perturbation will have a much larger impact on the thermal response of the system. Since
exact measurements of these parameters were no longer possible, we had to estimate these
parameters.
perfectly matched the real cell measured thermal response, as can be seen in Figure 4.
response of the pot were seen:
a) Without power no more heat generation in the anode, cathode and bath:
pot so that the heat convection on the outside of the shell is almost unchanged.
Inside the pot, no more heat is generated in the anode and no more hot process gas is
emitted. The still-running exhaust gas extraction system starts to cool down the covering
and anodes.
The stirring effect of the magnet field also breaks down, resulting in massive changes in
the heat transfer into the ledge.
At locations with a high metal and bath speed and a well-established ledge profile, the
speed-dependent heat transfer coefficient drops and less heat is conducted into the side /
end pier. The reduced heat flux results in a growing ledge thickness. This mostly occurs at
the centre of pot sides and ends.
At locations with low flow velocity, e.g. sludgy zones or stagnant areas, the changed
metal pad and slowed bath and metal movement can increase the heat transfer, resulting in
an increased heat flux into the lining. The effect can be seen often at the pot corners.
heat-transfer model running in parallel, these effects can be quantified and incorporated into
the global behaviour in the lump model.
This results in a drop in bath temperature and superheat, an increased concentration of excess
AlF3 through ledge formation and, depending on total bath acidity, reduced alumina
solubility, as shown in Figure 5a-c.