Light Metals 2007 Edited by Morten Sorlie

TMS (The Minerais, Metals & MateriaJs Society), 2007

PROCESS OPTIMIZATION IN BAKE FURNACE

Vinicius Pifferl, Paulo Miotto', Ciro Katol, Markus Meiei, Raymond Perruchoud/ and Peter Sulgei
I Consorcio de Aluminio do Maranhao (Alumar), Department of Carbon and Technology
BR 135, Km 18, Säo Luis, Brazil, Zip code 65065-604
2 R&D Carbon Ltd., P.O. Box 362, 3960 Sierre, Switzerland

Keywords: Baking capacity, anode quaLity, optimization, temperature distribution

Abstract
ALUMAR has three open top ring type anode bake fumaces, one
Consorcio de Aluminio do Maranhao (ALUMAR) is one of the of them being shut down at the time. Fumace I consists of 76 and
largest aluminium smelters of Latin America producing 440'000 fumace 2 of 68 sections, both with 8 pits to sized for 2 layers of 5
tons ofprimary metal. The plan to increase the line current to 228 anodes [2]. On both fumaces 3 fires have been in operation.
kA in all pots and to increase the anode cycle time from 26 to 28 Figure I shows the typical configuration of the firing equipment
days required a fundamental optimization ofthe baking process. prior to the optirnization work.

R&D Carbon was chosen as a technical partner to optimize the The reduction of the anode cycle time in the pots caused an
fumace operation with the goal to increase the production immediate increase in anode demand. In order to produce
capacity and to further improve the baked anode quality to the sufficient anodes for the new demand, the bake fumace had to
required level. reduce the fire cycle from 22 to 19.5 h. The new fire cycle was
very short and caused operational problems in terms of process
This paper describes the measures taken, namely the changeover control, emissions and quality distribution of the anodes. To
from three to four fires and the systematic optimization of the minimize the deterioration of the properties, ALUMAR has
baking parameters. The positive effect of these measures is adjusted the final baking temperature from 1225 to 1250°C and
demonstrated with improved production and quality figures. increased the soaking time from 90 to 96 h.

The extremely fast fire cycle in the fumaces generated opacity

Introduction problems in the pre-heating zone. In order to rninimize these
effects, the firing configuration was changed from 3 to 2
In January 2003 when ALUMAR operated at 216 kA, slotted preheating sections. This measure deteriorated the anode
anodes were introduced to improve the pot performance. properties even more, as the anode heat-up rate becarne excessive.
Although the results in the potroom were favourable, the slotted
anodes caused the carbon consumption in the pots to increase.
This extra consumption is attributed to the extra surface area of
the anode exposed to COz attack, which increases the carbon
consumption through carboxy reactions [I]. Lnaddition, the anode
weight was reduced by approximately 20 kg due to the slots.

ALUMAR pot room has continuously increased the aluminium

production through a load up optirnization strategy. One of the
main focuses ofthis strategy is the use of higher amperage, which
means higher anode consumption.

The combination described above has led to very thin butts which
started to create operational problems in the potroom. The anode
cycle time was then reduced frorn 28 to 26 days.

Figure I. Original fire configuration prior to optirnization with 2 pre-hearing sections

The new fast fire cycle led to several undesired side effects: Increase of the furnace productivity:
The daily furnace production can be calculated from the
High heat-up rate: The uncontrolled release of volatiles number of fires in operation, the fire cycle and the
increased the rate of cracked anodes
number of anodes per section:
Environmental emissions: The difficulty to control the
DailyProduction =
volatiles release led to high opacity levels in the off-gas
nFires. nAnodes
persection.24 h / FireCycle
during extended periods
Accordingly, the anode demand from the potroom can
Short total heating time: The heat could not reach the be used to calculate the required fire cycle:
anodes in the corners of the pits. Cold spots were
generated and the maximum temperature difference fiT FireCycle =
increased. nFircs. nAnodes
persectien. 24 h / Daily AnodeDemand
Reduced refractory life: The condition of the refractory With 6 fires (from the two fumaces), 80 anodes per
bricks was expected to deteriorate due to the higher final section and a demand of 590 anodes per day, the
baking temperature and longer soaking time at elevated required fire cycle is 19.5 h.
temperature
With the number of anodes per section being a constant
Deterioration of anode quality: The measurement of the and the fire cycle time already being squeezed, the
anode properties influenced by the baking process productivity could only be raised by adding a fire to the
revealed a substantial drop of the average and an existing ones. As the fumace ] of ALUMAR consists of
increase of the variation. 76 sections of which JO are basicaUy unused, this
opportunity was used to add one additional fire and
In the first half of 2005, the anode quality became the bottleneck squeeze the existing 3 fires without adding any extra
for a further increase of the line current in the potroom. It was sections.
recognized that a process optimization in the bake fumace was
needed to overcome the numerous problems. R&D Carbon was These combined measures defined the strategy to increase the
chosen as a technical partner to support ALUMAR to reach the anode production productivity and quality to the required anode
ambitious goals of increased line current and to increase the anode demand from the potroom.
cycle time in the pots back to 28 days.

