Você está na página 1de 8

ARTICLES

TECHNIQUES

ti on in
a
p timiz
o
d ing ants
len ke pl u)
l b r iena
Coa lor co Ma
e lor ) ys e de
Arc rce yrol
in (A de P
ch re
. D eple (Cent
S ot
ern ■ INTRODUCTION
P. P
The objective is to point out the best practices in coal blen-
ding within Arcelor coke plants in order to reduce steel pro-
duction cost by improving the coke quality and/or reducing
the blend cost. Comparing coal blends implies using single
tools for carbonization and common procedure for blend
Basically, different blends are prepared in the preparation and carbonization. This best practice exercise
Arcelor coke plants that incorporate coals of consists in two successive and complementary steps.
various ranks, LV, MV, HV, non-coking coals. The first step is a complete characterization of each blend
at laboratory scale :
A comprehensive comparison of these coal blends,
– measurement of the caking properties like F.S.I., fluidity
within the Arcelor coke plants, has been and dilatation ;
performed using two successive approaches :
characterization of each blend at laboratory scale, – petrographic investigations to determine the rank (distri-
bution of the mean reflectance of vitrinite) ;
coking tests performed in the CPM 400 kg
Movable Wall Oven. – some chemical analyses like determination of the alka-
lies contents (K2O+Na2O) in ashes are also performed on
CSR can be maintained to very satisfying values if the coke produced from these blends, because of their
coals entering in the blend composition are strictly effect on the coke reactivity.
selected as regards factors like rank and ashes
composition. The second step is a series of carbonization tests performed
with the CPM 400 kg Movable Wall Oven, using the same
coking conditions for all the blends tested.
Non-fusible components decrease the coking
pressure, impair the mechanical indices and the One of the main observations of these laboratory and pilot
CSR. This limits their incorporation below 10 %. scale tests is the possibility of adding low volatil non-
coking coals in the blends. The consequence is a reduction
Industrial results confirm the effect of the addition of the coking pressure and the possible decrease of the
blend cost.
of LV non-coking coals : the pushing forces are
reduced ; the flue temperatures must be increased The influence of non-coking coals incorporation is then
to keep the stability of the cokemaking parameters ; studied taking into account the process parameters and the
despite this temperature rise, the coke quality is evolution of the coke quality.
decreased.
Finally, the evaluation of the benefit encountered by this
An economical balance taking into account the modification of the blend is completed with regard to the
lower cost of non-coking coals and the influence of influence on the pig iron price in order to enable the global
the poorer quality of the coke on the blast furnace optimization of the blend.
fuel rates establishes that, following local
circumstances and the prices of raw materials,
the use of LV coals may lead to a reduction of the ■ BEST PRACTICE IN COAL BLENDING
cost of pig iron.
Blend characterization at laboratory

Blend composition

Blend compositions (all blends were prepared at CPM from


individual coals) have been described by each coke plant as
© La Revue de Métallurgie 2005. representative of its production (table I).

La Revue de Métallurgie-CIT Juillet-Août 2005 487


Optimisation des mélanges de
charbon dans les cokeries d’Arcelor
S. Deplechin (Arcelor)
P. Pernot (Centre de Pyrolyse de Marienau)

Une étude comparative complète des mélanges de char- Une sélection rigoureuse des charbons en termes de
bon des cokeries d’Arcelor a été réalisée avec deux cendres permet de garder un niveau de CSR satisfaisant.
approches successives : Les éléments non fusibles diminuent la pression en cours
– la caractérisation des mélanges en laboratoire, de cokéfaction, mais diminuent aussi les indices I40 et
I10, ce qui limite leur teneur à un maximum de 10 %.
– des essais de cokéfaction dans le four du CPM de Ces éléments détériorent également le CSR.
400 kg à panneresses mobiles.
Les essais industriels confirment l’effet de l’addition de
Les mélanges sont fondamentalement différents de par charbon LV non cokéfiant. On observe une réduction de
leurs teneurs en charbons australiens, le pouvoir réflec- l’effort de défournement, une augmentation de la tempé-
teur de la vitrinite, la fluidité. On peut distinguer : rature des émissions et une détérioration de la qualité du
– les mélanges constitués essentiellement de charbons coke.
LV et MV, avec des résistance à l’abrasion très satis- Le bilan des coûts, prenant en compte le prix moins
faisantes malgré une aptitude à la cokéfaction infé- élevé des charbons non-cokéfiants et l’effet de la qualité
rieure à celle des charbons MV ; réduite du coke sur le bilan thermique du haut-fourneau,
– les mélanges à forte teneur en charbons MV. Ils per- montre que selon les conditions locales, l’utilisation de
mettent de produire un coke de bonne qualité en charbons LV peut éventuellement réduire le coût de la
termes d’indices I40 et SCR, avec parfois des valeurs fonte.
non satisfaisantes d’indice I10 ;
– les mélanges qui contiennent des composants non
fusibles, du type charbon LV non cokéfiant.

