background image
MATHEMATICAL MODELING OF ALUMINIUM REDUCTION CELL POTSHELLS:
IMPROVEMENTS AND APPLICATIONS

Marc Dupuis, Jonquière
In its 2010 TMS conference paper [1], the author presented three types of ANSYS® based thermo-chimio-
mechanical potshell models, namely the "empty shell", the "almost empty shell" and the "half empty shell"
potshell models. All three types of models take into account the thermal loading coming from the thermal expansion
of the potshell steel structure itself considering the thermal gradients present in the steel structure and the internal
pressure coming from the cell lining expansion inside the potshell. The model versions presented in [1] were strain
forward redevelopment of the work the author presented quite some time ago in [2,3,4 and 5].
Since then, all three kind of models have been improved taking advantage of the contact elements facilities available
in ANSYS® 12.0. Those improved model versions will be presented here altogether with two applications. The first
application is the test of a potshell retrofit design aiming at eliminating the vertical deflection of monocoque
potshells. The second application is the test of the potshell and lining retrofit design the author proposed in [6] to
expend potlife of modern high amperage cells using graphitized cathodes blocks up to 3500 days.
Improved "almost empty shell" potshell model
The "almost empty shell" potshell model have been improved by decoupling the 2D potshell mesh from the 3D side
lining mesh and by reconnecting the two parts using ANSYS® CONTA174 and TARGE170 contact pair elements.
After the decoupling, it is possible to completely refine the 2D potshell mesh, this was not possible before the
decoupling. This is important because it was already demonstrated in [7] for the "empty shell" potshell model that
the initial thermo-electric model mesh [8,9] is too coarse to carry out accurate thermo-mechanical analysis, hence
the possibility to increase the potshell mesh refinement is a significant model improvement.
Figure 1 presents the resulting displacement solution, which is not significantly different from the one presented in
[1]. Apart for the possibility to further refine the 2D potshell mesh, a second significant improvement is the added
possibility to extract from the solution the pressure that the side lining is applying on the potshell through the
contact interface (see figure 2).
Improved "empty shell" potshell model
As discussed in [1], the main weakness of the "empty shell" potshell model type is the fact that the internal pressure
load has to be defined by the modeler as a boundary condition and that the modeler can only rely on semi-empirical
loading schemes established from measurement campaigns to do so. Now, with the improvement of the "almost
empty shell" potshell model, a new possibility has became available, it is now possible to apply as boundary
conditions to the "empty shell" model the contact interface pressure distribution extracted from the "almost empty
shell" model solution (see figure 3).
As we can see in figure 4, the resulting improved "empty shell" model displacement solution is quite different from
the one presented in [1] and is now quite similar to the one presented in figure 1. This clearly demonstrates that the
semi-empirical loading scheme that the author knew and used in [1] is quite different from the pressure distribution
presented in figure 2.
Testing a new potshell design aiming at eliminating the vertical potshell displacement with the improved
"empty shell" model
As discussed in [7,10 and 11], the vertical displacement of very long high amperage cell monocoque
potshells has a negative impact on the cell operation. For quite some years now, the author had a potshell
retrofit design idea aiming at preventing that vertical potshell displacement, the improved "empty shell"