Process Optimization
Changes Required to Achleve Targets
Implementation ofNew Fire
The increase of the line current and increase of the anode cycle
time in the pots are two contradicting developments, as norrnally The original arrangement of the firing equipment in furnace I
the cycle time would be lowered to compensate the greater anode required 21 sections per fire (as shown in figure I) and 3 to 4
consumption. In order to nevertheless achieve the ambitious sections were reserved for maintenance. In order to add a fourth
targets, combined measures were required in different areas of the fire, the configuration was reduced to 17 sections per fire which
anode production: allowed to keep 4 sections for maintenance after every second
fire. The reduction of 4 sections per fire required fundamentaI
Increase ofthe baked anode weight: changes in the baking philosophy. The new configuration is
As the anode length and width could not be changed, an shown in figure 2.
increase of the anode heigbt and density are the only
two measures to raise the anode weight. The maximum The reduction of 8 to 6 cooling sections was realized by
anode height can be increased by minimizing the readjusting the cooling arrangement which increased the cooling
variation of the anode height. The anode density can be efficiency and moved the zero-point to the desired location.
improved with a process optimization in the paste plant
[3] and in the bake fumace. The newly gained capacity ofthe fourth fire was partially utilized
Reduction of the net anode consumption: to increase the fire cycle from 19.5 to 24 h. This allowed to reduce
The minimum acceptable butts weight at ALUMAR is the baking zone from 5 to 4 burner bridges, by keeping the firing
240 kg. As the maximum anode height is limited, it is time constant at 96 h.
mandatory to reduce the net consumption, in order to
respect the critica] butts weight. The net consumption The section gained from the baking zone was used to increase the
can be reduced by optimizing the key anode properties pre-hearing zone from 2 to 3 sections.
that influence the buming behaviour, namely the CO2
and air reactivity, thermal conductivity and air An additional section was made available by optimizing the
permeability [4]. With the exception of the air refractory maintenance time.
permeability, these properties are rnainly influenced by
the baking process. Optimization of the final baking In order to adjust the baking curve and firing settings to the new
temperature and minimization of the variations in the situation, a systematic optimization was executed instead of a trial
pits are hence targets. and error approach. The methodology of the optimization is
outlined hereafter.
 LOADING PRE-HEATlNG COOUNG ZONE

Figure 2. New fire configuration of furnace I after optimization

Methodology to Assess and Optimize the Oistribution of the baked under weil controlled condinons in a laboratory furnace to a
Thermal Treatment in the Bake Furnace final temperature of 1050, 1150 and 1250°C respectively with a
soaking time of 20 h, see figure 4. Three sample cores of 50 mm
The optimum baking temperature is a characteristic parameter that
diameter were driLled from each pilot electrode for subsequent
depends on numerous variables such as anode raw material
testing for key properties.
properties, characteristics of the green anode, design and
condition of the bake fumace, capability of the firing system and
limitations of emissions. The determination of the optimum
baking temperature is hence not straightforward and very
individual depending on the plant conditions. The final baking
temperature is not a criteria for the anode performance in the pots,
but rather the combination ofthe relevant anode properties. For an
optimization of the bake furnace operating parameters it is hence
essential to know the actual distribution of anode properties.

