Preventive Treatment of Anode Effects Using

on-Line Individual Anode Current Monitoring

Lukas Dion, Charles-Luc Lagacé, François Laflamme,

Antoine Godefroy, James W. Evans, László I. Kiss, and Sándor Poncsák

Abstract
Anode effects (AE) are considered a nuisance for aluminium production due to the
numerous negative impacts that they generate in an electrolysis cell. Previously, great
efforts have been deployed to minimize the occurrence of this event. Using online anode
current monitoring, Alouette introduced an algorithm to detect abnormalities prior to an
AE, allowing sufficient time to apply a corrective action before its occurrence. Several sets
of strategies were tested to evaluate the best approach to correct the situation without
generating additional problems in the cells. Finally, individual anode currents measure-
ments were connected to the cell control system of two pots to automatically launch
preventive treatment of AE. A decrease in the total number of anode effects along with an
increased cell stability is noticeable. Other indicators, such as anodic incidents, alumina
dosage and metal purity were also compared to make sure that no deterioration in the cell
conditions occurred over time.

Keywords
Anode effect PFC emissions Individual anode monitoring Preventive anode effect
treatment

Introduction primary source of GHG (carbon dioxide) is inherent to the

production of aluminium because the carbon participates in
In the past decades, great efforts have been deployed to the electrolysis reaction. However, another kind of GHG is
reduce the greenhouse gases (GHG) emitted by the primary also associated with the primary aluminium production,
aluminium industry. For smelters using carbon anodes, the namely the perfluorocarbons (PFCs). CF4 and C2F6 are
known to have a respective global heating potential 6630
and 11,100 higher than the CO2 [1]. PFCs are periodically
L. Dion (&) produced during an event called anode effect (AE) occurring
Université du Québec à Chicoutimi, 555 boulevard de when insufficient alumina is present in the electrolytic bath.
l’Université, Chicoutimi, QC G7H 2B1, Canada For this reason, the molten fluoride electrolyte will dissociate
e-mail: [email protected]
to support the passage of electric charges and PFCs will be
C.-L. Lagacé
F. Laflamme
A. Godefroy generated during this reaction. The occurrence of an AE is
Aluminerie Alouette Inc, 400 chemin de la Pointe-Noire,
dependent on the electrolysis process conditions and the
C.P. 1650, Sept-Îles, QC G4R 5M9, Canada
total number of AEs can theoretically be minimized down to
J.W. Evans
zero, even if the alumina concentration is kept below satu-
Wireless Industrial Technologies, 4096 Piedmont Ave. #193,
94611, Oakland, CA G4R 5M9, USA ration in order to prevent sludge formation. Hence the reason
why the efforts of the industry have been able to reduce
L.I. Kiss
S. Poncsák
GRIPS, Université du Québec à Chicoutimi, 555 Boulevard de significantly the total amount of PFCs emitted between 1990
l’université, Chicoutimi, QC G7H 2B1, Canada and 2010 [2].

© The Minerals, Metals & Materials Society 2017 509

A.P. Ratvik (ed.), Light Metals 2017, The Minerals, Metals & Materials Series,
DOI 10.1007/978-3-319-51541-0_64
510 L. Dion et al.

