top of page
Search

This Startup Is Using AI To Reduce Emissions In Hard-To-Mitigate Industries

The best sustainability measures are the ones that pay for themselves.


That’s exactly the kind of work that’s the specialty of Fero Labs, a New York City-based manufacturing process optimization software company. Today they’ve announced the close of a $15 million growth round led by Climate Investment, bringing Fero Lab’s total funding to $30 million.


The company is developing AI-based software that helps identify ways to make industrial processes more efficient and therefore greener thanks to lower emissions, with a specific focus on manufacturing sectors that are difficult to decarbonize, such as steel, cement and chemicals. With wholesale decarbonization solutions such as complete process electrification and carbon capture and storage proving slow to materialize at industrial scale, Fero Labs is focused on ways to reduce those manufacturers’ emissions right now.


“If you think about it, the goal of reducing waste has been part of the manufacturing ethos for almost 70, 80 years,” said Berk Birand, co-founder and CEO of Fero Labs. “It goes back to the Toyota Production system, Lean, Six Sigma. Being in the tech world, (we’ve) borrowed a lot of things from manufacturing like Agile, Kanban and all these things. So now it’s like the climate world is borrowing more things from the manufacturing sector.”


Birand and his co-founders are focused on the next big idea for manufacturing waste reduction. Pooling their combined schooling and industry backgrounds in engineering, computer science, machine learning, and industrial strategy, they chose to concentrate on reducing manufacturing inefficiencies. “We estimate that factories, on average, consume about 9% more in resources than they absolutely need to to run their production,” Birand explained. “What do I mean by resource here? Resource could mean raw material consumption… Or it could be in terms of quality issues causing scrap… It could be in terms of processing time. All plants consume a lot more resources than they need to.”


Those are fairly typical examples of the kinds of wastes seen in industrial processes. What isn’t typical is the Fero Labs approach to identifying and reducing or eliminating those wastes. The company’s software employs AI- and machine learning-based software that interacts directly with the processes themselves, using the regular production data to ferret out the opportunities for improvement. And does it with a white-box model. “The biggest bottleneck in the adoption of AI in the industrial setting is in fact trust,” said Birand. “What we do very differently, and what has been the key to our success, is we use a specific type of machine learning which is explainable by nature. Our model, our software, is not a black box. Engineers can actually look at these models that have been trained, and then open them up and then look inside, see what the model has learned. They can almost have a conversation with these models.”

Fero Labs claims its software has already uncovered upwards of $20 million in savings for its customers and reduced more than 100,000 tons of industrial carbon emissions. One of their believers is Gerdau, a Brazilian global steel producer. The company used Fero Labs software to look at alloy optimization, a portion of its business that represents between 10% and 20% of its manufacturing costs. “Alloys are a very important aspect of what we do, in terms of how much can we save on our alloys, how much can we optimize our typical recipes,” said Carlo Travaglini, Director of Technology at Gerdau. “And that was the task. The vast majority of our steel is sold not based on specific chemistry, but based on what the chemistry is delivering in terms of mechanical yield and tensile strength. We did build a pretty strong model predicting those factors, which are key for us. The model is actually telling us what is the most cost-optimized chemistry that is sufficient to achieve and exceed our limits without overshooting.” Carlo is particularly impressed with the model’s ability to deliver actionable data in real time, and to deliver it effectively directly to the operators who are running the process.


The founders’ future plans for Fero Labs include both growing the business geographically–most of their operations are currently in North America–and by focusing on additional sectors of industry. “The industrial sector is very complex,” said Birand. “AI can dramatically have an impact on reducing emissions. We can also do this while reducing costs at the same time. One of the four goals of Fero is that, by reducing raw materials consumption, we’re reducing Scope 3 emissions. But we’re also reducing how much money they’re paying for raw materials… If you’re processing in a batch, and you process it for one hour less, that’s great for your Scope 2 emissions, but it also means that you can produce more batches with the same amount of time. Factories can not only reduce emissions, but also do this while also reducing costs.”


Follow me on LinkedIn. Check out my website

0 views0 comments

Comments


bottom of page