Slashing concrete emissions with AI
Cement production is one of the most polluting industries out there. This industry alone accounts for around 5% of all CO2 emissions globally. If we want to make an outsized contribution in terms of fighting climate change, with minimal effort, improving concrete production and distribution processes is a very low hanging fruit. A report found that investing in reducing emissions from cement production would do more for the climate than even investing in greener transportation technology or greener energy production¹.
The main reason why cement production is so bad for the environment is its very chemistry. In order to make 1,000 kg of cement, 900 kg of CO2 are emitted. This can be seen as the lower achievable limit of emissions. In reality, this chemical reaction only accounts for 50% of the emissions associated with concrete. The rest is mainly generated by burning fossil fuels for powering processing equipment and transporting the materials during their journey, from extraction, to the plant and to the construction site.
Some solutions
If we want to make an impact, there are several options. We could invent a new type of concrete that’s greener. We could also make everything involving concrete greener. Or we could simply reduce our use of concrete.
Although solution one can reduce emissions by up to 50%, we all know it won’t happen, not in thirty years. Solution two can also reduce emissions by 50%, but if we’re able to remove all gas-powered machinery from this world, we’ll have pretty much solved climate change already.
The best way to reduce concrete emissions is to ensure that the concrete is never produced in the first place. Want to cut global emissions by 1%, reduce concrete consumption by 20%!
Everything lies with the consumers of concrete. There are actually very few people who take the decision to consume large amounts of concrete. And you can be damn sure that any real estate developer or government will be happy to save 1% on their construction costs if you can help them identify how to save it. It’s much easier to convince a few thousand people per country to save 1% of their budget on concrete than it is to revolutionize an industry that simply doesn’t care for the planet (the producers). Money talks. If there is one thing we need to understand as environmentalists, it’s that it’s much easier to take the right decision when all other options are more expensive.
There are many ways you can reduce concrete consumption; choosing other materials, reducing waste, repurposing waste… So where does the AI come in? Planning! Planning to create as little waste as possible.
Two studies have tried to estimate the amount of wasted concrete in construction projects. One study found that between 1–13.2% of concrete is wasted on construction sites across the world². The other found that around 3% of concrete is wasted, with 58% of returned concrete associated with over ordering³.
The solution
I propose that we could use AI to solve this issue. By observing concrete orders and utilization, as well as plans for the construction project, we could very precisely estimate how much concrete to order. Nowadays, all plans for buildings are rendered in 3D. Even though construction sites should have a very clear idea of how much concrete to order, they evidently don’t. A lot of factors are at play here. With machine learning, we can discover these relevant variables and make estimates that match reality.
There seem to be two failure modes here, bad volume estimates and bad timing. To avoid bad estimates, more concrete is ordered, which obviously always leads to some concrete being wasted. As far as time management goes, the main problem here is the way construction is carried out. Delays happen, which means some concrete may harden in the truck before it can be poured, leading to it being lost.
I believe that by doing two simple things, we can greatly reduced the amount of concrete consumed. Firstly, we can precisely estimate how much concrete is needed. We should reasonably be able to predict this within much less than 1%. No over-ordering, just precise ordering. If we have a dataset that consists of structural plans, as well as quantity of concrete ordered, and returned, we should easily be able to train some regression model to tell us exactly what’s needed.
The second thing we can do is to better plan the surrounding operations upon which the concrete pours depend. By quantifying exactly how much time certain actions take, for instance setting up rebar and forms, we would be able to order the concrete to arrive at the precise time it will be needed.
The data collection process would be painstaking at first, but I think that it would not take too many example to even get to a good first solution. Materials and techniques are very well standardized in the construction industry. Everything important has been planned by an engineer and there are some traces of that, in the form of plans, estimated time and cost…
The reason I think this idea can have so much impact is that it’s software-based. It can be exported extremely easily. And once the dataset is built, it is not likely to change much, because of the fact that construction techniques evolve so slowly. This concept is infinitely parallelizable.
If we can reduce waste by even just 50%, we can reduce global emissions by roughly 0.1%. Although that may seem like a small number, it is absolutely huge. This amounts to the total emissions of countries like Nicaragua or Albania. All that, just by planning more carefully and being mindful. All that, by only adding one tool to construction planning work. And that tool would only get better with time.
Conclusion
This is a very low hanging fruit. Although there is a very hard limit on the minimum emissions we will ever be able to achieve with a product that essentially emits its own weight in carbon by its chemical process alone, there are ways to optimize its uses and delivery such that we can strive for that very limit. If we want to make an outsized impact for the climate, we can look at an industry that largely operates the same way across the world and that has a very conventional and simple business model. I believe that the concrete industry is the perfect candidate for bringing a large eco-friendly disruption to life.
