The incorporation of digital twins in construction enables progress towards more sustainable buildings through data‑driven material design aimed at reducing environmental impacts. The precise digitalisation of concrete components, combined with deep learning models, makes it possible to virtually reproduce the behaviour of the material and anticipate its compressive strength without relying on physical tests. This approach streamlines mix optimisation and reduces unnecessary use of resources and energy, key factors for improving the sustainability of the built environment.

The predictions obtained with deep neural networks capture the non‑linear interactions between cement, additions, water and admixtures, making it possible to clearly identify which variables determine concrete performance. This capability opens the door to reducing cement content  —the component with the highest carbon footprint— and replacing it with industrial byproducts with lower environmental impact. By simulating different compositions, it is possible to develop high‑performance mixes that enhance building durability and reduce energy and material consumption throughout their life cycle.

Thanks to these digital twins, concrete design becomes a more efficient, reproducible process aligned with the sector’s decarbonisation goals, facilitating the creation of construction solutions that support more resilient, durable buildings with a lower environmental footprint.