TABLE I : Structure of blends tested at pilot scale. – blends based on very different coals, using both low vola-
TABLEAU I : Mélanges de charbons pour tile (LV) and high volatile (HV) coals without medium
les essais pilotes. volatile (MV) coals. They have typical vitrinite reflectance,
with high standard deviation, illustrated in figure 2a ;

Blend composition A B C D E F
– blends, more homogeneous, with a higher amount of
medium volatile coals. The vitrinite reflectance is then as
LV non coking 10 5 3 presented in figure 2b. Such blends are generally charac-
Petroleum coke 10 5
terized by higher values of fluidity (300 to 800 ddpm)
despite of small amount of non-fusible components ;
LV coking coals 46 26 54 50 31 33
MV coals 36 37 8 20 39 40
HV coals 18 27 38 20 25 19

VM (d.b.) 24.4 22.2 26.3 22.9 23.6 25.5


Blend fluidity (ddpm) 97 420 35 250 350 830

Blend design

Some specific trends are observed in the blend composi-


tion, especially the amount of Australian coals, which is
very different from one coke plant to another one. Figure 1
illustrates this difference. Fig. 1 – Variation of Australian coals content in the blends
of the different coke plants.
The structures of the blends are also well distinct. Different Fig. 1 – Teneurs en charbons australiens dans les mélanges
types of blends are observed : des différentes cokeries.

488 La Revue de Métallurgie-CIT Juillet-Août 2005


COKERIE

a)

Fig. 3 – Fluidity of Arcelor blends.


Fig. 3 – Fluidité des mélanges de charbon d’Arcelor.

b) Coking tests at pilot scale

To compare coals blending practices in Arcelor coking


Fig. 2 – Reflectance histograms of two typical blends plants, coking tests were performed on each blend. The
a) using LV + HV coals, b) with MV coals.
same carbonization test facility (CPM pilot oven) was used,
Fig. 2 – Pouvoir réflecteur de deux mélanges types, with the same charging conditions (moisture, size distribu-
à base de charbons a) LV + HV et b) MV.
tion, density) and the same coking conditions. It was thus
possible to study intrinsic differences in coal blend proper-
ties through the coke quality obtained.
– blends which incorporate non-fusible components. Few
coke plants are using some amount (5 to 10 %) of non fusible
components, like LV non-coking coals or petroleum coke, to Blend preparation
limit the cost of the blend and increase the coke yield.
The charge preparation was similar for all the blends. Coal
As seen on the upper histograms, these blends have nearly size, moisture and charging density were respectively 70 %
the same mean reflectance. Mean values of reflectance still < 2 mm, 9 % and 730 kg/m3 d.b.
remain below 1.15 % to limit wall pressure.
Coking conditions
Blend caking properties
Blends were carbonized using the same coking conditions.
Fluidities of the blends are very different. Figure 3 shows These common conditions may slightly differ from those
that blends, incorporating non-coking components like used in the coke plant which can explain some differences
petroleum coke or LV non-coking coals, have the highest on coke quality between pilot and industrial ovens. Two
fluidity. This may be surprising when considering the distinct following coking conditions were applied on each
amount of these components, but it can be explained as a blend :
necessity to accept such non-fusible components. – high coking rate (Std1) with 16 hours gross coking time
and 1,100°C of heating wall temperature of the pilot
The very low value of fluidity (< 100 ddpm), measured for oven ;
blend C using LV + HV coals, is explained by the lower
fluidity of the coals used and the low over-lap of their ther- – lower coking rate (Std2) with 19.5 hours gross coking
mal domain of fluidity. Possibility to use non-coking com- time and 1,040°C of heating wall temperature to charac-
ponent in such a blend will be a challenge and studied later. terize the sensitivity of each blend to the coking rate.

