Detection of abnormal alumina feed rate in aluminium electrolysis cells using state and parameter estimation

Abstract The concentration of dissolved alumina in the electrolyte is one factor influencing the current efficiency in aluminium electrolysis cells. Too low concentration might lead to serious operational problems, like anode effect. In this paper measurements of the current through anodes in different parts of the cell are used to estimate states and parameters using an extended Kalman Filter. Faulty alumina feed rate causing abnormal alumina distribution might then be detected by examining how the expected model output differs from the state estimate from the Kalman Filter in combination with drift in the estimated parameter values.