For weeks, the lab had a crowd of scribbles, diagrams and chemical formulas in the whiteboard. A research team from the Olivity Group and the MIT concrete sustainability hub (CSHUB) were working sharply on a major problem: how can we reduce the amount of cement in concrete to save cost and emissions?
The question was definitely not new; Fly ash, byproduct of coal production, and byproduct of steelmaking, materials such as slag, have long been used to replace some cement in concrete mixtures. However, the demand for these products is moving forward because the industry sees its use to reduce its use by expanding its use, immediately searching for options. The team finds the challenge; He did not lack the candidates; The problem was that there were so many to rt the sort.
On May 17, a team led by Postdock Sorosh Mahjubi published an open access to nature’s cess paper Messing contents Outline their settlement. “We realized that AI is the key to moving,” Mahjubi notes. “There is a lot of data on the potential content – hundreds of thousands of pages! The literature. The sort of them would have taken many lifetime of work, the more content would have been discovered at the time!”
With larger language models delts, as many of us use every day, the team has created a machine-learning framework that evaluates and kind of candidate content based on their physical and chemical properties.
“First, there is a hydraulic reaction. Concrete is strong because the cement – ‘glue’ that keeps it together – hardens when contacted. So, if we replace this glue, we need to make sure that the substitute responds to the same,” Mahjubi explains. “Second, there is posolnisity. This is when a byproduct created by calcium hydroxide, a byproduct created when cement water is found to make the concrete hard and strengthen over time.
By analyzing more than 1 million rock samples. Mahjubi and his team find that the right material is available globally – and, in a more effective way, many can only be included in concrete mixes by just grinding them. This means that it is possible to save and save costs without the extra process.
“Some of the most interesting materials that can replace a piece of cement are ceramics,” Mahjubi notes. “Old tiles, bricks, pottery – all of these materials can have high reactivity. This is what we have observed in ancient Roman concrete, where ceramics were added to help waterproof structures.
The possibility of everyday materials, such as industrial dysfunction materials such as ceramics and mine tailings, is an example of how concrete materials can help enable the circular economy. Otherwise, by identifying and reprodaining the materials that end in landfills, researchers and the industry can help other life as part of our buildings and structural facilities.
Looking forward, the research team is planning to upgrade the framework to enable more content to evaluate, while experimentally recognizing some of the best candidates. “We are excited to see how the AI tools have received this research in a short time, and we are excited to see how the latest developments in the larger -language models are,” says Professor Elsa Olivetty, a senior working author and member of the MIT Department of Materials Science and Engineering. He serves as MIT Climate Project Mission Director, CSHUB Principal Investigator and Leader of Olivate Group.
“Concrete is the spinal cord of a built environment,” says Randolph Kirchen, co-author and CSHUB director. “We hope to support industry efforts to make the industry efforts to make more sustainable, without compromising strength, safety or durability, by applying data science and AI tools in material design.
In addition to Mahjubi, Olivetty and Kirchen, co-authors at work, MIT Postdock Vinith Venugopal, IPEC Bensu Human SM ’21, PhD ’24; And CSHUB Deputy Director Hassam Azarijafri.
