CONCLUSIONS
Numerical algorithms that perform cell voltage noise removal and cell voltage (or
resistance) slope calculation have been compared.
High frequency voltage noise can be successfully removed using quite simple
averaging and curve fitting techniques.
A more complex and more CPU demanding parabolic curve fitting scheme
produced a noisier and less accurate cell voltage slope estimation than a simpler straight
line fitting scheme.
Next step would be to test the cell voltage noise removal algorithm "on-line" as
part of a continuous tracking alumina feeding control algorithm in a dynamic cell
simulator.
REFERENCES
(1) BEARNE, G., 1999.
The development of aluminium reduction cell process control, JOM, 51 (5) (1999),
pp. 16-22.
(2) BONNY, P. AND AL, 1984.
Process and apparatus for accurately controlling the rate of introduction and the
content of alumina is an igneous electrolysis tank in the production of aluminium, US
patent number: 4,431,491.
(3) LEROY, M., 1987.
Process for accurately maintaining a low alumina content in an electrolytic smelting
cell for the production of aluminum, US patent number: 4,654,129.
(4) MACAUDIERE, Y., 1988.
Recent advances in process control of the potline, TMS, p. 607-612.
(5) ROBILLIARD, K. R. AND ROLOFS, B., 1989.
A demand feed strategy for aluminium electrolysis cells, TMS, pp. 269-273.
(6) TRACY, G. P., 2001.
Insights into parameters affecting current efficiency, Proceeding of the 7
th
Australasian Technology Workshop.
Process simulation, TMS Course on Industrial Aluminum Electrolysis.