About this Abstract |
| Meeting |
2026 TMS Annual Meeting & Exhibition
|
| Symposium
|
Aluminum Reduction Technology
|
| Presentation Title |
Leveraging the Power of Flexible Discriminant Analysis (FDA) to minimize Decision-making uncertainty when Spotting Sidelining True Risk to Failure? |
| Author(s) |
Joseph Ndjebayi Nloga, Philippe Theberge |
| On-Site Speaker (Planned) |
Joseph Ndjebayi Nloga |
| Abstract Scope |
Pot life is a key element of both productivity and strategic planning at RTA smelters. Capital and operational expenditures depend heavily on the cathode longevity. Since pot relining makes up over 50% of the CAPEX, it requires strategic planning and oversight. Both the annual capacity and the associated expenses limit the cell relining process, and deciding which cell to take out usually includes uncertainty and may lead to wasted cathode life. Cell failure can occur through three mechanisms, including cathode problems, short-term temperature drift, and sidewall premature erosion. Although some companies have developed cell replacement strategies based on the former failure mode, the unpredictability of the latter mechanism could cause many smelters to make the wrong cell choices, taking into consideration the number of concurrent occurrences. Could supervised learning algorithms, like Flexible Discriminant Analysis (FDA), improve decision-making when spotting sidelining true risk to failure? |
| Proceedings Inclusion? |
Planned: Light Metals |
| Keywords |
Process Technology, Aluminum, Other |