Carbon Re's David Boyd considers the role of artificial intelligence in the cement sector's journey to net-zero.
Global Cement (GC): Please can you introduce Carbon Re?
David Boyd (DB): Carbon Re was founded in 2020 to develop the work carried out by Daniel Summerbell during his PhD at the University of Cambridge, as well as researchers at University College London (UCL). They had developed artificial intelligence (AI) algorithms that used large data sets to improve cement pyroprocess efficiency.
The first system that Carbon Re has brought to market is Delta Zero, which uses the latest machine learning techniques to develop and deliver models that optimise cement pyroprocesses. It goes beyond traditional expert control systems by finding new opportunities that cannot be identified by human intuition.
GC: How does it do this?
DB: Delta Zero looks at relationships across the whole range of data available from the preheater, kiln and the cooler. It develops a model of the pyroprocessing system as a whole, rather than a collection of different models that each looks at a one-to-one relationship. The amount of data the system considers is a step up compared to previous approaches.
Another key difference from expert controls is that the models are not fixed until a human adjusts them. This means no drop-off in perfomance, a common issue with expert systems. In contrast, the machine learning algorithm constantly develops and deploys new and improved models based on incoming data. The models optimise themselves.
GC: What benefits can a cement plant expect?
DB: A typical cement plant can reduce its energy use for pyroprocessing by 5 - 10% with Delta Zero. Assuming that the same fuels are being used as prior to implementation, this would mean that its fuel-derived CO2 emissions would also be 5 - 10% lower.
GC: What are the most fertile markets are for AI-led approaches to decarbonisation?
DB: In 2022 Carbon Re commissioned a White Paper with a researcher from the University of Oxford that sought to identify the best opportunities for cement plant decarbonisation, and how AI compares with other technologies. The paper found that Carbon Re's approach, which relies on data already available in the plant, can provide value to cement plants in almost any market. AI models are developed in the cloud and provide optimisations of kiln control parameters. There is no set-up or installation of new sensors at the plant. The only pre-requisite is a reliable high-speed internet connection.
GC: How important will AI be to the over-arching goal of cement sector decarbonisation?
DB: AI is not a cure-all for decarbonisation. However, it is something we can do today in the fight against climate change. This is crucial, because timing is so often overlooked in the race to net-zero. If we really care about global temperature rises, we should decarbonise as much as possible as quickly as possible. Every 1t of CO2 emissions avoided before 2030 is worth 4t of CO2 emissions avoided before 2050, due to the cumulative energy build-up in the atmosphere.
AI is one of the quickest approaches to decarbonise the cement sector at scale, as well as one of the lowest cost approaches. Carbon Re's approach is on the pyroprocess side, but there are similar systems for electrical use, particularly in grinding. There are also systems that can predict the reactivity of clinker - and hence strength development - far more effectively than in the past. Users can then use as little clinker as necessary to limit emissions and avoid over-specification. Combined, each of these relatively small gains, aggregated across a plant, group, or country, could prevent a lot of CO2 from reaching the atmosphere.
Two other levers that producers need to 'max out' right now are alternative fuels (AF) and supplementary cementitious materials (SCMs). There is still so much we can do with existing technology.
The recent advances with calcined clays and Portland limestone cement could easily be replicated across the board. The same is true of AF, although the commercial availability of SCMs and AF is not a given.
GC: How do other decarbonisation approaches fit into this story?
DB: There are so many exciting technologies out there for cement sector decarbonisation that I've started to lose count. There are novel cement chemistries, artificial SCMs, CO2 capture, storage and utilisation (CCUS), (re)-carbonisation, CO2 injection, cement via algae, electrical cement, solar cement... and more. However, these will take time to reach a meaningful scale in terms of deployment. This makes them competitors in the race towards long-term decarbonisation, i.e.: fundamentally weaning ourselves off CO2 emissions from cement and concrete production over the second half of this century. These technologies may each have a place, but for the world of 2050 and beyond, rather than the next 20 years.
GC: Which novel approach - other than AI - is most likely to be scaleable before 2030?
DB: The biggest potential mover for me is graphene. While it is very expensive compared to cement, you only need a very tiny amount to have a profound effect on performance. This allows users to significantly reduce the amount of clinker in cement - and the amount of cement in a concrete structure. There may even be options to eliminate steel reinforcements in some applications, which would be a big win for construction CO2 emissions as a whole. The major concern, again, is scaleability, although graphene production is a well-proven technology, unlike some other potential avenues.
GC: How can policy makers help decarbonisation, especially in markets that are lagging?
DB: Economically, fuel prices are now so high - in part due to the war in Ukraine - that decarbonisation is being driven in the cement sector in every market, regardless of policy. Positive steps that decision makers can take are to develop waste-sorting schemes to facilitate the adoption of AF. Systems can also be developed to encourage the economic use of SCMs. This should include moves away from prescriptive standards for cement and concrete, which reflect a different time in our sector's history and urgently need to be updated. Performance-based standards would open up markets for higher SCM use, including for synthesised SCMs.
GC: What role will CO2 taxes and trading schemes play?
DB: While they have been around for a while, particularly the EU Emissions Trading Scheme (ETS), the effective use of trading schemes and taxes to drive decarbonisation is still some way off. The EU ETS even contains flaws that act against its intended aims. This is because free allowances - which can in theory be re-sold at the market rate (currently Euro85 - 95/t of CO2) are not provided to an individual plant until it makes at least 50% of its rated clinker capacity.
The effect? Plants that would otherwise be mothballed become economic to run, resulting in clinker production in the wrong places. This is why Europe has the highest clinker factor of any world region and why cement exports from the EU have risen since the ETS came into effect. Anyone looking to replicate the EU ETS should bear this in mind if the aim of the game is decarbonisation.
Finally, the statement 'cement is hard-to-abate' is only true to some extent. Overstating this can be unhelpful - a label that is hard to remove. Policy makers, particularly in less CO2-conscious markets, risk giving the sector a 'free pass' by using this term and may lead to a self-defeating mentality that is not compatible with meaningful change.
As your readers will appreciate, there are many ways for producers to decarbonise - both right now and in the future. At Carbon Re, we see AI as one of the best routes to rapid and effective decarbonisation - as well as associated cost savings - wherever your plant is in the world.
GC: Thank you for a very interesting discussion.
DB: You are very welcome. It was great to talk!
About David Boyd
David Boyd is Chief Commercial Officer at Carbon Re, a UK-based developer of AI for cement pyroprocess optimisation. He has more than 20 years of experience in mechanical and civil engineering, the steel sector, digitalisation and mergers and acquisitions.
Carbon Re was founded by Daniel Summerbell (Chief Solutions Officer), Aidan O'Sullivan (Chief Technology Officer), Buffy Price (Chief Operations Officer) and Sherif Elsayed-Ali (Chief Executive Officer). They are supported by an international team of experts in machine learning, statistics, software, engineering and cement production.
