Unlocking the Potential of Digital Transformation in the Metals Industry: Insights from the Digital Transformation Forum 2023
30 Mar 2023 - Production, Technology, Industry 4.0
During AIST's Digital Transformation Forum 2023, in Pittsburgh, PSI Metals gave a presentation on improving production quality with advanced analytics. Together with other sponsors and presenters, they discussed relevant topics of digital transformation in the steel industry. The dominant subject was the use of data-driven AI technology.
AIST (Association for Iron & Steel Technology) is a non-profit organization with 15,500 members from more than 70 countries with a clear mission: to advance the technical development, production, processing and application of iron and steel. With this in mind, they regularly organize events related to the steel industry, such as the Digital Transformation Forum. It provides a great opportunity to share knowledge, acquire technical information, network and collaborate.
In addition to technical presentations, this year's Digital Transformation Forum featured two panels, one for producers and one for vendors. Highlights of the event included the following topics:
- change management
- knowledge transfer
- the meaning of digital transformation
- the future of the steel industry
A Grand Return: Presenting a Use Case for Improving Production Quality With Advanced Analytics
Luc Van Nerom, Innovation Manager at PSI Metals, stated during his presentation:
It is amazing how hungry the steel industry is for data-driven AI technology. Most, if not all, US steel producers are experimenting with machine learning technology.
With AI being the proclaimed future of digital transformation in the steel industry, this event was dominated by advanced presentations of research collaborations and use cases of machine learning and machine teaching technology.
Some steel producers in the U.S. presented how they are using machine learning and machine teaching to improve welding quality, classify strip breaks, create workflows to reduce production risks and cost, and many more solutions.
PSI Metals also made a presentation titled “Improving Aluminum Quality through Advanced Analytics.” Although it was an aluminum use case, aluminum production is similar to steel production and the same AI defect detection and prediction methods can be applied in steel. The aluminum industry has made significant improvements in digital transformation, making them a great example to emulate.
PSI Metals' presentation was on using machine learning, with a focus on automated machine learning (AutoML), to detect and predict defects based on historical production and defect data. The objective is to support defect definition. The presentation further explained the use case of a data-driven prescriptive pipeline, including data curation and statistical analysis as well as how unified data model was used.
Advanced analytics is being increasingly used in the area of quality control and optimization. In this use case, a machine learning prescriptive approach was applied, with the underlying predictive model wrapped within an outer algorithmic layer oriented to provide the "prescriptions", e.g. the remedies for the predicted defect occurrences. The internal machine learning part is designed within a data-driven approach, where a representative amount of historical data is processed and fed into the predictive methodology, resulting in a high-quality statistical model.
On the other hand, the outer layer provides optimization capabilities, where the space of chosen process variables influences, for example, the production quality. The task of the optimization layer is to exploit the predictive potential of the machine learning model and to correct the current production setting so that certain objectives, like avoiding the defect occurrence, are achieved. Generally, an algorithm is expected to cope with numerous criteria, often conflicting with each other. An example could be productivity maximization vs. the minimization of defect occurrences.
However, the machine learning is trained on the historical data set and thus, incorporates the picture of various production regimes as completely as possible.
This entire workflow was analyzed through the process data collected on a production line at the Turkish aluminum plant, ASAŞ. The performance has been promising, including the explanation of how the profitability analysis turns the predictive model measures into business value. The presentation was received with great interest and numerous questions were asked about the potential uses of this technology in the steel industry.
What Digital Transformation Means to Us: Discussing the Journey and Challenges
Following the panels, there were open discussions which allowed individuals to explain what digital transformation meant to them. Some viewed digital transformation as a goal to achieve or what was required to achieve their actual goal.
However, there are certain issues and challenges, which surround achieving this transformation, such as budget approval for digital transformation projects and change management.
Many people believed that budget approval for digital transformation projects depends on whether the company's leadership prioritizes financials or technological advancement. The second identified challenge is convincing experts in the older generation to participate in the creation of new methods that come with digital transformation.
The producer panel, which included representatives from Tata Steel Europe, ArcelorMittal Global R&D, Steel Dynamics, Commercial Metals Company and ATI Specialty Rolled Products, further discussed how they build trust between the older generation of workers and AI, as well as how to reassure them that AI will not replace them. The panelists encourage them to participate and share their knowledge and experience, as it is required to achieve success with these digital transformation projects.
The panelists also jointly expressed their enthusiasm towards the increase of younger workforce. They further acknowledged the extensive information and training available in engineering and IT which better prepare younger generations to join the steel industry’s workforce. Jeffery Parks, Director of Automation at ATI Specialty Rolled Products reiterated the importance of focusing on people and explained that:
Technology is challenging but getting people to work together is even more challenging.
Looking to the Future: The Road to AISTech 2023
PSI Metals Sales Directors, Scott Wilson and Javier Nadal attended the event. Scott showed his appreciation for this gathering organized by AIST and said:
Another great AIST event, bringing bright minds together to discuss the latest advancements in technology that can be utilized for the betterment of the steel industry!
PSI Metals' presentation was well received which shows the openness of the industry to build a quality expert machine learning workbench and to commodify AI. This would assist in building trust between product managers and AI.
The upcoming AISTech 2023 is another opportunity to discuss solutions and progress between companies, both those producing steel and those providing services for steel production.
Chidi Sybil Aku
Marketing and Document Manager
PSI Metals GmbH
Chidi Aku is the Marketing Manager for North America at PSI Metals. She joined the company in August 2021 and is working on campaigns to increase PSI Metals North America Inc.’s presence in unclaimed parts of the market. She is also in charge of document management for sales documents such as annexes and proposals.
Follow Us on