This paper describes the most recent efforts deployed by phase of its formation, that is to say when a small number of
Aluminerie Alouette to reduce significantly the AE fre- anodes are affected. The International Aluminium Institute
quency, hence the total PFCs emissions. Using on-line anode [4] recently adopted the term Low-Voltage Anode Effects
monitoring, it is possible to use an algorithm to anticipate (LVAE) to describe all PFCs emissions occurring below the
abnormalities that are indicating upcoming anode effects. voltage of detection, set by each smelter (generally 6 or
Once the information is known that an AE is imminent, it is 8 V). If the threshold is reached, all PFCs generated by this
important to react quickly and efficiently to eliminate the cell are then considered as High-Voltage Anode Effects
abnormalities in the cell without generating perturbations (HVAE) for the duration of the AE.
that can generate other problems over a long-term period. Knowing that the cell voltage is a bad indicator to detect
Manual tests were performed with three different preventive LVAE or foresee upcoming HVAE, additional information
AE mitigation strategies to evaluate their efficiency at is necessary to eventually detect these abnormalities. Rye
unblocking alumina feeders that were not distributing alu- et al. [5] demonstrated that a low alumina concentration can
mina properly. have an impact on the current distribution among the anodes.
Finally, the system has been running automatically for More importantly, it was shown that prior to an anode effect,
almost a year on two cells equipped with online anode several anodes will significantly change their current load,
monitoring and very good results were achieved. Key per- indicating that the event (AE) can be detected prior to the
formance indicators prior and during the test are shown and usual shift in cell voltage. Other studies confirmed [6, 7] that
described in the last section of the article. online measurements of individual anode currents is a viable
solution to preventively detect anode effects. Figure 1
illustrates that the local current drops significantly for some
Early Detection of the Anode Effects anodes during LVAE. In this case, as the CF4 concentrations
rose from 0.2 ppb up to 1.5 ppm, the current from anode
The production of primary aluminium requires the passage A16 and A17 respectively decreased by 50 and 15%. After
of a constant electric current through an electrolysis bath. this quick surge in emissions, CF4 remained under A16 and
Passage of these electric charges is made possible by the A17 and thus increased the local electrical resistance. Hence
redox reaction 1 that occurs in the presence of alumina. the current has been redistributed according to the respective
However, if the alumina concentration is not sufficient, resistances of all anodes and the production of aluminium
reaction 1 alone cannot withstand the passage of current and continued under these new conditions. The same phe-
other reactants will be electrolyzed by reactions 2 and 3. nomenon occurred two more times before the cell finally
reached an HVAE six hours later. Only at that moment, the
2Al2 O3 þ 3C ¼ 4Al þ 3CO2 ð1Þ anode effect stopping procedure was started to resolve the
problem. Any similar procedure launched during any of
4Na3 AlF6 þ 3C ¼ 4Al þ 3CF4 þ 12NaF ð2Þ the three LVAE that occurred before the HVAE could have
eliminated several hours of PFC emissions as well as the
2Na3 AlF6 þ 2C ¼ 2Al þ C2 F6 þ 6NaF ð3Þ
HVAE itself.
When CF4 and C2F6 are generated, they will most likely For this reason, an algorithm was developed to detect any
stay under the anodes due to the high adhesion of those sudden change in individual anode currents that are related
gases under carbon anodes. Because these gases have an to an anode effect, with the main focus being the prevention
important electrical resistivity, the electrical current will be of HVAE. Two different sets of conditions need to be
redistributed among the other anodes, increasing their reached for the cell to be considered close to an AE. The first
respective current density. Consequently, the alumina will condition is similar to the methodology published in 2015 by
deplete faster in that area, eventually leading to the spreading Dion et al. [7]. It involves the use of a short-term (1–5 min)
of reactions 2 and 3 in the cell. Hence the overall mechanism and a long-term moving average (15–25 min) for each
that generates PFCs is auto-amplifying if no significant individual anode. It is possible to observe in Fig. 2 that the
change is applied to the process strategy. current behaviour is significantly different in the short-term
As the number of anodes under which PFCs are trapped moving average (STMA) with comparison to the long-term
increases, it will lead to an increase of the overall cell moving average (LTMA) used as the reference. Henceforth,
voltage as well. However, Wong et al. [3] calculated that the once the ratio STMA/LTMA has reached a predetermined
voltage increase is not significant if PFCs are only located threshold for a specific number of anodes, the first condition
under a limited number of anodes. It was calculated that of the algorithm becomes satisfied.
when the voltage of detection of an AE is reached at 6 V, Using exclusively the first condition is sufficient to detect
almost half of the anodes are already in contact with PFCs. upcoming anode effect but it can also generate false alarms
For this reason, PFCs are very hard to detect in the initial by detecting other events in the cell such as anode change,
Preventive Treatment of Anode Effects Using on-Line … 511

