Towards Low-Carbon Cement: Clinker Reduction in Blended Systems through Experimental Investigation, Physico-Chemical Analysis, and AI-Based Predictive Modeling of Mechanical Performance

The reduction of clinker content in cement production is a key strategy for decarbonizing the construction industry, as it significantly lowers CO₂ emissions while promoting the use of supplementary cementitious materials (SCMs) such as fly ash and natural pozzolans. Nevertheless, clinker substitution alters hydration kinetics, microstructure development, and early-age mechanical behavior, which may raise concerns regarding durability and structural performance.

This study investigates three cement formulations: a clinker-rich reference system, a fly ash-based blended cement (SC1), and a hybrid fly ash–natural pozzolan system (SC2). The experimental program focuses on the evaluation of compressive strength at 2 and 28 days in order to assess both early-age and long-term performance. The obtained results are further interpreted through a physico-chemical approach, highlighting the role of hydration mechanisms, including C–S–H formation and pozzolanic reactions, in strength development.

In addition, a data-driven modeling framework is developed to improve predictive capability. Machine learning techniques are employed to estimate compressive strength as a function of clinker content and mixture composition. The proposed model demonstrates satisfactory predictive accuracy and enables a better understanding of the relationships between formulation parameters and mechanical performance.

The findings reveal a clear trade-off between reduced early-age strength and enhanced long-term performance. Overall, this work provides valuable insights and practical guidelines for optimizing low-clinker, low-carbon cement formulations while maintaining adequate mechanical properties.