Asset Reliability and performance improvement of Aluminium smelting



Challenge Summary

To find out the empirical relationship between Aluminium Fluoride (AlF3) fed & Aluminium Fluoride (AlF3) set on the basis of available pot-room data, to reduce time in computing feed of Aluminium Fluoride to augment manpower, increase the stability of the extraction system, prevent pot leakages, reduce the usage of soda bags.

Challenge Scenario

In the process of Aluminium smelting through electrolysis Aluminium Fluoride (AlF3) is to be fed into the pots as per prescribed quantity in each of the pots. The existing control logic implemented leads to excessive or very deficient feed of Aluminium Fluoride to the pots leading to very adverse effects such as pot leakage which incurs heavy costs to repair, higher or lower bath temperature, lower pot service life, excessive use of soda due to excessive feed of Aluminium Fluoride. A single manufacturing unit contains anywhere between 600 to 1500 individual pots with individual controllers within separate rooms each holding up to 100 pots. For each pot the Aluminium Fluoride (AlF3) feed calculation is to be done and set in the controller based on measurements made on parameters associated with the pot such as Pot Bath temperature trend, Pot Noise, Pot age, Alumina dump, Excess AlF3, Cathode type, CaF2 content for each pot. An autonomous intelligent control system in place is required to find the empirical relationship between feed and set which can augment manpower, and increase the reliability of the system.

Profile of the End-User

Aluminium Extraction process manager :

Existing method : A process manager is assigned to a room, in a shift of 8 hours almost up to 5 hours is spent on calculation of the feed for Aluminium Fluoride based on past recordings of the parameters of the pot room for each pot and then setting the correct amount of feed rate to the pot. All the data recorded from each pot is collected in a MES system via cloud, then using necessary software tools the regression analysis is done with manual intervention for each pot based on data recorded for the past 3 months, 6 months and 1 year.

Gaps : A process manager is assigned to a room, in a shift of 8 hours almost up to 5 hours is spent on calculation of the feed for Aluminium Fluoride based on past recordings of the parameters of the pot room for each pot and then setting the correct amount of feed rate to the pot. All the data recorded from each pot is collected in a MES system via cloud, then using necessary software tools the regression analysis is done with manual intervention for each pot based on data recorded for the past 3 months, 6 months and 1 year.

Functional Requirements of the End-User

  • To find out the empirical relationship between Aluminium Fluoride (AlF3) fed & Aluminium Fluoride (AlF3) set on the basis of available pot-room data.
  • To reduce the standard deviation in Aluminium Fluoride (AlF3) fed & Aluminium Fluoride (AlF3) set in pot-controller.

Functional & Operational Capabilities

  • A self updating regression algorithm for daily changing operational data.
  • Predict the correct amount of AlF3 to be fed to the pot based on the operational data.
  • Calculate the current efficiency

Operational constraints

  • Different Cathode types used in the process of extraction.
  • Availability of networking infrastructure
  • Interface with SAP and storage system used for recording data
  • Integration of hardware if any with existing controller system.

Expected Tangible Benefits and Measurable Gains

  • Reduction in Alf3 consumption in pot.
  • Improvement in pot stability.
  • Prevention of possible pot leakages.
  • Reduction in consumption of soda bags.
  • Complete process automation possibility , reduction in time consumed for computing feed for each pot according to the parameters measured.

Performance Metrics or Outcomes

  • Accuracy of prediction of AlF3 to be fed into the pot.
  • Reduction in pot leakages and instability instances of the system.
  • Increase in current efficiency
  • Increase in productivity greater than 60 %

​​​