Fig. 1 Individual anode currents

and CF4 concentration from a A19 A18 A17 A16 A15 CF4
single electrolysis cell during
LVAE 1.5 4

1.3
3.5
1.1
3

Anode normalized current

CF4 concentra on (ppm)

0.9
2.5
0.7

0.5 2

0.3
1.5
0.1
1
-0.1
0.5
-0.3

-0.5 0
8:52 9:00 9:07 9:14 9:21 9:28 9:36 9:43 9:50 9:57

Fig. 2 llustration of a cell

behavior leading to preventive
AE treatment. Upstream side.
(Stars represents maximum
current values reached for each
individual anodes during the short
period)

metal tapping or cell instability. To maintain the same period. Moreover, all the maximum values were measured
detection sensitivity, an additional condition was established within a two-minute interval.
which can be attributed most likely to the generation of Finally, if both conditions are satisfied for a specific cell,
PFCs. This second condition investigates the maximum it is highly plausible that PFCs are generated under its
current value of each anode within a short period similar to anodes. Hence, if adequate actions are undertaken at this
the STMA. It was observed that the current value of each moment, the PFC emissions from this cell can be signifi-
anode tends to reach a higher value than its reference state cantly reduced.
during an AE. This phenomenon is linked to the current
redistribution that occurs when resistive gas is moving under
the anodes and can occur very quickly (less than 5 s). Hence Establishment of the Preventive AE Treatment
it might have no significant effect on the average value of
each respective anode. Figure 2 clearly demonstrate that 6 Numerous elements have to be considered to develop a
anodes out of 10 temporarily reached a maximum higher preventive AE treatment. It has to react quickly enough to
than the anode that is driving the most current in the LTMA eliminate any imminent AE but the treatment itself must be
512 L. Dion et al.

