The Challenge
Reducing raw material costs without risking quality
Gerdau is one of the world’s largest producers of long steel, special steel, and iron ore, with mills in 10 countries and over 30,000 employees. The company makes made-to-order special steel products for the most demanding applications in the automotive, commercial vehicle, agricultural, construction, and energy markets.
After decades of running a strong process, they, like many other steel firms, faced a distinctly modern challenge: with raw material prices skyrocketing, they had to lower costs quickly in order to remain competitive, without risking the high-quality steel their customers expected.
Using traditional process improvement tools, Gerdau’s engineers had already been able to optimize production as much as possible. Yet these optimizations were conservative to ensure high quality production under all circumstances.
With multiple sources of variation affecting production, including scrap quality and rolling mill temperatures, raw material limits had been carefully calculated to avoid even the slightest chance of failure. To further reduce costs, the team needed a tool that could adapt in real time to shifting conditions on the factory floor.
"We're running a lean process--we have much less room for error than ever before. We need to reduce raw material costs, while delivering the quality customers expect. By combining process knowledge with Fero Lab's ML tool, we not only optimized our alloy usage and saved costs--we also reduced the workload on the process engineers and made huge product capability improvements."
Jena Kreuzer, process engineer, Gerdau
The Solution
Live optimization in the melt shop
Since the team was already used to optimizing processes with Six Sigma, using Fero felt like a natural extension of those analyses. Using the Fero software, engineers at a Gerdau plant were able to create a digital twin of their unique process in minutes, without needing to write a single line of code.
Fero’s white-box machine learning models then determined the optimal amount of raw materials for each heat and delivered these recommendations to operators in real time. Using Fero’s recommendations, the team was able to save $3 per ton of steel in raw material costs and reduce quality variation by 15%.
Before building this live recommendation model, the engineers wanted to start out by establishing that the model had a basic understanding of their process. In order to do this, they uploaded their historical alloy addition data into Fero and built a static model to investigate the relationships between melt shop alloy additions and end-of-line mechanical properties.
The white-box ML model reinforced their understanding of the process, as well as providing them with a new tool to investigate what was happening in the mill in a new way and assess what next steps they could take. Quantifying uncertainty around predictions and input factors allowed them to gain trust in the analysis quickly and understand where they could draw clear conclusions as opposed to general abstractions.
Once the Gerdau engineers had a ML model they trusted, they tested its reliability by simulating changes in rolling mill parameters. This allowed them to evaluate potential regrading scenarios and realize value from what would have otherwise been scrapped heats.
With a black-box solution, they would only have been able to make simple predictions. But with Fero Labs, they were able to see exactly what would happen in every scenario thanks to the software’s underlying white-box machine learning technology.
To realize the full potential of Fero, Gerdau connected it to their live data and had ladle chemistry samples sent to the software in real time. Fero was able to analyze those samples and recommend the lowest-cost alloy addition that would guarantee quality, in mere seconds.
Crucially, Fero Labs was able to give actionable insights in the short but critical time window when the operators could still alter the chemistry: all they had to do was look at a screen in the control room and they’d find the Fero recommendation for that heat.
The result
$3/ton in savings...and a more efficient process
With Fero Labs showing operators the minimum amount of alloy they needed to add to each heat in real time to meet quality specifications, the Gerdau team has been able to save $3 per ton of steel in raw material costs. This number has only increased as a larger share of production uses the new workflow, gaining trust in the insights of white-box machine learning.
Moreover, by reducing raw material consumption, Gerdau saw the added benefit of lowering its emissions footprint.
Reducing unnecessary alloy additions through Fero eliminated the need for mining and refining 500,000 pounds of raw material. If the software were employed for additional use cases, such as scrap rate reduction or furnace oxygen intake optimization, or galvanizing line energy minimization, the steel giant could see even more sustainability improvements hand in hand with increased profitability.
Not only has Fero’s machine learning solution helped save costs and emissions, it also improved quality consistency. Once operators began adjusting the chemistry on a heat-by-heat basis, based on the software’s recommendations, the team observed a 15% reduction in quality variation.
This finding attests to the increased flexibility of the operations team and their newly proactive approach to the process. The team is no longer in “firefighter mode,” but rather adapting seamlessly to raw material fluctuations and other natural sources of variability within the steel process, which has had a positive impact on many other aspects of their operations management.
Gerdau’s success highlights that a streamlined operation, equipped with the appropriate tools, can lead to improved product quality and significant cost savings.
Since 2018, Gerdau has expanded the use of Fero Labs to five other plants in North America and now leverages the tool as a scalable solution for other complex process challenges beyond raw material reduction, such as minimizing defects, improving energy efficiency, and enhancing production yield.
This proactive approach ensures that Gerdau remains at the forefront of innovation and drives long-term success for the company and its customers.