The high value of fluidity (> 400 ddpm) of the other blends Coking pressure during carbonization
is explained by the high content of coals which exhibit
higher fluidity, like MV coals (2,000 to 4,000 ddpm) and Examination of the internal gas pressure and wall pressure
some HV high fluidity coals (until 20,000 to 30,000 ddpm confirms that all these blends can be considered as safe
of maximum fluidity). blends even using the higher coking rate.

La Revue de Métallurgie-CIT Juillet-Août 2005 489


Despite the high amount of low volatile coals, some blends composition (1, 2, 3). Moreover, these blends do not incor-
do not induce high gas or wall pressures (< 5 kPa). This is porate some LV hard coking coals which can be disastrous
due to Australian low volatile coals, including a lot of iner- for CSR.
tinite, and the presence of small amount of LV non-coking
coals which tends to decrease the wall pressure.
Sensitivity of blends to the coking rate
Coke quality
Each blend was also carbonized using the second coking
Before examination of the results of this best practice, it rate (lower coking rate Std 2) that allows determination of
is necessary to mention that the blend quality through the the sensitivity of the blend to the coking rate (table III).
coke quality obtained may strongly differ from industrial
results according to some differences in the coking condi- As expected, the decrease of the coking rate induces a
tions. It is the reason why another best practice only focussed significant improvement of the I40 on the order of +2 or +3
on the characterization on some industrial cokes sampled in points of I40 per hour of coking time at 900°C (fig. 4).
each coke plant was also performed.
This effect seems to depend on the blend (until +5 or +6
Effect of blend composition points per hour of coking time for blends C and D). As
usual, the decrease of the coking rate impairs I10, but this
Main results are summarized in table II for the higher trend seems to be attenuated for blends which exhibit
coking rate Std1 (1,100°C - 16 hours). higher fluidity.

The highest abrasion strength (I10 = 18.9) in coke plant C The maximum gain of CSR measured is around 1 point per
could be explained by the use of some HV coals which hour of coking time at 900°C. This observation confirms
leads to good I10 when they are combined with LV hard previous works (4), but the trend is not observed for all
coking coals (experience derived from coking tests perfor- blends (fig. 5).
med on simplified blends). I10 index measured in pilot
oven is always pessimistic (+1 to +1.5 point), according to
the edge effect of the pilot oven. TABLE III : Main results of the coking tests performed
on each blend using the lower coking rate.
TABLEAU III : Résultats des essais de cokéfaction lente
TABLE II : Main results of the coking tests performed sur chaque mélange.
on each blend using the higher coking rate.
TABLEAU II : Résultats des essais de cokéfaction rapide Coking rate Std1 (low) A B C D E F
sur chaque mélange.
Coking pressure (kPa) 4.8 0.7 3.7 1.6 4.7 1.1

Coking rate Std1 (high) A B C D E F Coking time at 900°C (h) 16.4 15.8 16.6 16 16;6 16.5

Coking pressure (kPa) 2.9 1.4 3.4 3.1 4 1.8 I40 50.4 48.7 54.3 56 55.8 48.3

Coking time at 900°C (h) 14.4 14.1 14.9 14.1 14.2 14.3 I10 20.9 19.9 20 21.6 20.4 23.2

I40 45.7 42.8 44 45.1 50.8 40.2 CSR 54 59 58 59 56 54

I10 19.9 22.7 18.9 22 19.6 22.8

CSR 58 61 58 63 55 52

The good fragmentation strength (I40) observed in blend E


can be explained by the significant amount (25 %) of some
MV coals for I40 previously selected using coking tests on
individual coals. Such MV coals with Ro(%) between 1.02
and 1.08 % are characterized by very satisfying caking pro-
perties (typical maximum fluidity is between 2,000 to
4,000 ddpm). These coals lead to rather poor I10 despite
these rather high caking properties and this trend is also
detected in blends.

