Skip to content

How AI Helps Accomplish Green Steel Goals

By: Fero Labs Logo light
Adobe Stock 528204115

Steel represents 7% of global emissions—more than any  other industrial sector. Around 75% of steel is still made in high-emitting coal-fired blast furnaces. If the world is to meet its 2050 climate goals, emissions from steel must be halved and then continue falling, according to the International Energy Agency (IEA).

Enter: green steel. Global Google searches for green steel have quintupled in the past 20 years. By 2032 the market size is projected to reach $6.24B, driven both by steel producers' net-zero goals and end users' sustainability priorities.

“Green steel is becoming incredibly important,” Nucor CEO Leon Topalian told attendees at the Global Steel Dynamics Forum in New York this past June, highlighting market demand for “cleaner and cleaner steels” and “a true net-zero product.”

Nucor launched its green steel offering, Econiq, in 2021 and delivered its first Econiq coil to General Motors in 2022. Over the next year, Topalian said, the company expects to supply more than a million tons, primarily to automotive and higher-end manufacturers.

The automotive sector represents around 12% of global steel demand. Anticipating customer demand and a competitive edge in the near future, automotive manufacturers have been inking green steel deals left and right. Volvo recently committed to using 100% green steel by 2050; similarly, Ford pledged to use 10% green steel by 2030 as part of the World Economic Forum’s First Mover coalition. Overseas, Swedish truck maker Scania recently announced a partnership with H2 Green Steel for “sustainably produced steel.”

AI and green steel

Green steel is typically defined as steel produced without the use of fossil fuels, such as by switching to hydrogen-powered or electric arc furnaces (EAF). These heat sources reflect a major step forward in emissions reduction. However, they don’t address sources of indirect emissions, such as those resulting from steel companies using mined components in their raw ingredients.

These emissions—classified as Scope 3—are both the hardest to measure and the most pernicious. Green EAF plants in the US produce nearly as many Scope 3 emissions as their traditional counterparts, despite having reduced their carbon footprint from the furnaces themselves. As this type of emissions must be carefully monitored and tracked throughout the manufacturing and supply chain process, reducing them is more complicated than reducing direct emissions from a furnace, and typically not on companies' radar.

That's where artificial intelligence and machine learning (AI/ML) can come in. By using AI/ML methods to make production more efficient, steelmakers—whether they operate via blast or EAF furnace—can reduce the amount of raw materials they use, thus reducing indirect (Scope 3) emissions.

AI and ML solutions make production more efficient by using historical data to determine the necessary amount of raw material for a given process. In this way, producers can avoid using more than they need, without compromising quality. Once the models have been trained on enough historical data, they can work dynamically, adapting to each heat in real time. For example, as alloys are added to steel, these solutions can calculate the minimum amount of each alloy needed to achieve the target strength of the finished product. This method can save substantial indirect emissions, adding to the direct emissions saved by switching to greener production methods.

AI and ML can also make steel production more efficient—and “greener”—by minimizing the amount of energy needed for a given process. Traditionally, operators err on the side of caution and run processes to a point where they can be certain of avoiding failure. However, AI/ML algorithms can shorten process times by figuring out how long operators actually need to run the process in order to achieve the desired quality.

The focus on quality is understandable. When steel quality doesn't match the requirements, the heat must be scrapped—which also consumes significant energy. With AI/ML ensuring that every heat matches quality standards, producers can minimize scrap and stabilize production to minimize rework, which also lowers emissions and ensures that the final product is truly green steel.

Today, the world produces about 450 MT of steel using EAFs. If scaled globally, AI and ML solutions could mitigate at least 40 MT CO2e, the equivalent of the fossil fuel usage of Portugal. This would benefit the steel industry by making their steel even more appealing to companies such as automotive firms that are trying to use more environmentally friendly steel to attract consumers. As green steel increases in consumer demand, we can expect indirect emissions to become a bigger part of the discussion in the years ahead.