Encyclopedie de la recherche sur l'aluminium au Quebec - Edition 2014 | Page 19

PRODUCTION DE L’ALUMINIUM // ALUMINIUM PRODUCTION

Real-time control of the anodes quality
using acoustic methods

17

CONTRÔLE EN TEMPS RÉEL DE LA QUALITÉ DES ANODES PAR MÉTHODES ACOUSTIQUES

Contrôle en temps réel de la qualité des anodes
par méthodes acoustiques
REAL-TIME CONTROL OF THE ANODES QUALITY USING ACOUSTIC METHODS

Moez Ben Boubaker1, Carl Duchesne1, Jayson Tessier2, Houshang Alamdari3 et Mario Fafard4
1

Département de Génie Chimique, Université Laval, Québec, QC, G1V 0A6, Canada
Canada Primary Metals, Aluminerie de Deschambault, 1 Boulevard des Sources,
Deschambault-Grondines, QC, G0A 1S0, Canada
3 Département de Génie des mines, de la métallurgie et des matériaux, Université Laval, Québec, QC, G1V 0A6, Canada
4 Département de Génie civil et de génie des eaux, Université Laval, Québec, QC, G1V 0A6, Canada
2 Alcoa

1 : Introduction

2 : Problem & objectives
Impact on reduction cells:




Fig. 2: Anode degradation
Fig. 1: Defects

Electrical resistivity
Carbon consumption
Set cycle
.

 Develop a rapid and non-destructive method to inspect 100% of
anodes produced :
 Intercept faulty green anodes before baking (save carbon, baking
and energy consumption).

How to intercept faulty anode
before they deteriorate
process performance and
increase production costs

 Intercept faulty baked anodes before set in reduction cells
(save carbon and energy consumption).

3 : Methodology

Fig. 5: Example cut of an anode slice

Fig. 3: Schematic of the methodology

Fig. 4: Sliced baked anode scanned using X-ray tomography
Fig. 6: Example X-ray image of a slice

S.1 Acquiring ultrasonic and acoustic emission signals through anode material.
S.2 Preprocessing and storage of signals.
S.3 Relevant features sensitive to various types of defects will be extracted from acoustic signals using
various methods.
S.4 Acoustic features will be analyzed using multivariate statistical methods to classifying them
according the defect types or to correlated them with anode properties (e.g. densities).

Fig. 9: Acoustic data acquisition through a small block

Fig. 7: Data acquisition through different slices

Fig. 8: Acoustic data acquisition through a separated corridor

S.5 Results validation.

4 : Results

Fig. 10: Selection of corresponding X-Ray images

Fig. 11: Separation of scan position images

Fig.12: Example of image processing of one scan position

Fig. 13: Classification using discrete wavelet decomposition

Moez Ben Boubaker
Carl Duchesne
Département de
génie chimique,
Université Laval
Houshang Alamdari
Département de génie des
mines, de la métallurgie et des
matériaux, Université Laval

Fig. 14: Sequential excitation with several frequencies

Fig. 16: Regression model between X-ray
image features and those of acoustic
signals
Fig.17: The influence of defects on the classification of corridors

Fig.18: The influence of defects on the classification of blocks

Fig. 15: A single excitation with suitable frequency sweep

5 : Future work
• Build a larger acoustic setup for application on full scale anodes.
• Perform control tests over multiple anodes and track them in electrolytic cells using voltage
differences and their final states using butts images.

6 : Acknowledgement

Jayson Tessier
Alcoa Canada,
Alum