Page 2

IN SITU

ALUMINIUM CELL CONTROL

A new approach to controlling alumina feeding in a H.H. cell, called In Situ

aluminium cell control has been recently proposed [1, 2 and 3]. The core principle is
quite simple: it turned out that it is possible to extract the values of both the dissolved
alumina concentration and the ACD at the end of a 5 to 10 minute no feed observation
period by using only the standard cell amperage and voltage data.

Once this is done, it is possible for an extended period of time to use a PID

controller to regulate the rate of alumina feeding using only the normalized cell voltage
to calculate the error between the estimated dissolved alumina concentration and a target
value. After an extended period of 8 to 12 hours, a new no feed observation period must
be performed in order to reestablish the value of both process variables. The detailed
procedure is described below.

Calculation of the normalized cell voltage

The first step is a calculation of the normalized cell voltage using raw sampled

cell amperage and voltage data points in order to remove the effect of the line amperage
fluctuations on the cell voltage fluctuations. Calculating normalized cell voltage is an
alternative to calculating cell pseudo-resistance. This variable can then be used to assess
the change of cell voltage/resistance due to changes in cell bath chemistry and ACD
which are independent of the cell amperage.

= ( V BEMF ) / I * I

+ BEMF

(1)

Where:

V

is the normalized cell voltage mostly free of fluctuation due to the amperage

fluctuation (V)

is the raw fluctuating cell voltage (V)

is the raw fluctuating cell amperage (kA)

is the nominal or target cell amperage (kA)

BEMF

is the extrapolated voltage at zero amperage usually set to 1.65 (V)

The Dyna/Marc cell simulator can be used to illustrate the efficacy of this

procedure. It can be set up to produce a noisy cell amperage but also a cell resistance
without MHD and bubble release fluctuations. In this way raw cell voltage noise is only
dependant on the amperage noise as shown in Figure 1.

As already demonstrated in [7], employing an incorrect value for the BEMF will

lead to the calculation of a noisy cell pseudo-resistance as can be seen in Figure 2, but
also of a cell pseudo-resistance having a significantly different value. As we can see in
Figure 3, using an incorrect value for the BEMF will lead to the calculation of a noisy