Development and predictive modeling of eco-friendly SCC incorporating supplementary cementitious materials

Purpose This study aims to investigate the development of eco-friendly self-compacting concrete (SCC) by partial replacement of Portland cement (PC) with crushed dune sand (CDS) and ground granulated blast-furnace slag (GGBFS). The effects of these materials on the fresh and hardened concrete properties are evaluated, and the optimal mixture providing a balanced combination of performance and sustainability is identified. Design/methodology/approach The experimental design consisted of 15 SCC mixtures based on a three-factor mixture-design methodology. The fresh properties included slump flow, T500 flow time, L-box ratio and segregation resistance, while the hardened properties comprised 7 and 28-day compressive strength and 28-day porosity. The experimental data were analyzed using Design-Expert 13 to develop predictive statistical models describing the relationships between mixture proportions and SCC performance. The developed models were evaluated using analysis of variance (ANOVA) and performance indicators, including R2, RMSE, MAE, SI and MAPE. A multi-objective optimization procedure based on the desirability function was then used to determine the optimum SCC mixture. Unlike recent machine-learning-based approaches proposed for predicting the performance of advanced cementitious materials and SCC, the present study adopts an interpretable mixture-design-based statistical framework for prediction and optimization. Findings The developed predictive models showed high reliability, with R2 values ranging from 89.36% to 95.42%, and were statistically significant according to ANOVA (p < 0.05). Validation results demonstrated high prediction accuracy, with low RMSE, MAE and SI values and MAPE values mostly below 3%; the close agreement between predicted and measured responses confirmed the applicability and reliability of the models within the investigated mixture-design domain. Increasing PC content improved matrix compactness, thereby increasing compressive strength and reducing porosity, whereas increasing GGBFS and CDS enhanced sustainability while maintaining acceptable SCC performance, despite a slight reduction in early-age strength. SCC14 (5% CDS) was identified as the optimal mixture, while SCC7 (10% GGBFS + 5% CDS) provided an effective alternative with reduced cement consumption. Originality/value This study provides an assessment of the combined use of GGBFS and CDS as alternative materials for SCC production. Beyond experimental characterization, it demonstrates the effectiveness of predictive statistical modeling and multi-objective optimization for identifying sustainable, economical and high-performance SCC mixtures. The proposed framework provides an interpretable and experimentally efficient approach for predicting SCC performance and optimizing mixture proportions, with experimental validation supporting its applicability within the investigated mixture-design domain and complementing recent machine-learning-based methods reported for advanced cementitious materials.