" 0 CJ
-8 0 10 20 30 40 50
Total inerts >3mm, wu.
FIG 19. Correction to Kicum K40 index for size of inerts,
(a) >3mm, and (b) <0.12 mm.
14
10
.•
I 11 13 12
Coal 9 Purge gas inlet
2 Coal retort tube 10 Gas outlet
3 Silica brick 11 Tar trap
4 Silica rods 12 Track
5 Spring 13 Drive motor
6 Clamp 14 Furnace
7 0- rings 15 Thermocouples
8 Flange
FIG 20. Diagrammatic illustration of small pilot oven.
Specimen
CO~ Inlet
High frequency coil
! ------=-.;----
Position of discharge
FIG 21. Discharge apparatus used for etching.
Specimen support
Reaction vessel
To rotary pump
a
b
FIG. 22 Blectron micrographs showing (a) etched and (b) fractured coke surfaces at low IDII8nification. pores being at P.
0.6
0.3
O!-
0
01-0.6
0.30
0.60
0.30
0.60
0.30
0.60
0.30
0.60
0.30
t-
I-
I-
-
-
t-
.... L. .. c
.... L. ... c
bI a. >-....
u III ~ e 0
ID u..
r I I
I I
I r
I
I
I
L.
.2 .. .... E 0 e u.. -l
~ Mozaics 0 .... ,
r \ .~
bI E bI a.. V1 V1 ::> 0 L. L. e e " bI
t... .... 0 0 .. c 0 U u 2 u.. '"
I
-
I
.. .... e
" .. ... c
E bI E -'" ::> L- L. ... e " ""
>-.... L-0 "" c c ..
U 2 u.. > \ J •
Granular
FIG 23. Variation of textural composition with rank of
coal carbonized.
204
301a
301b
401
501
602
.>< c e 0 L- C
... 0" o 0 U u
a
b
c
FIG. 24 Electron micrographs showing the flat textural component in (a) etched and (b and c) fractured coke surfaces.
a
b
c
FIG . 25 Electron micrographs showing the lamellar textural component in (a) etched and (b and c) fractured coke surfaces .
b
c
FIG. 26 Electron micrographs showing the intermediate textural component in (a) etched and (b and c) fractured coke surfaces .
a
b
c
FIG. 27 Electron micrographs showing the granular textural component in (a) etched and (b and c) fractured coke surfaces.
a
b
FIG. 28 Electron micrographs showing large carbonaceous inerts in Ga) etched and (b) fractured cake surfaces.
FIG. 29 Electron micrograph showing splayed and folded lamellae in an etched pitch-coke surface.
a
... ~ . ,..,. --
-0.-
b
-"
c ~-
c
~~- .~
Gc·
--
/r'
"/ • .,/'" f
./ \ , , ,
FIG. 30 Electron micrographs showing microcrac)ts in (a) lamellar, (b') medium granular and (c) fine granular coke carbon.
I
QOIL )
FIG. 31 Electron micrograph showing microcrack traversing pore wall to carbonaceous inert particle.
.I,~ .. :.J ,~
" I /7.. -... - .*
O~" .
a
,~ '.
< lOO)lm >
b
< lOO)Jm >
FIG. 32 Electron micrographs showing (a) extended microcrack and (b) fracture crack networks.
· · . · . •
--· . · . 16 26
· . · . . ... . · • . · .
c: 29 30 0 .. u
"" · . L. . . · . "0
C7' c: · . "0 · · 0
. · · . · 0 ....J 36 46
.. . . · ., .,
46 54
.... .. . . . . · 0° ... •
· . · . · . 55 62
FIG 33. Flaw distribution diagrams for 'as-received' specimens. The
numbers refer to the number of flaws per specimen.
. . • I • .. .. , ..
0" . . • • .. . · . · . 'l
34 35
.' .. •
" .. ( ,
", .. .. , · .. . . " . . .. c 35 41 0
'';:; u ... . .. L. . . . '0 .. 0> .. c
" '0 ' .. . • a ........ a 43 46 --l
.' "t· 0:.l.". " .. '" "
" .. .. 57 61
.. "
65 71
FIG 34. Flaw distribution diagrams for 'stressed' specimens.
, ':j , , ' , , . ,
,. , , .. .. .. 31 31
:-;< .. .. ' , , ..
' .. .. "
44 49 c 0 .... u
'" ..
I-
"0
0> C
"0 0 0 51 51 -l
60 62
'\ . : ... . .. ' ':: ::t: : ': ~", : . .
I ••• , -, , , '
73 83
FIG 35. Flaw distribution diagrams .for 'stress-relieved' specimens.
c o .;; u .. L..
"0
Cl c: "0 C o ...J
3.10
3.37
3.98
5.52
6.47
3.23
3.84
4.59
6.13
6.94
FIG 36. Fracture crack diagrams. The numbers give the observed
specimen strength ( MPa ).
