Concrete Proof¶

Client: Gideon Farrell and Oliver Greet, Converge.

Concrete is everywhere: it helped build the Roman world and enables today’s skyscrapers and infrastructure. It is also a huge decarbonisation challenge, with enormous potential for better, lower-carbon mixes. A concrete project specification can contain dozens of requirements describing mixes needed for different parts of a building: strength, durability, exposure conditions, cement and binder types, water/binder ratios and more. These requirements are often buried in lengthy PDF documents, while concrete suppliers may have databases of potential mixes and their measured properties. Your challenge is to build a prototype that can read a specification, work out what each mix needs to satisfy, and search a supplied database for suitable candidates. The hard part is making sense of messy technical language and turning it into structured constraints that a computer can reason about. How much can be automated, and how should a user check or correct what the system has understood? You could explore document parsing, natural-language processing or AI, constraint satisfaction, multi-objective ranking, and visualisation. We will provide example specifications, concrete mix data and relevant standards, leaving you free to design the approach. The goal is a decision-support tool helping engineers, suppliers and contractors get from specification to suitable concrete mixes.

Background reading:

• What Pompeii reveals about ancient Roman building technology (MIT News)
• Making concrete change: the cement sector and carbon emissions (Chatham House)