There are different methods to determine the temperature

distribution in the bake fumace, however only one to obtain also
the optimum anode property distribution:
Multi-stage thermocouples have the advantage of fast
results, but only information about the temperature of Figure 3. Production of pilot electrodes
the packing coke is obtained. Due to the short service
life (typically 3 to 4 cycles) this method is relatively
expensive and no information is obtained about the
actual anode quality.
The determination of the crystallite size Lc of green
coke filled in graphite crucibles placed in stub holes is a
standardised method according to ISO 17499. However,
the resulting so-called "Equivalent temperature" is 70 to
150°C higher than the real temperature [5] which needs
to be taken into account when results are interpreted. As
above, no information is obtained about the actual anode
quality.
The determination of key properties on sample cores
taken from baked anodes and the comparison with
calibration curves obtained from pilot electrodes allows
Figure 4. Baking ofpilot electrodes
to draw a map of the temperature and anode property
distribution. This method was applied by R&O Carbon
at ALUMAR and is described hereafter. Test anodes produced from the same paste as the pilot electrodes
were sampled after baking according to a pre-defined sample plan
and tested for the same key properties as the pilot electrodes.
250 kg ofproduction paste was collected after the mixers by using
the by-pass slide for scrap paste. A total of 36 test electrodes
The cores from the pilot and production anodes were analysed for
were produced in a pilot press with this paste, with a diameter of
the apparent density, specific electrical resistance, flexural
146 mm and a height of approximately 180 mm, as shown in
strength, thermal conductivity, CO2 reactivity, air reactivity,
figure 3 [3]. Three sets of 12 green pilot electrodes were then
density in xylene and elements (XRF). The evaluation of the location of the pitch bum could hence be improved and the
results gives infonnation about the homogeneity and consistency efficiency of the combustion of the pitch volatiles was increased,
of the production anodes, as weil as about the heat up phase and which both reduced the opacity problems in the off-gas.
soaking temperature ofthe baking process.
Furthennore, the improved heat-up rate had a great impact on
The analysis of the pilot electrodes allowed to draw calibration several anode properties, such as the variation of the specific
curves that are utiIized to assess and optimize the baking process electrical resistance and flexural strength (see Table I). In
ofthe production fumace, An example ofthe calibration curves is addition, the better control of the pitch bum helped to reduce the
given in figure 5. baking loss.

Oensity In Xylene [kg/dm') AIr Reactlvlty Oust ["10) Final Baking Temperature
2120 ~--~~--~----~----~ 8.0 The optimization measures resulted in a substantial reduction of
the maximum temperature difference within the pit from 215 to
2.105 t--"""'-I::---+--h~---1 6.0 112°C. Pit regions with temperatures lower than 1080°C could be
eliminated completely (see figure 7).
2.090 t---t-----''"'!oo.;;;;;''''--t-----1
4.0

2060
2.0

0.0
Before
Optimization ra ;--

a
r-

a
1050 1100 1150 1200 1250
Flnal Baklng Temperature [OC]
1250·C a a a
'--
-
- '----
Figure 5. Influence ofthe final baking temperature on anode
properties
1200·C a ~
f1~ë
1150·C I
Results I'"-.
-
a 11"" b,..
a ~
Baking Curve 1100·C
The implementation ofthe fourth fire and subsequent optimization
of the entire baking process resulted in a great increase of the tota!
1050·C
ra -
a ~
hearing time from 135 to 168 h, as shown in figure 6. This
allowed to reduce the final flue gas temperature from 1250 to
1225°C. Fuel consumption savings could be observed after this
reduction.
1000·C
'--
a -
a a
'--
r-- -
950·C a a
Temperature rC)
1400.-----~---- __--_,----_,r--r----_.
l<pc re
1200
After
Optimization ~ c
a
-
1000
800 ~--+-,~>-__;;I~_---i--+- -Old Curve
Figure 7. Final baking temperature distribution of anodes in the
600 ~ -New Curve pits before and after the optirnization
400 ~ t .. The elimination ofhot spots of up to 1250°C is beneficia! for the
200 +- ~f---+--t---
life of the refractory material, as weil as to avoid desulfurization
O+--r--~~~--T-~--r-~~~ effects that are deleterious for the anode quality and off-gas
o ~ ~ ~ ~ 1001~1~1W1~200 emissions.
Time(h)
Anode Ouality
Figure 6. Target baking curve before and after the optimization
Table I shows the anode properties before and after the bake
Heat-up Rate fumace optimization work.

