Application of Taguchi's optimization techniques for enhancing the fracture characteristics and brittleness of self-compacting alkali-activated concrete

Abstract

Alkali-activated concrete has emerged as a promising material for energy-efficient construction, offering a technically viable and eco-efficient alternative that aligns with global sustainability goals. This study explores optimizing fracture properties in self-compacting alkali-activated concrete (SAAC) through controlled variations in maximum aggregate size (d<inf>max</inf>) and fly ash. A systematic approach incorporating Taguchi's design of experiments (DOE) and ANOVA analysis was employed to identify optimal mix proportions that enhance fracture performance and ductility. The study employed the Weight-Compensated Work of Fracture Method (WWFM) based on curtailment of the tail of the P–? curve to determine the size-independent fracture energy (G<inf>F</inf>), enhancing the reliability of SAAC in structural applications. Additionally, the Two-Parameter Fracture Model (TPFM) evaluated the critical stress intensity factor (Ks<inf>Ic</inf>) and critical crack tip opening displacement (CTOD<inf>c</inf>), while the MATLAB-based Box-Counting Dimension Method (BCDM) assessed the fractal dimension (D). The findings revealed a higher fracture performance with 0 % fly ash and 16 mm d<inf>max</inf> (G<inf>F</inf> of 206.3 N/m and Ks<inf>Ic</inf> of 1.91 MPa?m), suitable for structural applications requiring maximum fracture energy and toughness. The study further tailored a higher ductility mix with 50 % fly ash and 16 mm d<inf>max</inf> (CTOD<inf>c</inf> of 0.032 mm and D of 1.144) offering a balanced solution for non-structural applications, providing sufficient strength with enhanced ductility. The closed-form predictive design (CPD) model enables the prediction of f<inf>t</inf> and K<inf>Ic</inf> under a specified maximum fracture load, offering engineers a practical tool to optimize SAAC formulations by adjusting aggregate sizes and binder proportions for specific project needs. Regression models aligned strongly with experimental and existing literature results, affirming the reliability of predictive performance for future SAAC mix designs. © 2025 Elsevier Ltd