Skip to Content

Green concrete: Using AI to reduce concrete’s carbon footprint

  • In collaboration with researchers at the University of Illinois at Urbana-Champaign, we have developed a new AI model that optimizes concrete mixtures for sustainability as well as strength. 
  • In early field testing, carbon emission was reduced by 40 percent, while strength requirements were exceeded. 
  • Cement in concrete accounts for approximately 8 percent of carbon emissions globally. If successful, the impact of this work could reach well beyond data center construction, as it applies more broadly to the general construction industry.

Recommended Reading

Data centers power the internet and make it possible for people to connect with the communities that are central to their lives. However, the construction and operation of these data centers can also be a source of greenhouse gas emissions. Last year, Meta achieved net zero emissions in our operations, and we are now turning our attention to our ambitious goal to achieve net zero emissions across our value chain by the end of 2030. This commitment includes addressing the indirect environmental impacts of our business, such as from the embodied carbon found in our buildings. Embodied carbon includes the upstream and downstream emissions from the manufacturing, transportation, maintenance, replacement, and decommissioning of building materials. Among these building materials, concrete is a major contributor to the embodied carbon of our data centers.  

Concrete is the most widely used engineered material in the world, with billions of tons produced annually. Cement in concrete is a major source of carbon emissions globally, accounting for 8 percent of carbon emissions. Thus, reducing the emissions from concrete will have an impact that reaches well beyond data center construction. We have found a way to do this by using artificial intelligence (AI) to generate a concrete mix design that reduces the use of cement — without compromising strength requirements.

Concrete is a complex mixture of dozens to hundreds of ingredients. Conventional concrete is typically a mix of cement, construction aggregate, water, and performance-enhancing chemicals known as admixtures. There are many possible materials that can be used as aggregate, including gravel, crushed stone, sand, slag, recycled concrete, and geosynthetic aggregates of different types and sizes. Since cement is the most carbon-intensive ingredient in concrete, one potential way to reduce the carbon footprint of concrete is to reduce the amount of cement and/or replace it with other cementitious materials like fly ash, slag, and ground glass, which have a lower carbon intensity. 

But getting the proportions of cement substitutes right is tricky when we are also trying to adhere to stringent technical performance requirements. Manually optimizing a concrete formula for both sustainable and technically sound outcomes for different end uses is a formidable challenge, one that can be time-intensive. Mathematically, each of the potentially hundreds of possible ingredients is an added dimension in a complex space of all possible concrete formulas. To efficiently and quickly optimize concrete for sustainable performance, and to generate new concrete formulas, we’ve turned to AI. 

Accelerating concrete formula discovery with AI

AI can be used to learn and optimize for specific outcomes within high dimensional spaces, where data sets with many attributes can be modeled. When a valid data set is available, AI can be used to estimate or “learn” the feasible high-dimensional space in terms of relevant parameters, such as strength and sustainability. To develop this type of model, we leveraged the work of researchers at the University of Illinois at Urbana-Champaign, including AI expert Prof. Lav Varshney from the electrical and computer engineering department, and concrete expert Prof. Nishant Garg from the civil engineering department. 

Varshney and Garg trained a model using the Concrete Compressive Strength data set, openly available from the UCI Machine Learning Repository. This database has 1,030 instances of concrete formulas and their validated attributes, including seven-day and 28-day compressive strength data (i.e., how the concrete gained strength seven days and 28 days after pouring). The embodied carbon footprint associated with the concrete formulas was derived using the Cement Sustainability Initiative’s Environmental Product Declaration (EPD) tool. EPDs are a standardized way of accounting for the environmental impacts of a product or material, including carbon emissions over its life cycle. 

Using the input data on concrete formulas along with their corresponding compressive strength and carbon footprint, the AI model was able to generate a number of promising new concrete mixes that could meet our stated data center requirements with a lower embodied carbon impact than the industry standard.

UIUC researchers perform a slump test to measure the consistency and workability of one of the AI-generated formulas.

That model generated a list of potential formulas, from which we selected the five best options. We then further refined those to be used in the real world. We leveraged an iterative process: The potential formulas were tested in the lab, refined, and retested. This process was repeated until we identified formulas that could be handled and poured easily, while also exceeding requirements for seven-day and 28-day strength with a 40 percent reduction in emissions. Manually searching for concrete mixes that simultaneously optimize for multiple attributes, like strength and sustainability, can be challenging, as technical performance has traditionally been prioritized in the concrete industry, and there is a general perception that optimizing for sustainability may adversely impact performance. In the absence of AI resources, discovering similarly good, sustainable formulas that also meet technical performance criteria could require several time-intensive iterations. With AI, we were able to accelerate the discovery process and validate good formulas within weeks. In this effort, the low-carbon concrete formulas generated by the model entailed significant replacement of cement (upwards of 70 percent) with a combination of two types of low-carbon substitutes, namely fly ash and slag.