33 ~
, 30 ~
~ 27
,.0' 0" ~ ,~~ ,~.
24
\ \ , ,
I ,
\ \ I
(9 \ , \ , I
, (9 , ,
\ , , \
\ 8' .. \ I
\ • • I , ,
• , , , , \ , ' . , , , , , \ • , , \ , \
, • ,
6.25 6.00 5.75 5.50 5.25 5.00 4.75 4.50
FIG. 37 Compositions of blends of coals A-C-F lie at centres
of circles bearing (a) coke numbers and (b) coke tensile
strengths. Dotted lines on (a) and (b) are iso-volatile
matter content and iso-strength lines respectively.
I I
·-G~-f::\17· - - - -----\J
f;\- - ------- \J
_--r:0-------- \J
1 1 , 1
1 ,
,~I
~: ,~I
~: 'Q ~I ,
:8 I I.
I
, I·
1
I .
5.75 5.50 5.25 5.00 4.75
I , I
, , , 4.50
27
FIG. 38 Compositions of blends of coals A-E-F lie at centres
of circles bearing (a) coke numbers and (b) coke tensile
strengths. Dotted lines on (a) and (b) are iso-volatile
matter content and iso-strength lines respectively.
33
30 \
\
27 \
\
• • •
24
5.75
\
\ '.
'-. CV CV '--\ \
\ \
• , 5.5 5.25
• , 5.0
FIG. 39 Compositions of blends of coals A-B-E lie at centres
of circles bearing (a) coke numbers and (b) coke tensile
strengths. Dotted lines on (a) and (b) are iso-volatile
matter content and iso-strength lines respectively.
27
30
33 ,
24
, ,
1::> ,
, ,
•
, ., , , , ,.
, .. .... 8 ,
."
5.50 " ,
, ,
. 5.75
, 6.00
, , , , , .. ' ." 6.25
,
FIG. 40 Compositions of blends of coals A-C-E lie at centres
of circles bearing (a) coke numbers and (b) coke tensile
strengths. Dotted lines on (a) and" (b) are iso-volatile
matter content and iso-strength lines respectively.
--8- 24
. Cl. ----0-----8---
-- - -- -83 ----
5.25
,5.50 ,
5.75 .'
, .6.00
8" ,
, ,6.25 , , ,.
,8" .. ,6.50 .. , • , , , ,., ,.8' , .-.- , .. , , .. .. ,
, .. , , , , , , .. , , ,. , ,
FIG. 41 Compositions of blends of coals A-D-E lie at centres
of circles bearing (a) coke numbers and (b) coke tensile
strengths. Dotted lines on (a) and (b) are iso-volatile
matter content and iso-strength lines respectively.
'" p, :s: .Q +' bO
'" '" H +' IJ)
'" .... .~
IJ)
'" '" +'
"" '" H
" IJ)
'" '" :s:
8.0
7.0
/ 6.0 ./
:./.: :
5.0 / 4.0
4.0 5.0 6.0 7.0
Calculated tensile strength, MPa.
FIG. 42 Comparison of measured coke tensile strengths with those
calculated using an MLR equation which estimates the
strength with a standard error of 0.443 MPa.
,
< >
FIG. 43 Polarized-light micrograph showing the plain anthracitic textural component.
•
< 20~ >
FIG. 44 Polarized-light micrograph showing the patterned anthracitlc textural component.
< >
FIG . 45 Polarized-light micrograph showing the broad-flow textural component.
< 20~ >
FIG. 46 Polarized-light micrograph showing the striated-flow textural component.
< >
FIG. 47 Polarized-light micrograph showing the granular-flow textural component.
< >
FIG. 48 Polarized-light micrograph showing the coarse-D06aic textural component.
< >
FIG. 49 Polarized-light micrograph showing the medium-mosaic textural component.
r
< >
FIG. 50 Polarized-light micrograph showing the fine-mosaic textural component.
.,
• • • ..
. , • , •
< 20l'm )
FIG. 51 Polarized-light micrograph showing the i&otropic textural component.
FIG. 52 Polarized-light micrograph showing a large inert particle.
