Hydration-Mechanism-Based Strength Modeling and Binder-Level Inverse Design of Low-Carbon Slag–Fly Ash Ternary Concrete

Reducing carbon emissions in concrete production while improving structural performance has become a priority in the transition to carbon neutrality. High-volume fly ash and slag systems significantly reduce carbon emissions. However, their complex hydration interactions alter strength development, rendering conventional empirical strength models inadequate for reliable low-carbon binder design. To overcome this limitation, this study proposes an integrated hydration-mechanism-based framework that links strength prediction with carbon-oriented binder optimization. A unified hydration model is developed based on the coupled evolution of capillary water and calcium hydroxide, enabling a consistent description of cement hydration, slag latent hydraulic reaction, and fly ash pozzolanic reaction in ternary binder systems. A single set of material-specific kinetic parameters is applied across the investigated mixture proportions and curing ages, without recalibration for each mixture. Building on the predicted degree of reaction, a compressive strength model centered on effective reaction contributions is calibrated using 1030 experimental data points encompassing 3–365 days and 2.33–82.60 MPa. Within the calibration database, the model reproduces early-age strength reduction, later-age compensation, and strength crossover behavior induced by mineral admixtures. The strength model is subsequently embedded into a genetic algorithm framework to perform theoretical binder-level low-carbon inverse design under 28-day strength and composition-domain constraints. The optimization results reveal boundary-dominated solutions, with mineral admixture replacement ratios approaching upper limits and a reduced water content contributing to further calculated emission reduction. Because aggregates, paste volume, superplasticizer dosage, and workability are not included, the resulting binder compositions are theoretical candidates rather than complete concrete mixture designs. This work presents a physically interpretable and optimization-ready framework for binder-level low-carbon design within the adopted model domain.