Step 1: Researchers vet a database of concrete formulas and performance characteristics.
Step 2: Experts use the AI model to generate new concrete formulas, optimizing for attributes including compressive strength and carbon emissions.
Step 3: The AI-generated formulas are mixed at the UIUC lab so that their properties can be assessed.
Step 4: The AI-generated formulas undergo slump testing at the UIUC lab as part of their initial performance assessment and refinement.
Step 5: The compressive strength of the refined AI-generated formulas is tested.
Step 6: The concrete supplier, Ozinga, inspects the refined formulas, and makes adjustments based on material availability, environmental conditions like temperature and wind, and additional performance considerations.
Step 7: Ozinga validates the final formula in their lab.
Step 8: The low-carbon concrete is poured at the DeKalb data center location in the construction personnel office space.
Step 9: Further developments and tests are planned in order to scale the impact of this innovation to other concrete applications at the data center campus.
1 of 9

Bringing the innovation to the field

We wanted to test the formulas in the field, and selected our data center in DeKalb, Illinois, as an ideal location, given its proximity to the University of Illinois at Urbana-Champaign. The resulting formulas were shared with the concrete supplier for our DeKalb data center, Ozinga, which further refined them based on:

  1. Ozinga’s expertise as an industry practitioner
  2. Local material availability 
  3. The need to enable adequate strength gain under the cold weather conditions expected at our data center location

After the final formula was tested in Ozinga’s labs, we agreed to test them on multiple (noncritical) structures at our DeKalb data center, namely the guardhouse floor slab and the floor slab of the construction management team’s temporary office space building. This refined concrete formula entailed upwards of 50 percent cement replacement with a combination of fly ash and slag.

The new concrete formula was tested in the floor slab of a guardhouse, an ancillary building at the DeKalb data center campus. Left: Concrete being poured in the floor slab. Right: Finished guardhouse

The field tests confirmed that the resulting low-emission concrete formula exceeded the seven-day and 28-day strength requirements, with a carbon impact 40 percent lower than the regional benchmark. However, to enable faster, more efficient construction required for our priorities, we need the concrete to meet specific strength thresholds earlier than seven days. We are working to improve the early strength performance of the concrete at the three- to five-day marks, and account for the impact that variations in environmental conditions, such as temperature and wind, can have on concrete performance. Depending on material availability, there is significant potential for the general building industry to adopt the use of concrete mixes similar to the ones tested in our DeKalb data center, particularly for construction projects that aren’t subject to compressed construction schedules.

An Ozinga ready-mix concrete truck transferring concrete into a pump hopper.

Next steps

With the help of this AI model, we’ve been able to successfully design and use concrete that meets our long-term strength requirements and has a 40 percent lower carbon impact than the regional benchmark. But we are just getting started. While we are encouraged by the success of this pilot, further developments and tests are needed to scale the impact of this innovation. This includes a number of factors: 

  1. First, we need to understand and optimize the performance of the mixes under different weather conditions, such as the cold weather in Illinois. Each data center location has unique attributes that must be accounted for in the design and testing of concrete. 
  2. There are improvements to be made to optimize for early strength gain (i.e., to have it set faster) to accommodate a tighter construction schedule. There is an opportunity to directly optimize for such logistical considerations. 
  3. Finally, there is a need to understand how novel materials could be used in lieu of cement in concrete, particularly as the supply of traditional cement substitutes like fly ash and slag is projected to decrease in the long run. Material availability is an important factor in construction; identifying novel materials that could be used in concrete can open new opportunities. Additionally, novel materials can further enhance the strength and sustainability of new concrete formulas. 

The resulting concrete mixes from our model can be used outside of data center construction, and there is an opportunity to further develop this model to address other use cases. Our exploration of innovative solutions to reduce data center construction emissions is not limited to concrete. There are opportunities to reduce the emissions of other materials. We are also exploring innovative data center designs as another way to improve sustainability. 

For more information about this work, please read our paper for ACM COMPASS 2022

To help personalize content, tailor and measure ads, and provide a safer experience, we use cookies. By clicking or navigating the site, you agree to allow our collection of information on and off Facebook through cookies. Learn more, including about available controls: Cookies Policy