Higher values of CSR (CSR > 60 for B and D) are obtained


Fig. 4 – Effect of coking rate on I40.
with blends which incorporate high amount of some LV
and MV coals well adapted according to criteria like ashes Fig. 4 – Effet de la vitesse de cokéfaction sur l’I40.

490 La Revue de Métallurgie-CIT Juillet-Août 2005


COKERIE

Fig. 7 – Effect of the incorporation of LV non-coking coals


Fig. 5 – Effect of coking rate on CSR. on the I40 (coking tests at pilot scale).
Fig. 5 – Effet de la vitesse de cokéfaction sur la réactivité Fig. 7 – Effet de l’incorporation de charbons LV sur l’indice I40,
au CO2 (CSR). mesuré sur installation pilote.

■ SPECIFIC EFFECT OF LV
NON-COKING COALS
Some complementary coking tests were performed on
blends with and without incorporation of non-fusible com-
ponents, especially LV non-coking coals. Such a practice
tends to increase the mean reflectance of the blend but
without any risk of increase of wall pressure according to
the nature (inert) of this addition.

Results clearly show that an LV addition even significantly


reduces the coking pressure (fig. 6). Unfortunately, such an
addition always impairs the coke quality, especially I40 and
I10 (fig. 7). This detrimental effect limits the incorporation Fig. 8 – Effect of the incorporation of LV non-coking coals
on the CSR (coking tests at pilot scale).
of LV to a maximum on the order of 10 %.
Fig. 8 – Effet de l’incorporation de charbons LV sur le CSR,
Impairment of CSR is also measured especially when ini- mesuré sur installation pilote.
tial value of CSR is high (fig. 8).
■ INDUSTRIAL OBSERVATIONS
Thus an addition of LV non-coking coals must be limited to
a maximum of 10 % in blends charged by gravity (wet). The incorporation of LV non-coking coals at the Sollac
coke plant in Dunkerque was carried out to keep a safe
blend when standard soft-coking coals were not available in
sufficient quantity. In addition to the lower price and coking
pressure, this addition enables to keep a low volatile matter
content. This point is of interest regarding the coke yield
and the carbon deposition rate.

Due to blast-furnaces constraints, industrial coke quality


must be kept between narrow margins. When the coke qua-
lity decreases, process and blend adaptations must be com-
pleted in order to ensure a rapid recovery of the coke quality.
The first modification is the reduction of the non-coking
coals content in the blend even if the quality worsening is due
to process alteration or out of range quality of other coals.

Fig. 6 – Effect of the incorporation of LV non-coking coals In order to give as confident as possible conclusions, two
on the wall pressure measured at pilot scale. periods have been defined. Each period comprises LV non-
Fig. 6 – Effet de l’incorporation de charbons LV sur la pression coking coals levels ranging from 0 to 15 %. The first period,
de paroi, mesurée sur installation pilote. named hereafter period A, corresponds to the introduction

La Revue de Métallurgie-CIT Juillet-Août 2005 491


of LV non-coking in the blend, from October 2000 to TABLE IV : Heat consumption and flue temperature
December 2001. The second period, named period B, evolutions for a LV non-coking percentage increase
extends from January 2002 to November 2003. of 10 %.
TABLEAU IV : Évolutions des besoins thermiques et
The following observations are based on industrial values, de la température des gaz émis pour une augmentation
each one corresponding to a different blend. They are obtai- de 10 % du taux de charbon LV.
ned by battery, except the coke quality which is measured
on the whole coke production. Period A Period B
B6 B7 B6 B7
Pushing forces Heat consumption (%) + 0.3 - 1.0 + 2.3 + 0.8

Flue temp. (°C) +10.3 + 6.8 + 6.7 + 4.6


In agreement with the results of wall pressures measured at
the 400 kg Movable Wall Oven, the pushing forces decrease
as the LV non-coking content of the blend is raised.