controlled so that it does not bother the cell significantly additional false alarms, and so on. For these reasons, actions
over a long period if numerous preventive AE treatments are from the feeders should be defined with great care and at the
launched. AE are essentially caused by an insufficient same time, other ways to route alumina should be
amount of alumina dissolved in the bath in a small region or considered.
in the entire cell. This lack of alumina can be caused by Quenching procedure: movement of the anode beam has
conditions such as low bath height or low bath temperature been used for years in the industry to help cure AE. Gen-
in the cell that makes the transport or dissolution of particles erally, quenching is composed of several up-and-down
difficult in certain regions of the cell. The preventive AE motions of the anode beam to help clear out the gas from
treatment should be designed to assure a decent amount of under the anode. In addition to this positive effect, it
alumina dissolves into the bath under a short time period to increases the movement of the bath, in favor of the disso-
correct the situation. lution of alumina. Moreover, as the anodes are immersed
In another case, an insufficient amount of alumina can be into the bath, the liquid level is higher and bath can get more
fed to the cell to compensate the necessary consumption easily in contact with the crust. This crust is composed of a
from the electrolysis. This problem generally occurs when substantial amount of alumina which then contributes to the
an alumina feeder hole is obstructed by an excessive amount alumina input. However, excessive movements of the anode
of alumina. This phenomenon occurs when bad dissolution beam can cause severe disruptions in the anode cover,
conditions are present and thus the alumina becomes com- exposing the anodes to increased oxidation while threatening
pacted by the repeated actions of the crust breaker. Under the thermal balance of the cell. Downward movements can
such conditions, the alumina is shaped like a mound, starting also compress the anode against the side ledge, causing
from the cathode up to the alumina feeding chute. Simula- solidified bath to stick under the anodes, increasing insta-
tion performed by Dassylva and al. [8] showed that if the bility and eventually leading to anodic incidents (spikes).
alumina is clumped, the dissolution efficiency is significantly Finally, in the worst case scenario when bath levels are low
reduced. In the reality, such circumstances cause even poorer and when anode beam movements are badly set up (e.g.
conditions than that suggested by the simulations, therefore significant displacement upwards), it could lead to a rupture
the alumina dissolves slower than it is fed, causing an of the circuit (anode out of the bath); an event that must be
accumulation. Usually, a manual intervention is necessary to avoided at all cost. Therefore, movements of the anode beam
resolve the problem in such situation. However, the pre- need to be set correctly for the treatment to assure an ade-
ventive AE treatment should be designed to unblock the quate solution.
majority of feeders facing this problem. Additional energy input: Another adjustable element is
the energy input in the cell. It is possible to increase the
anode-cathode distance (ACD) for several hours in order to
Preventive AE Treatment Parameters increase the internal heat in the cell generated by Joule
effect. Additional energy will play a positive role in the
To design the preventive AE treatment, the ensemble of dissolution of the alumina. Moreover, the additional volume
parameters that can be controlled are divided into three of bath under the anode (higher ACD) will favor the trans-
different categories. For each of them, there are specific port of alumina to a region further away from the feeders. On
considerations to take into account when establishing the the other hand, additional energy might disrupt the thermal
preventive AE treatment. balance of the cell. More importantly, it can melt away the
Alumina feeder actions: once an alarm is sent to the cell side ledge and expose the sidewalls of the cell to the elec-
control system, commands such as “stopping or changing trolyte. Energy inputs are not a short-term solution to pre-
the rate of injection” can be sent to the crust breakers and the vent imminent AE and could be used as a final step to the
alumina feeders. Knowing that a lack of alumina is probably preventive AE treatment.
the root cause of the problem, launching an overfeeding
appears as the logical step to quickly solve the problem.
However, such action could easily make the cell’s conditions Efficiency of AE Treatments to Unblock
worse for two sets of circumstances that are likely to occur. Feeder Holes
If a feeding hole is blocked, this solution will only increase
the pile of alumina above the crust. It will also increase the By considering the numerous elements presented in the last
risk of blocking other feeders. Additionally, if a false alarm section, Aluminerie Alouette tested three different preventive
occurs, the additional alumina could increase the alumina AE treatment algorithms to evaluate their respective effi-
content in the bath close to the solubility limit. Therefore, the ciency to unblock feeder holes. During the duration of all the
preventive AE treatment could generate muck in the cell, 3 treatments, no alumina is fed into the cell. However, the
which would likely increase the instability and might cause action of the crust breaker is maintained active in order to
Preventive Treatment of Anode Effects Using on-Line … 513

push some alumina into the bath while movements of the number of quenching cycle is the most important parameter
beam allows for an increased contact between the electrolyte in favor of unblocking feeder holes. Moreover, a certain
and the undissolved alumina. The main difference of the amount of time is necessary to allow for the dissolution of
treatments resides in the quenching pattern illustrated in anode cover material and alumina. The results from treat-
Fig. 3. The goal was to test a “non-aggressive” (A) and two ment C indicate that a larger amplitude of anode beam
“aggressive” treatments (B & C) respectively. Treatment B movement is not as efficient as a longer quenching period.
has a high number of quenching cycles and treatment C has Bath level measurements performed during the tests
a larger amplitude in the movements of the anode beam. No indicate that the bath level difference occurred after treat-
additional energy was added to the cell after these ment B was applied to the cell. On the other hand, the
treatments. average bath level difference between each group is signif-
During several weeks, manual tests were performed to icantly smaller than the respective standard deviation of each
evaluate the capability of each treatment to unblock feeder treatment. For this reason, the average bath level difference
holes. When a feeder was observed in conditions similar to between each group can’t be considered as significant.
Fig. 4A, a preventive AE treatment was manually initiated It is important to take into considerations that these tests
on this cell to observe its effect on the feeder hole. In represent the conditions subsequent to a preventive AE
addition, the bath level was measured prior and after the treatment. In some cases, it is possible that the feeding holes
treatment for supplementary information. might be blocked again, thus the interest in temporarily
The results shown in Table 1 indicate that the three pre- adding some energy to the cell. Moreover, the preventive AE
ventive AE treatments were successful in unblocking feeder treatment will not be efficient enough if a mechanical
holes for the majority of the cases. Results from treatment B problem is present in the cell (e.g. feeder malfunction) or if
showed an important increase in the efficiency of the pre- the bath level remains significantly too low. However, the
ventive AE treatment to unblock the holes while the results preventive AE treatments are registered as an intervention in
from treatment C were slightly lower than the results from the cell control system. Hence, multiple preventive AE
treatment A. Therefore, it is possible to determine that the treatments will send an alarm to the process technicians and

