Economic volatility, environmental concerns, and increased regulations are critical challenges faced by global aluminum producers. Improvement in aluminum smelters demands efficient and intelligent use of information available within the smelter. Customer satisfaction, quality and cost are key. Improvement in these key elements ensures protection against market pressures and allows high quality products to be produced. By introducing digital technologies in smelters, a digital manufacturing roadmap will deliver a significant leap in productivity, provide deep insights into the production process and unleash new revenue sources. Aluminum prices have flourished in all the metal exchanges globally over the last few months, providing an excellent opportunity to transform traditional smelting operations into digitally enabled operations.
IOT and Digital use cases across aluminum smelters
Artificial neural networks and machine learning
Aluminum reduction cells often encounter substandard performance that can result in a dramatic increase in energy consumption, and increase the emissions of fluorocarbon gases. Artificial neural network and machine learning can facilitate in understanding and identifying the noise patterns like metal pad roll, and short circuiting sound, which in turn proactively notify operators to take appropriate action.
Another significant use is estimating the alumina concentration and excess Alf3 % in the cells. The process comprises of performing variance analysis and applying analytics, which in turn helps in estimating parameters precisely and in detailed diagnosis (neural network has two steps for this use case-variance analysis and analytics. First, variance analysis is performed and then analytics is used to determine and predict the possible outcomes).
A mathematical model of the process is used to model a system based on the relationship between input and output. The difference between the actual parameters and that of the modelled parameters are calculated and named as residuals. These residuals are further evaluated to estimate the optimum process parameters.
Now, prescriptive analytics is applied by using optimization and simulation algorithms to estimate the precise process parameters. It can be used to quantify the effect on reduction cells by varying the alumina concentration and excess Alf3 %. It can assist in recommending the possible outcomes before the decisions are actually made. It not only predicts what will happen to the cell temperature, but also provides insights on why it will happen and prescribes for optimum cell temperature and excess Alf3 %.
Image analysis of cover alumina, and anode assemblies
Alumina and crushed bath are mixed together and supplied to pot-tending-machines (PTM) or to alumina or crushed bath hoppers. This composition plays a significant role in maintaining a good crust integrity, ensuring good productivity and low gas emissions. Higher gas emissions can lead to improper heat balance in the pot resulting in abnormalities. Image analysis can be used to identify the proper composition and detect uneven mixing that may cause quality issues. It can also facilitate in keeping stringent control over the heat balance in the pots, which will eventually drive the increase in current efficiency and high quality molten metal.
The anode assembly is often inspected manually. The use of digital image analysis can save costs in the manual process and provide benefits such as identifying defective anodes, and determining whether the reduction cells and anode production parameters require adjustment based on the anode characteristics data.