Performance-driven mix design and multi-response optimization of ambient-cured fly ash-GGBFS based geopolymer concrete

Fly ash (FA) and ground granulated blast furnace slag (GGBFS) are widely investigated as alternative precursor materials for alkali-activated geopolymer concrete (GPC). However, systematic mixture proportioning is required to achieve desirable fresh and mechanical performance under ambient-curing conditions. This study presents a performance-driven optimization framework for waste-based FA-GGBFS GPC using Response Surface Methodology (RSM) integrated with a desirability function approach. An experimental programme comprising 18 unique mixtures arranged in a 3 × 3 × 2 full-factorial structure was conducted using three GGBFS replacement levels (25%, 50%, and 75%), three alkaline solution-to-binder (AL/B) ratios (0.45, 0.50, and 0.55), and two predefined mix-design levels (M1 and M2). Five performance responses, including slump, 7- and 28-day compressive strengths (CS), and 7- and 28-day splitting tensile strengths (SPT), were investigated. The results demonstrated that GGBFS replacement was the dominant factor influencing the investigated mechanical responses, while the AL/B ratio exhibited a non-monotonic influence on strength. CS and SPT generally increased with increasing GGBFS replacement within the investigated range, whereas mechanical performance generally improved as the AL/B ratio increased from 0.45 to approximately 0.50 and declined when it was further increased to 0.55. Slump increased with increasing AL/B ratio and decreased with increasing GGBFS replacement. The hierarchically reduced response models exhibited R 2 values ranging from 0.9302 to 0.9829, with predicted R 2 values ranging from 0.8000 to 0.9028 and Adequate Precision values between 16.1109 and 26.3199. Multi-response optimization identified constrained optimum at the upper investigated GGBFS level of 75%, with AL/B ratios of 0.508 for M1 and 0.494 for M2, yielding overall desirability values of 0.945 and 0.976, respectively. Comparison with the available confirmation results showed reasonable agreement between the experimental and model-predicted responses within the investigated design space. Overall, the proposed framework provides an experimentally supported approach for predicting and optimizing the fresh and mechanical performance of ambient-cured FA-GGBFS-based GPC within the investigated mixture-design domain.