Machine Learning Model Optimizes Concrete Strength and CO2 Uptake in Carbon-Cured Cement
bioengineer.orgResearchers developed a machine learning framework using Least Squares Boosting to predict compressive strength and carbon dioxide uptake in CO2-cured concrete. The model was trained on 423 strength and 131 uptake datasets from 15 studies, achieving correlation coefficients above 0.95 on test data. It identified curing gas concentration, humidity, and exposure time as key drivers for strength, while binder chemistry governed uptake potential. Monte Carlo simulations suggested that curing for 48 to 72 hours with CO2 concentrations above 10% and relative humidity above 60% can yield up to 38.8% CO2 uptake in certain mixes. This approach could help reduce cement's carbon footprint by enabling more effective carbonation curing, which stores CO2 in concrete while accelerating strength gain.