The increase of the pre-heating time from 39 to 72 h was an

important starting point to reduce the heat-up rate. The
measurement and eva!uation of the temperature, underpressure
and flue gas components in the pre-heating zone allowed to adjust
the baking cUIVe in an optimal marmer. As a consequence, the
 Table I: lmprovement ofthe anode quality figures related to the bake furnace adaptations
Properties Unit Before Optimization After Optimization

Average 20" Average 20"

Baked apparent density kg/drrr' 1.56 0.04 1.58 0.02

Sulfur % 1.90 0.50 2.25 0.25
Spec. electr. resistance IlQm 53 9 52 4
Flexural strength MPa 11.9 5.6 12.8 3.2
Air reactivity residue % 64 6 68 3
Air permeability nPm 1.0 0.8 0.5 0.4
Density in xylene kg/drrr' 2.115 0.030 2.090 0.019

With an unchanged green density level, the baked density has The baked anode weight could be increased by 10 kg as a result of
improved as a result of minimized desulfurization occurring less sulfur and nitrogen losses during baking.
previously during baking. The slower heat-up rate and the lower
pit temperature difference resulted in a better consistency of the
specific electrical resistance and flexural strength. The air Butts Recycling and Carbon Consumption
reactivity residue was also improved as the micro-porosity
A significant improvement of the net anode consumption was
resulting frorn the desulfurization has been minimized.
observed in the potroom and therefore the butts weight increased
remarkably, as shown in Table 11.
Before
Optimization
The increase ofthe butts recycling by 10 % rel. was beneficial for
the anode weight as weil as for the air permeability. With the
70.0% better air permeability and reactivity levels an improvement ofthe
net anode consumption of JO kgC/tA1 could be achieved. As 30 kg
of carbon is consumed per day and anode, the bake furnace
67.5% optimization was a major step towards the goal of increasing the
anode cycle time back to 28 days.
65.0%

Conclusions and Outlook

62.5%
The use of pilot electrodes to assess the baking temperature and
,.....- - - subsequently optimize the entire baking process confirmed to be a
60.0% 0 0 0 systematic methodology to maximize the anode properties and
furnace productivity in view of best performance in the pots.
j- -- --
57.5% 0 0 0 As a side effect, the baking 1055, fuel consumption, S02 ernissions
'--- - '--- and opacity problems in the offgas could be reduced, and the
r--- - refractory He potentially increased.
55.0% 0 65% 0 1-

After
0
Optimization ~ -
Figure 8. Distribution of the air reactivity residue before and after
the optimization

T a bi e 1I Companson 0f an od ean db utts firgures fior 26 cvc Ida

e s
Unit Before Optimization After Optimization

Baked anode weight kg 1056 1066

Net anode consumption kgC/tAi 414 404
Butts weight kg 273 300
Besides the better economical savings related to the lower carbon
consumption [6], the fumace optimization allowed to reach 50 %
of the anode cycle time target. Trials in the green side have
demonstrated recently that potentials are left ro gain
approximately 20 kg of anode weight by:
Squeezing process variability to get more consistent
anode height
Concomitant increase ofthe anode height target
Better mixing through longer residence time in mixer
Implernentation of air bellow in vibrocompactor to
control anode density level

The opportunity of a longer mixing time (35 instead of 25 min) is

related to the fact that in the case of 28 anode cycle days in the
pots there is 8 % less paste to be produced. A plan to increase the
availability ofthe batch mixers from 6.0 (current minimum) to 7.5
is under completion to rnaximize the mixing operations to the
requested level.

This will provide the possibility to continue the bake fumace

optimization work as the lower anode block demand with 28 cycle
days will result in an average increase of the fire cycle close to 2
hours.

References
I. Markus W. Meier, Rayrnond C. Perruchoud, Wemer K.
Fischer, "Production and Performance of Slotted
Anodes", Light Metals 2007
2. Paulo Miotto, Marcos Aurelio Silva, Ciro Kato, Roberlito
Silva, "New Method to Start Up Fires in Baking
Fumaces", Light Metals 2005, 683 - 688
3. Raja J. Akhtar, Markus W. Meier et al, "Dynamic Process
Optimization in Paste Plant", Light Metals 2006, 571 -
575
4. Wemer K. Fischer, Felix Keiler and Raymond
Perruchoud, "Interdependence Between Anode Net
Consumption and Pot Design, Pot Operating Parameters
and Anode Properties", Light Metals 1991,681 - 686
5. L.P. Lossius, I. Holden and H. Linga, "The Equivalent
Temperature Method for Measuring the Baking Level of
Anodes", Light Metals 2006, 609 - 608
6. M. Meier, T. Mueller, "Influence of Anode Quality on
Aluminium Production Cost", 3rd Rodding Conference,
lceland, 2005