Fig. 3 llustration of three different preventive anode effect treatment procedure

514 L. Dion et al.

Fig. 4 llustration of a feeder

before and after a preventive AE
treatment. a Blocked feeder.
b Unblocked feeder after a
successful treatment

Table 1 Success rate and the Treatment A Treatment B Treatment C

difference in bath levels for the
tested preventive AE treatments Success rate (%) 68 90 60
Average bath level difference (after—before) (cm) −0.25 0.30 −0.39
Bath level difference standard deviation (cm) 0.73 1.30 0.95
Number of feeders tested 57 30 25

actions can be undertaken to address the problem. Hence- level, temperature target, etc. were also similar for all the
forth, the preventive AE treatment allows for additional time cells.
to operate without the negative impacts of AE occurring in For comparison, four different periods were selected.
the cell. Period #1 was the reference period where no actions were
taken on the cell related to the preventive AE treatment. In
period #2, preventive AE treatment was active for both test
Key Performance Indicators of Cells cells in the preliminary phase of the project. Then in period
with Automatic AE Treatment #3, optimisations were performed on the system to increase
the detection efficiency. For periods #2 and #3, preventive
To evaluate correctly the effectiveness of the preventive AE AE treatment A was activated once an imminent AE was
treatment, the algorithm was connected to the cell control detected. Finally, in period #4, the same parameters
system on two cells to launch automatically the preventive remained for the detection of AE but preventive AE treat-
AE treatment when required. A close follow-up of the per- ment B was active. However, one of the two test cells was
formance indicators of the selected cells will allow for a excluded from period #4. For this cell, the results were not
better evaluation of the overall efficiency of such treatment representative of the preventive AE treatment’s real potential
during several months of operations. due to a significant number of mechanical problems that
To perform this analysis, two cells were implemented occurred in this cell.
with automatic AE detection and preventive AE treatments. The specific duration of each period is listed in Table 2
For comparison, 5 other cells were monitored as a reference. and the key performance indicators for all periods are listed
All of these 7 cells are part of the AP40LE technology, using in Table 3.
four point-feeders and prebaked anodes. The current inten- The principal performance indicators that were investi-
sity of all these cells remained at the same level for the entire gated are the total number of anode effects and the anode
period. Finally, process parameters such as bath and metal effect overvoltage. The results (illustrated on Fig. 5) clearly
Preventive Treatment of Anode Effects Using on-Line … 515

Table 2 Details on the four periods selected for testing the preventive AE treatments
Periods Active treatment protocol Start date End date Total number of days
1. None 2015/09/01 2015/12/10 101
2. Treatment A 2015/12/17 2016/03/31 106
3. Treatment A (with optimized detection) 2016/04/01 2016/06/14 75
4. Treatment B (with optimized detection) 2016/06/15 2016/08/20 66

