Encyclopedie de la recherche sur l'aluminium au Quebec - Edition 2014 | Page 37

PRODUCTION DE
Multiphase modelingL’ALUMINIUM // ALUMINIUM PRODUCTION
of the
cryolite bath with solidification

35

MODÉLISATION MULTIPHASE DE LA CRYOLITE AVEC SOLIDIFICATION

Modélisation multiphase de la
MULTIPHASE MODELING OF THE CRYOLITE BATH WITH SOLIDIFICATION
cryolite avec solidification
Dave Martin12, Patrice Goulet12, Jean-Loup Robert12,
Donald Ziegler3, and Mario Fafard12
1Department

2NSERC/Alcoa

of Civil Engineering, 1065 avenue de la Médecine Université Laval, Québec, QC, G1V 0A6, Canada
Industrial Research Chair MACE3 and Aluminium Research Centre – REGAL Université Laval, Québec, QC, G1V 0A6, Canada
3Alcoa Primary Metals, Alcoa Technical Center, 100 Technical Drive, Alcoa Center, PA, 15069-0001, USA

Interface

Introduction
Phase change phenomena in the cryolite bath have
an important effect on anode changes and ledge
thickness which mainly impacts on energy
efficiency and cell life span. A better understanding
of the processes will help increase performance and
lower operating costs. However, the harsh
environment prohibits detailed experiments and
numerical simulations were an evident alternative to
circumvent these difficulties.

The main challenge of this project is managing
the interface. To account for change in behaviour
at the interface, the XFEM technique is used. It
allows us to maintain a sharp interface without
remeshing.

Level Set solution

Preliminary Results: Testing the Method
The following results show the solidification of water in a corner. The triangular
elements shown are used to generate the integration scheme and display the
results properly.

Problematic and objectives
A numerical model must include the heat transfer
and fluid dynamics in the cell and the electrolyte’s
phase change. The important density variation
(~30%, whereas water is ~8%) introduces special
challenges. The excess mass as the liquid solidifies
acts as an flux boundary for the fluid phase.
Furthermore, the phase change is driven by the
jump in heat flux (q) across the boundary. These
jumps in the solution require special numerical
attention.

Heat and mass transfer at interface

Theory

Analytical solution

Here is the flow of water around a cylinder in a channel (no mass flux on lower and
upper boundaries). A pressure differential at each end drives the flow.

Conclusion
• Phase change model reproduces cryolite physics
• Viable tool in predicting cell behaviour
• Discontinuities can move in time without remeshing → shorter solve time
• Error controllable using numerical tools

Future work

Dave Martin
Patrice Goulet
Jean-Loup Robert
Mario Fafard
Département de génie civil,
Chaire de recherche
industrielle
CRSNG/Alcoa MACE3,
Centre de recherche
sur l’aluminium - REGAL,
Université Laval
Donald Ziegler
Alcoa Technical Center,
Alcoa Primary Metals

•Complete 2D tests
•Add mass flux boundary condition (fluid flow)
•Couple phase change and fluid flow problems

Journée des étudiants – REGAL process uses a very corrosive electrolytic bath at about 960°C.
The Hall-Héroult

Le procédé Hall-Héroult utilise un bain de cryolite très corrosif à une température
18 novembre sur
d’environ 960 
°C. Lors du changement d’anode, une couche solide se forme 2014
l’anode froide. De plus, une couche solide est maintenue sur les parois de la cuve
pour la protéger. L’épaisseur de ces couches solides modifie considérablement
l’efficacité de la réaction. Prédire ces phénomènes à partir d’un modèle numérique
est donc un élément clé afin de réduire les coûts d’exploitation de l’usine. Par
contre, les discontinuités physiques associées au changement de phase posent
des défis pour la simulation numérique. La méthode par éléments finis étendus
(XFEM) a été exploitée pour surmonter ces difficultés.

During anode changeout, a solid layer forms on the cool anode. Furthermore, a
solid layer of electrolytic bath is maintained on the walls of the cell to protect
it. The thickness of these solid layers significantly alters the efficiency of the
chemical reaction. Predicting these phenomena from a numerical model is
therefore a key element in lowering the operating costs of the plant. However,
the physical discontinuities associated with the phase change involve important
numerical challenges. An extended finite element method (XFEM) was used to
overcome these difficulties.