Figures 9 and 10 show the evolution of the maximum force Thermal considerations
applied on the coke cake and of the total power delivered to
the pushing ram. Due to the poorer caking properties and the need to keep
the coke abrasion strength, the batteries flue temperatures
The mean reduction of the pushing force for an increase of were slightly increased. Table IV gives the evolution of the
non-coking coal of 10 % is equal to 6.5 t for period A and heat consumption and of the flue temperatures.
1.6 t for period B. In the same way, the power required is
lowered of 9 and 2 percent. The higher thermal level of the batteries is roughly counter-
balanced by the smaller thermal need for the LV coals. This
results in a small variation of the batteries consumption.

Influence on coke quality

Due to continuous process and blend adaptations, the indi-


vidual influence of LV non-coking coals on the coke abra-
sion and fragmentation strengths is not easy to determine.
Nevertheless, taking long periods in consideration, a gene-
ral trend can be given :
– the coke quality is impaired by the incorporation of LV
non-coking coals (fig. 11-12) ;
– the increase of the I10 during the studied periods can be
Fig. 9 – Influence of the LV non-coking content on
the maximum pushing force (battery 6). estimated to 0,61 and 0,26 point. The decrease of I40 is
equal to 0,71 and 0,97 point.
Fig. 9 – Influence de la teneur en charbon LV sur l’effort
d’extraction maximal (batterie 6)

■ ECONOMIC EVALUATION
The economic evaluation of the LV non-coking coals incor-
poration in coke blends is not easy to perform. The usual
estimation based on coke yield and impurities content is not
sufficient, as the decrease in the coke strength must be
taken into account on the blast-furnaces ratios. The main
difficulties arise from the determination of the coke quality
modification resulting from the use of non-coking coals
and from the influence of this degradation on the blast-fur-
nace parameters.

The short calculation described here supposes that the qua-


Fig. 10 – Influence of the LV non-coking content on the total lity of the coke can be completely estimated via the I10 and
pushing power (battery 6). I40 indexes and that the sole blast-furnace ratio influenced
Fig. 10 – Influence de la teneur en charbon LV sur la puissance by the coke quality is the coal injection rate. The last hypo-
d’extraction maximale (batterie 6). thesis is the linearity of all the properties.

492 La Revue de Métallurgie-CIT Juillet-Août 2005


COKERIE

of coke. This supposes that the process parameters (heat


losses, global reductant rate, pig iron quality) are kept
constant by the adaptation of the coal injection rate to the
coke quality.

Generally, publications relating the influence of coke qua-


lity on coal injection indicate an increase of 5 to 10 ‰ of
the coal injection for an increase of the I40 index of one
point. The value of 7.5 ‰coal is kept for the rest of this eva-
luation.

Coke value estimation


Fig. 11 – Influence of the LV non-coking content on I10.
Fig. 11 – Influence de la teneur en charbon LV sur l’indice I10. The present work is limited to the influence of the LV non-
coking coals. Therefore, the estimation of the coke value
will only take into account the influence of the decrease in
coke strength.

The impurities content such as ashes, sulphur, phosphorus


and alkalis of the non-coking coal are assumed equal to the
content of the blend. The economical study keeps the coke
and by-product yields constant. Economically, this sup-
poses either a blend modification to keep the volatile mat-
ter constant, either internal or external valorization prices
equivalent to the cost of coke.

Supposing that the coking coal is 15 € more expensive than


the steam coal, the use of 10 % of LV non-coking coal
decreases the cost of the blend of 1.5 €/twet coal. As a result,
Fig. 12 – Influence of the LV non-coking content on I40. the use of LV non-coking leads to a benefit of 2.2 €/tcoke.
Fig. 12 – Influence de la teneur en charbon LV sur l’indice I40.

