AI-driven Concrete Mix-Design

Concrete is the most widely used construction material in the world, and cement production remains one of the major industrial sources of CO₂ emissions. Since cement strongly influences both the cost and the environmental impact of a mix, even modest reductions in cement content can lead to significant savings when applied at the scale of concrete production.

This project applies machine learning to concrete mix design by combining mixture-composition data with environmental information from production sites, including weather-related variables. The models estimate key performance-related parameters, such as compressive strength and workability, directly from the proportions of the constituents and the surrounding production conditions.

By replacing part of the traditional trial-and-error process with fast, data-driven predictions, the approach supports a more efficient exploration of the mix design space. This makes it possible to identify mixtures that meet target mechanical and workability requirements while reducing unnecessary cement use, lowering cost, and decreasing embodied carbon.