special - ALUMINIUM-Nachrichten – ALU-WEB.DE
special - ALUMINIUM-Nachrichten – ALU-WEB.DE
special - ALUMINIUM-Nachrichten – ALU-WEB.DE
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
SPECIAL<br />
<strong><strong>ALU</strong>MINIUM</strong> SMELTING INDUSTRY<br />
Theory<br />
The theory is described in a more detailed<br />
manner in reference [4]. Let us briefly mention<br />
that the alumina distribution in the bath<br />
is determined by the following partial differential<br />
equation:<br />
c: Alumina concentration in the bath<br />
D: Alumina diffusion coefficient<br />
u: Bath velocity field generated by MHD<br />
Lorentz force and by the release of bubbles<br />
If we consider fluctuations around the stationary<br />
state:<br />
(2)<br />
The velocity field is generated by Lorentz<br />
force field and by the bubbles. When the<br />
number of bubbles produced, per m 2 and per<br />
second, is too large, we cannot use a numerical<br />
approach describing the motion of each<br />
bubble separately.<br />
There are essentially two standard ways to<br />
overcome this difficulty. The first consists of<br />
performing some kind of averaging over the<br />
equations and over the corresponding fields.<br />
The second bypasses the averaging and directly<br />
postulates the flow equations for each<br />
phase.<br />
One of the main difficulties encountered<br />
when performing an averaging process is related<br />
to the possible jumps that fields can suffer<br />
at the boundaries between the two phases.<br />
One way to overcome this problem consists<br />
in extending the domain of definition of each<br />
motion equation to the domain occupied by<br />
the two phases. This is achieved by multiplying<br />
each equation by the characteristic function<br />
corresponding to its domain of definition.<br />
Derivatives are then performed in the sense<br />
of distributions which allows us to keep track<br />
of these discontinuities in the averaging process.<br />
Whatever choice we make, the resulting<br />
equations will contain terms which reflect<br />
the interaction between the two phases. The<br />
exact shapes of these terms are not known;<br />
they have to be defined through constitutive<br />
equations.<br />
(3)<br />
Taking into account the first law of Fick, eq.<br />
3 becomes: in Fig. 3. From the two figures, the alumina<br />
concentration field appears as only slightly<br />
modified by the velocity field. However, when<br />
An estimation shows that D turb is of the order<br />
of 0.2 m 2 /s.<br />
Industrial cell<br />
(4)<br />
The problem has been solved for a 180 kA cell<br />
using two point feeders. On the feeders, the<br />
alumina concentration is set to 5% of the bath<br />
weight. When presenting a stationary solution,<br />
this assumes continuous feeding. But we can<br />
easily analyse the<br />
impact of dump<br />
feeding. Fig. 1 corresponds<br />
to the<br />
stationary alumina<br />
distribution,<br />
when the velocity<br />
is neglected. The<br />
concentration is<br />
shown under the<br />
anodes. The two<br />
feeder locations<br />
appear clearly in<br />
the figure. The<br />
asymmetry of the<br />
diffusion pattern<br />
reflects the larger<br />
channel width at<br />
the feeders. A difference<br />
of close<br />
to 2.5% alumina<br />
concentration can<br />
be observed at<br />
the surface of the<br />
anodes. The vertical<br />
variation of<br />
alumina is 0.5%<br />
under the feeders,<br />
but it is negligible<br />
away from<br />
the feeders.<br />
Fig. 2 shows<br />
the velocity field<br />
generated by<br />
the bubbles and<br />
Lorentz force in<br />
this particular<br />
cell.<br />
The impact of<br />
the velocity field<br />
on the stationary<br />
solution of the<br />
alumina concentration<br />
is shown<br />
considering the concentration evolution, the<br />
time needed for reaching the stationary state<br />
is reduced by a factor 2 when the velocity field<br />
is acting. Therefore the velocity field plays an<br />
important role in the feeding process (alumina<br />
dumps).<br />
To highlight the role of the velocity field,<br />
Fig. 4 shows the difference between the alumina<br />
concentration field due only to the diffusion<br />
compared with that in presence of the<br />
velocity field. The greatest differences are observed<br />
at the ends of the cell, due essentially<br />
to the MHD effects. High negative values re-<br />
Fig. 1: Alumina concentration at the anode surface assuming no velocity field<br />
Fig. 2: Velocity field in the bath of the cell<br />
Fig. 3: Alumina concentration at the anode surface in presence of the velocity field<br />
Fig. 4: Alumina concentration variation due to the velocity field<br />
© Kan-nak<br />
<strong><strong>ALU</strong>MINIUM</strong> · 1-2/2013 59