Table 3 Comparison between the test group and reference cells over four periods
Period #1 Period #2 Period #3 Period #4
Reference Test Reference Test Reference Test Reference Test
cells cells cells cells cells cells cells cells
Total number of AE (AE/cell*day) 0.89 0.90 0.68 0.36 0.49 0.18 0.68 0.16
Standard deviation 1.51 1.48 1.17 0.79 1.02 0.45 1.11 0.44
Anode effect overvoltage (mV) 1.81 2.39 0.92 1.05 2.08 0.39 1.18 0.70
Standard deviation 6.25 7.30 2.44 5.23 21.85 1.37 3.06 2.90
Cell instability (nanoOhms) 121 94 102 64 95 55 124 58
Standard deviation 74 71 56 24 58 28 63 40
Additional energy input (nanoOhms) 0.056 0.046 0.051 0.034 0.049 0.032 0.053 0.030
Standard deviation 0.036 0.028 0.028 0.024 0.049 0.018 0.023 0.018
Daily mass of bath produced (kg/cell*day) 50 29 52 42 94 130 48 60
Standard deviation 292 224 334 209 290 260 297 230
Number of alumina doses (doses/day) 5692 5684 5753 5781 5628 5617 5679 5688
Standard deviation 347 323 358 218 398 258 362 203
Parasite alumina feeding indicator 5.81 6.55 4.85 4.55 6.50 7.41 5.93 5.88
Standard deviation 3.80 4.25 4.01 1.94 3.67 2.67 4.56 1.63
Iron level in aluminium (ppm) 650 475 506 516 678 582 800 720
Standard deviation 182 106 123 72 139 162 123 263
Number of anodic incidents (Normalized to 165% 110% 131% 72% 81% 65% 146% 7%
potline average)
Standard deviation 329% 223% 279% 182% 234% 187% 308% 55%
Bold values represent average value for each parameters and italics represents the standard deviation values of the same parameter

Fig. 5 Performance indicators

related to anode effects for the 2.00 2.50
overvoltage (mV/cell)
Number of AE
(AE/cell*day)

2.00
Daily anode effect

four periods of testing 1.50

1.50
1.00
1.00
0.50 0.50
0.00 0.00
period #1 period #2 period #3 period #4

Number of AE (Reference) Number of AE (Test)

Anode effect overvoltage (Reference) Anode effect overvoltage (Test)

indicate that a significant change occurred once the system in periods #3 and #4. It illustrates the effectiveness of the
was fully operational. The daily number of AE was similar algorithm to detect AE as well as the efficiency of the pre-
to the reference cells in period #1 and dropped significantly ventive AE treatment to eliminate the threat before the event
516 L. Dion et al.

