Meta uses AI to design eco-friendly concrete • The Register
Facebook parent company Meta has a new project rooted in the physical realm: using artificial intelligence to discover new formulas for green concrete.
Concrete is the foundation of modern construction, but the manufacturing process is a huge source of carbon emissions. The manufacture of cement, one of the main ingredients of concrete, is responsible for around 8% of global carbon emissions, which would make it the largest industrial emitter. And that’s just the cement – lots of other things with their own trace of carbon emissions go into the concrete.
Meta, which is currently working on eight data center projects, is pouring a lot of concrete.
This makes it harder for the company to keep working towards its sustainability goals, like eliminating carbon emissions from the supply chain by 2030. Greening concrete isn’t easy either, said Amruta Sudhalkar, Meta Sustainability Program Manager.
“Manually optimizing a concrete formula for results that are both durable and technically sound…is a formidable challenge,” Sudhalkar said. Factors that go into making good concrete, especially one that tries to replace its cement content, mean not only considering ingredient ratios, but also external factors such as weather and location.
If trying to discover specific formulas from vast pools of potential ingredients sounds familiar, that’s because it’s a lot like drug discovery, where machine learning excels. Using machine learning to develop new concrete formulations that reduce carbon emissions and retain strength is seen as a natural extension of this technology.
Meta teamed up with University of Illinois at Urbana-Champaign professors Lav Varshney and Nishant Garg, of electrical and computer engineering and civil engineering, respectively, to build an AI model to do just that.
The AI was trained on concrete compressive strength test data that included formulas and considered cure times, and the carbon footprints for the different formulas were derived from the Environmental Product Declaration Tool of the Cement Sustainability Initiative.
The research team selected five potential candidates for real-world testing, each “involving significant replacement of cement (more than 70%) with a combination of two types of low-carbon substitutes, namely ashes flywheels and slag,” Sudhalkar said.
Ozinga, Meta’s concrete supplier for the DeKalb project, poured the AI-formulated concrete as the foundation for an on-site guardhouse and temporary office space for the construction management team. According to Meta, the test concrete resulted in 40% lower carbon emissions than regional benchmarks and also proved stronger in seven- and 28-day strength tests.
However, not everything went perfectly. While the new concrete was strong given the time, it still wasn’t hardening fast enough to meet Meta’s construction priorities for the site. “We strive to improve the initial strength performance of the concrete after three to five days and to take into account the impact that variations in environmental conditions, such as temperature and wind, can have on the performance of the concrete. concrete,” the company said.
While Meta works to address these issues, the company also noted that it views its Green Concrete as usable across different industries and for different use cases, and is particularly well suited for situations without the time constraints that require faster hardening.
To that end, Meta has published its work on the project and will present its findings at the Association for Computing Machinery’s Compass conference in late June. ®