Benefit estimation
Blast-furnace constraints In the case of the Dunkirk plant, the local coke rate is
approximately 200 ‰. As a consequence, the cost reduc-
As for coke quality, it is always difficult to give the own tion of the pig iron when the blend includes 10 % of LV
effect of one change on blast furnace performances because non-coking coal amounts to 0.44 €/tpig iron.
many parameters vary simultaneously. A presentation of
blast-furnace engineers from different Arcelor plants (5) According to the industrial results, the coke quality index
gives an idea of the influence of the burden on the BF decreases of approximately two points. The influence of
ratios. This work is related here because it leads to a prac- this degradation is a 15 ‰coal rate reduction of the coal
ticable relation and a well-accepted coefficient. injection rate. The economic evaluation of the benefit resul-
ting from the PCI directly gives the interest of the incorpo-
Based on the study of blast-furnace permeability, this pre- ration of non-coking coals.
sentation enables to give the relative influence of abrasion
and fragmentation indices on the blast-furnaces ratios. In the present case, this incorporation presents an interest if
Permeability is one of the main process indicators on blast the profit of coal injection is less than 29.3 €/tcoal. This level
furnaces. It is directly linked to the maximum productivity depends on operational costs, coal and coke market prices.
and coal injection rate achievable.

The shaft permeability was statistically correlated to the ■ CONCLUSIONS


coke quality at Cockerill-Sambre. The index used is a linear
function of the I10 and the I40. The quality index determi- In spite of tight rank variation, high number of components
ned is : I40-3,4*I10+100. in the blends, it was possible to explain some differences in
coke qualities obtained especially by comparison with
The next step is to determine the influence of the coke qua- some results from an experimental database of individual
lity index on the pig iron cost. This estimation is based on coals (and simplified blends) carbonized using a standard
the fact that the benefit or loss in the pig iron cost can be procedure.
summarized by the amount of coal injected in substitution

La Revue de Métallurgie-CIT Juillet-Août 2005 493


The highest abrasion strength observed using LV and HV ■ REFERENCES
coals in blends could be explained by analogy to some
other simplified blends by the association of some adapted (1) KERKKONEN (O.), MATTILA (E.), HEINIEMI (R.) –
coals. The correlation between reactivity and ash mineralogy of
coke. Ironmaking Conference, proceedings (1996), p. 275-
The high fragmentation strength (I40) observed in some 281.
other blends can be explained by the high amount of some (2) TODOSCHUK (T.-W.), GRANSDEN (J.-F.) – Develop-
adapted MV coals. Such MV coals with Ro(%) between ment of coke strength after reaction (CSR) at Dofasco. Iron
1.02 and 1.08 % are characterized by very satisfying caking & Steel Technology (March 2004), p. 73-84.
properties (typical maximum fluidity is between 2,000 to
(3) KERKKONEN (O.), ARENDT (P.), KÜHL (H.) – Effect
4,000 ddpm). of coke structure on an increase in CSR range during the
production period. Ironmaking Conference, proceedings
Higher values of CSR are obtained with blends which (2002), p. 393-404.
always incorporate coals with low ashes basicity index. In
all cases, coals which are disastrous for CSR are absolutely (4) VOGT (D.), DEPOUX (M.) – Coke reactivity prediction by
texture analysis. Fuel Processing Technology, 24 (1990),
avoided in these blends. p. 99-105.
Blends which incorporate a small amount of non-fusible (5) FRANSSEN (C.), GOIMARD (J.-C.), NEGRO (P.), PETIT
components like LV non-coking or petroleum coke lead to (C.), PIERRET (H.) – Observations of blast furnace opera-
poorer indices of quality especially I10. Complementary tions in regard with the coke quality. Young European
Blast-Furnace Engineers. Scunthorpe (May 16-17, 2002).
coking tests performed with and without increasing content
of non-fusible components confirm that such an addition
must be limited to a maximum of 10 % and that it decreases
the coking pressure.

The industrial use of LV non-coking slightly modifies the


operating parameters. These coals enable to limit the
pushing forces and power.

Despite a higher batteries temperature level, the coke qua-


lity is impaired. An incorporation of 10 % of LV non-
coking coal decreases the coke quality index of 2 points.
Such a decrease can lead to a loss of 15 ‰ of the coal injec-
tion rate at the blast-furnaces.

A linear economical balance based on the coke quality


degradation and the blast-furnaces parameters shows that,
depending on the raw material prices, the use of non-coking
coals can lead to a decrease in the pig iron cost.

494 La Revue de Métallurgie-CIT Juillet-Août 2005

Você também pode gostar