STATISTICAL INVESTIGATION AND MODELING OF BATH LEVEL IN HALL-HÉROULT CELLS

摘要

Industrially, aluminium is produced through the decomposition ofrnalumina (Al_2O_3) in metallurgical reactors known as pots.rnBasically, smelter grade alumina powder has first to be dissolvedrnin an electrolytic bath, which is a molten mixture of cryolite andrnfluoride salts. Dissolved alumina is then dissociated to producernaluminium. Bath level is a leading indicator of great importance tornsmelter operators as it directly affects different key performancernindicators related to current efficiency, energy consumption andrnmetal purity. This paper presents an investigation of the variablesrnhaving an impact on bath level control. It is presented how potrnmanipulated and state variables and other parameters arernstatistically linked to bath levels variations. Moreover, differentrnstatistical models are used in order to attempt a prediction andrnforecast of bath level.