Cheuk-Yi Cheung

Frequency response analysis of anode current signals as a diagnostic aid for detecting approaching anode effects in aluminum smelting cells

Modern cell control aims to prevent anode effects by controlling the alumina feeding rate based on the change in cell resistance or voltage and the preset limits of these values. Success of this approach depends on the uniform distribution of dissolved alumina across the cell and the anode current distribution. As this is not always the case in practice, the control procedure sometimes fails and the cell undergoes anode effect. Monitoring of the anode current signals has been suggested as an alternative way for early anode effect detection. This paper presents frequency response analysis of anode current signals obtained from an operating cell and shows the ability for early detection of an anode effect. It has been found that the frequency response peak associated with bubble dynamics of the corresponding anode disappears as it undergoes partial anode effect prior to the cell approaching full anode effect. The results show that the analysis can provide further information to identify a localized anode effect which can facilitate cell control for more effective anode effect prevention.

Monitoring Local Alumina Dissolution in Aluminum Reduction Cells Using State Estimation

In the Hall-Heroult process, alumina dissolution rate is dependent on a number of process variables. One major variable is the superheat, especially in modern reduction cells which operate at low cell voltage. During the cell operation, routine practices and abnormalities give rise to energy imbalance at different parts of the cell, and consequently local superheat variations. This leads to variations in local alumina dissolution rate, and affects process efficiency and performance. This paper presents a method for monitoring local cell conditions by estimating the rate constant of alumina dissolution and alumina concentration using the extended Kalman filter. The method employs a combined mass balance and cell voltage model, which estimates the effective local superheat-dependent dissolution rate constant and local alumina concentration from individual anode current, cell voltage, alumina feed rate and anode-cathode distance. The results are verified with alumina concentration and superheat measured during an experiment conducted in an industrial cell.

Detection of Local Cell Conditions Based on Individual Anode Current Measurements

The application of individual anode current measurements in the Hall-Heroult cells has been investigated to aid cell monitoring, in addition to the conventional use of cell voltage measurements. Its advantages are significant, especially in high amperage cells, where information from the voltage signal is heavily diluted. One common use of individual anode current measurements is to identify process faults that can cause a significant change in the local current flow paths between anode busbar and the metal pad. While this detection is simple and direct, there are other consequential changes in cell conditions that impact the current distribution in the vicinity of a problem anode. This paper presents a Moving Window Kernel PCA-based method to extract spatial information from the individual anode current signals by incorporating other known process variables.

Impacts of Anode Set on the Energy Re-distribution of PB Aluminum Smelting Cells

Since the introduction of prebaked anodes technology in Hall-Heroult process, anode setting has become one of the routine work practices. As all anodes in different parts of the cell are changed in turn at short regular intervals, the operation is always subject to a different degree of disturbances. This paper presents a dynamic thermal model that can be used to simulate the impact of anode setting on the local thermal balance and hence the overall operating condition by incorporating individual anode current signals as model inputs. This is done by discretizing the bath into multiple subsystems based on the position of each anode. The model can predict the local thermal conditions during the increase of current pick up of a newly replaced anode when based on online measurements of current distribution. A model incorporated with anode current distribution as model inputs can be employed as a foundation for future development of an online fault diagnostic system to help isolating other disturbances, hence improve early detection of impending abnormal conditions.