occurs. Moreover, the AE overvoltage significantly dropped Conclusion

alongside the total number of AE indicating that the mag-
nitude of the anode effects that still occurred was not worse A successful preventive treatment procedure was presented
than usual. Hence by considering that the total PFC emis- in this paper to minimize the occurrence of anode effects.
sions are calculated based on a linear relation between a This procedure is based on an algorithm that detects the
predetermined factor and this indicator, the results confirms variation in the individual anode currents that indicates an
that fewer PFCs are produced when using this system. anode effect is about to occur. It is effective to detect both
However, to accurately quantify the reduction, on-line gas low voltage and high voltage anode effects. This algorithm
measurements monitoring should be performed to account then sends an alarm to the cell control system and a pre-
for the low voltage anode effects as well. ventive AE treatment can be initiated on the cell to avoid the
A positive effect was observed on the cell stability when negative impact caused by an AE.
using preventive AE treatments. Even if the instability of the Three different preventive AE treatment procedure were
test cell were lower than that of the reference in the control tested to evaluate their ability to unblock feeder holes.
period, the difference increased while the preventive AE A success rate ranging from 60 to 90% was observed.
treatment was active. More importantly, the standard devi- Additionally, two different preventive AE treatment proce-
ation of this variable was reduced significantly indicating dure were tested and automatically launched on two cells for
that values are less scattered and closer to the average, hence several months to assess the overall performance of these
confirming the increase in cell stability. With the active cells on a long-term basis. The results indicate that a sig-
preventive AE treatment, it is launched in the beginning nificant decrease in the total number of AE is observed. This
phase of LVAE or HVAE. Therefore, the cell can recover outcome is also observed in the anode effect overvoltage
quickly and remains stable, instead of being unstable for which confirms that fewer PFCs are generated from high
plausibly several hours. Additionally, when the cell is voltage anode effect on the cells with the preventive AE
unstable, a resistance increase is introduced into the cell by treatment active.
the control system, consequently increasing the ACD and Moreover, a positive effect was observed regarding the
thus adding some extra energy to the bath. In periods #2, #3 cell stability and for this reason, a smaller amount of extra
and #4, this additional energy input was significantly energy was added to the cell. However, bath production
reduced due to an increased stability. increased with the preventive AE treatment. No negative
The quenching effect of the procedure is designed to effect from the preventive AE treatment was observed
consume part of the anode cover. For this reason, the daily regarding the total number of alumina doses, the total
production of bath is higher with the active preventive AE number of anodic incidents or regarding iron contamination.
treatment. However, when this treatment will be applied to With the positive results obtained in the initial phase of
multiple cells, it will be possible to adjust the crushed bath this project, Aluminerie Alouette plan on stepping up the
and alumina ratio from the anode cover material to account number of cells using the proposed methodology in the near
for the effect of the preventive AE treatments. A correlation future. More cells will allow to correctly evaluate the eco-
can be observed between the daily mass of bath produced nomic gain from such procedure for a whole pot room,
and the parasite alumina feeding indicator. Therefore, even if hence justifying a part of the investment cost necessary to
the number of alumina feedings slightly decreased for some implement continuous anode current monitoring across the
period, it was simply replenished from another source. entire smelter.
Overall, the alumina fed to the electrolysis cell remained
similar to the reference cells, hence we can assume that the Acknowledgments The support from Aluminerie Alouette to publish
metal production was equivalent as well. and present these results is greatly appreciated.
Finally, the quenching of the anode beam could lead to an
increased bath exposure of the anode stubs or to an increase
in the frequency of anodic incidents. However, during our References
tests, no undesirable difference between the reference and
the test cells was observed. Therefore, it indicates that under 1. G. Myhre et al., Anthropogenic and natural radiative forcing, in
Climate Change 2013: The Physical Science Basis. Contribution of
the tested conditions, the preventive AE treatment generates Working Group I to the Fifth Assessment Report of the Intergov-
no negative effect regarding these indicators. It is important ernmental Panel on Climate Change, ed. by T.F. Stocker et al.,
to mention that some data were intentionally excluded from (Cambridge University press: Cambridge, United Kingdom and
this analysis if the iron contamination came from an exces- New-York, NY, USA, 2013)
2. M. Fischedick et al., Mitigation of Climate Change. Contribution of
sive exposure of the stubs that was not a result of the Working Group III to the Fifth Assessment Report of the Intergov-
quenching procedure. Such exposure could be caused by a ernmental Panel on Climate Change (Cambridge University press:
failure of the clad or if the anode detached from its stubs. Cambridge, United Kingdom and New-York, 2014)
Preventive Treatment of Anode Effects Using on-Line … 517

3. D. Wong, A.T. Tabereaux, P. Lavoie, Anode effect phenomena 6. B.J. Welch, Quantifying PFC emissions from smelter cells, in
during conventional AEs, low voltage propagating AEs & Proceedings of the 10th Australasian Aluminium Smelting Confer-
non-propagating AEs, in Light Metals (San Diego, USA, 2014), ence (Launceston, Tasmania, 2011)
pp. 529–535 7. L. Dion et al., On-line monitoring of anode currents to understand
4. International Aluminium Institute, in Workshop on Low Voltage and improve the process control at alouette, in Light Metals
Emissions of PFC, Workshop attendees (London, UK, 2015) (Orlando, USA, 2015), pp. 723–728
5. K.Å. Rye, M. Königsson, I. Solberg, Current redistribution among 8. V. Dassylva-Raymond et al., Modeling the behavior of alumina
individual anode carbons in a Hall-Heroult prebake cell at low agglomerate in the Hall-Héroult process, in Light Metals (San
alumina concentrations, in Light Metals (1998), pp. 241–246 Diego, USA, 2014), pp. 603–608