Performance and alkalinity control in geopolymer coral aggregate concrete via experiments and machine learning

Abstract

In ocean engineering, the combination of coral aggregate concrete (CAC) and fiber-reinforced polymer (FRP) composites offers potential for cost-effective, durable structures. However, the highly alkaline environment of cement-based matrices degrades FRP composites via resin hydrolysis, leading to performance deterioration of FRP-reinforced concrete structures. To address this challenge, fly ash-slag composite geopolymers were utilized to develop low-alkalinity geopolymer coral aggregate concrete (GPCAC). A systematic analysis was conducted to evaluate the effects of alkaline dosage and slag/fly ash ratio on the mechanical properties, microstructures, and pore solution alkalinity of GPCAC. The results revealed that increasing the alkali dosage from 4 % to 10 % enhanced the 28-day compressive strength by 33.64 MPa and splitting tensile strength by 2.72 MPa, whereas raising slag content from 30 % to 70 % boosted these strengths by 11.11 MPa and 0.71 MPa, respectively. At 7 days, GPCAC specimens with 10 % alkaline content exhibited a 0.47-unit higher pH value than the 4 % alkali group, and specimens with 70 % slag achieved a 0.25-unit pH increase over the 30 % slag group. However, these disparities diminished by 28 days. Notably, the pore solution pH of GPCAC measured 0.13–0.48 units lower than cement-based CAC, creating a low-alkalinity environment advantageous for mitigating FRP composite degradation. In addition, GPCAC exhibited a higher gel pore proportion than CAC, and optimizing alkali and slag content further increased gel porosity, yielding a denser matrix. An optimal pore structure was achieved with a 6 % alkali dosage and 50 % slag content, with gel pores accounting for 51.42 %, which was 2.15 times that of cement-based CAC. Ultimately, a snake optimization algorithm-enhanced random forest model accurately predicted GPCAC compressive strength with a prediction error below 5 %.