Concrete A I(nnovation)

Concrete mix technology is demanding and, all too often, thankless. We ask our people to guarantee the structural integrity of the design and minimize costs despite the host of bad things that can happen with mix components and transit. When things go right, our folks are invisible. Yet when things go wrong, they are the first to bear the brunt of the problem.

Now help is on the horizon via a new crop of applications based on machine learning, commonly referred to as artificial intelligence (AI). The premise is to use the vast history of mixes and material tests to optimize for cost or carbon dioxide (CO2)—and in many cases, both. Let’s take a closer look at one of the offerings, Concrete-AI.

Mathieu Bauchy, Ph.D., and Gaurav Sant, Ph.D., are professors at the University of California, Los Angeles (UCLA) Samueli School of Engineering and Applied Science. Several years ago, they started working with the Federal Highway Administration (FHWA) and a consortium of construction materials suppliers and concrete producers to find ways to reduce concrete’s carbon footprint through increased use of alternatives to portland cement. The project was successful, and as a follow-up, they channeled their experience and knowledge of artificial intelligence into a new company dedicated to the optimization of concrete mixes: Concrete-AI. 

Concrete-AI is working with several concrete producers across North America and has analyzed more than 100,000 unique mixes with their associated material tests. For each mix, the Concrete-AI engine attributed CO2 and cost to each component or grouping of materials at the plant and then “learned” the behavior of the mix as a function of its proportions. The engine now can take desired properties, such as strength and slump, and suggest the best mix design to minimize cost or CO2.

BUT DOES IT WORK? 
Initial testing done with the five supporting producers indicates Concrete-AI’s suggested mixes perform within +/- 500 psi. Admittedly, more validation needs to be done, but all initial indications are positive.

Bauchy says Concrete-AI technology tackles three primary applications. First, it can optimize current mixes. Second, as mix components and their material properties change, mixes can be adjusted. Third, it can greatly shorten the time required to specify and validate project-specific mixes.

To be clear, Bauchy emphasizes that Concrete-AI is an “accelerant” for mix technologists and quality control/quality assurance (QC/QA) teams. They will still need to test and possibly tweak the mix for final acceptance. However, it should significantly improve the first try. He further reports that the not entirely unexpected consequence is that, in many cases, both cost and CO2 are minimized.

WHAT ARE THE ECONOMICS? 
We must run profitable businesses to stay in business, period. If we add up all the money various sales folks tell us we could save by using their product, collectively, they would be paying us to produce concrete! Intuitively, Bauchy does not want to fall into this trap and is commendably cautious about making financial predictions.

He notes that while times will change, right now there is very little direct financial incentive to reduce carbon footprints in the U.S. Thus, the burden of adoption remains focused on cost reduction, and to this point Concrete-AI is targeting north of $1 per cubic yard. While these are early days, the target is in sight with the frequently added benefit of carbon reduction.

Consider for a moment what would happen if global producers adopted a similar model offered by Concrete-AI. Bauchy predicts widespread adoption would reduce carbon emissions by an amount equivalent to what France currently produces. Based on both the company’s economic and carbon footprint progress, it makes sense to keep close tabs on Concrete-AI.

AI DRIVES MIX DESIGN PERFORMANCE, CO2 PROFILES
ConcreteAI applies pioneering manufacturing and plant automation technologies to concrete production, quality control and embodied carbon management. Artificial intelligence and machine learning algorithms analyze existing and newly logged data to predict engineering performance indicators with increasing accuracy and certainty. With recurring aggregate, binder and admixture material inputs, for example, the “performance prediction” module calculates the expected slump, shrinkage and strength development (one- to 90-day) properties. 

The technology emerged from a UCLA Samueli School of Engineering effort whose members also participated in the development of a carbon dioxide mineralization and sequestration method: CarbonBuilt. That project is on its own commercialization path via an independent entity, CarbonBuilt Inc., which was the Grand Prize Winner of the NRG COSIA Carbon XPRIZE global competition that ran from 2016 until 2021.

COST VS. CARBON OPTIMIZATION
Producers and practitioners mindful of embodied carbon in finished slabs and structures can use Concrete-AI technology to craft mix designs light on portland cement but ultimately reaching or exceeding target compressive strength. Alongside an “embodied carbon” module, the technology was launched with “cost optimization” and “mix generator” modules.


Craig Yeack has held leadership positions with both construction materials producers and software providers. He is co-founder of BCMI Corp. (the Bulk Construction Materials Initiative), which is dedicated to reinventing the construction materials business with modern mobile and cloud-based tools. His Tech Talk column—named best column by the Construction Media Alliance in 2018—focuses on concise, actionable ideas to improve financial performance for ready-mix producers. He can be reached at [email protected].