> , >
<i +>
" " " o
'" " 8
-o
ID +>
" " +>
" o o
.40 -20
-
.. 40 r-
20
~
.40
r-.2 0
r--
.4 0
.2 0
r-
.4 0
.2 0
Co. I
r---
A r--
t=l -,
.r-- /
B
r-- "-r-
~. t----1
r-r- c
r-
r--=:J h .--- -
D - r-
=- h r-
r--
E
r- h .. --=:r:= ~ -
40 F
20
--=- rL F L I Gc Gm Gf Gvf Fb+s Fg Hc Km Hf
. Fig. 53 Comparison of SEM and PLM textural compositions of single-coal cokes.
t--I
Coal A
.60
" -" 0 .... .., .40 .... .,....-+
(I)
o > Po, I'l > .20 0
,-I--
0 (I) ..... (J)
"' .... ~ 0 ::l 0 .., " ..... (J) '" .60 .., 0
0
'---0
Coal C - 4r-
~ ..... 1
'" (J) -" ..... o bO .40 .... " ..., ..... 0 Ul
'" .20 J: 1J. r-
h AIC coal blends
..... .015 '" I'l " 0 0 .... ~ .., ....
0 .010 III (I) ~ (J) ........
0 .005 '0 0 (J) (I)
"''''' (J)
~ 0 bO ~ 7'/ ~
::l .., cc o " .d ..... (J) 0 cc.., 0" ..... -.005 I 0 ~ o 0
/ ILJ
:..L '0 ..... 1 (J) cc '0 -.010 ~ ~ (J) ::l::l'O (I) .., " III " (J) (J) (J) ..... -.015 :-:..,.0
Fb Fs Fg Mc Mm Mf I
Fig. 54 Influence of blending low-volatile coal A with " medium-volatile coal C on coke textural composition.
" o .... ..., .... (J) • o > p.., s > o o -
(J) .... QJ "' ..... .. 0 => 0 ..., " .... QJ '" ..., 0
o .... 1
" QJ " .... o bJ .... " ..., .... o (J)
" ....... '" 0
.60
.40
.20
.60
.40
.20
.015
.... " § ~ .010 .... .. .., ..... o " (J) • 005 .. QJ ..........
o ] 0 Ul .., ..... QJ
~o~ => ..., " o "..cl -.005
.... QJ 0 "'.., 0
o " .... 18 8 -.01
-0 .... 1 QJ ,,-0 .. .. QJ
~ .;; -g -.015 '" " QJ QJ QJ .... =-:...,.0
Coal A Coal A Coal A
~ -~ ~
- -~ --' - f-r- -
I---, -Coal D Coal E - Coal F ~ - - ~
~ ---~ -
n - I-
AID coal blends AlE coal blends A/F coal blends
~ V ~ / r /
~ LYJ --t ~
Fb Fs Fg Mc Mm Mf I Fb Fg Fs Mc Mm Mf, I Fb Fs Fg Mc Mm 'Mf I
Fig. 55 Influence of blending low-volatile coal A with high-volatile coals D, E and F on coke textural composition.
'" 0 .,... ...., .,... Ul o > "-, ~ :> 0 ()
Ul ~ '" ",,,,, k 0 " () +' ,,~
Q) " ...., 0 ()
~ I
'" '" " ..... 0 t<l .,... " ...., .,...
() w '" k"'"
'" 0
~
" " ~ 0 0 .,... k +' ..... ()
" Ul k Q)
.... "" 0 -0 ()
Ul. Q)
+' ..... Q)
'" 0 t<l ~ k
" +' '" () " .a ~ Q) ()
"+' o ~ ~.
I 0 ~ () ()
-o~ I Q) ",-0 k k Q)
" ,,-0 UJ+, " '" " Q) Q) Q) .....
::=::~..o
-- Coal -.60
.40
.20
Coal
.60
.40
.20
r-1 B/E coal
. 015
.010
.005
B
E_ --r--
il-blends
+- Coal C ~
Coal E __ --
CIE coal blends
-- Coal C ~
~ . - -.,. I
.
-l
Coal F -
I-
CIF coal blends
771 !"Tl tr"'7"I I;-; ..,....., l; lLLI U
-.005
. -.010
-.015
Fb Fs Fg Mc Mm Jlif I Fb F~ Fg Mc Mm. Jlif I
Fig. 56 Influence of blending medium-volatile coals with high-volatile coals on coke textural composition.
~ ILL U
Fb . Fs Fg Mc Mm 11£ I
FIG 57. Carbon composed of small, near-perfect graphitic crystallites.
----~~------~------~~~~--
==-=-=--/~---=--=--=--=-/-=-A;==== ~-----~-:/ /
FIG 58. Carbon composed of extensive disordered carbon layer planes with areas of crystallite perfection.
, , ,
\
I
, , ,
b
, ,
\
I \
a
Vibration direction
c
FIG 59. Colours of isochromatic areas in mosaics depend on the orientation of the constituent crystallites (dashes) relative to the vibration direction (see c). Interfaces only bear extinction contours (black lines ) if crystallites adopt an intermediate alignment (see b) which is parallel or perpendicular to the vibration direction.
Vibration +---- ----7)
direction
/ Vibration
direction
/
FIG 60. Colour and extinction contour changes on altering the orientation of a flow structure relative to the vibration direction .
B
"
