A review on the engineering performance and durability of geopolymer concrete under extreme conditions: High-temperature, carbonation, and freeze-thaw resistance

Geopolymer concrete (GC), a low-carbon cementitious material with strong potential for industrial waste utilization, is a promising alternative to ordinary Portland cement. This paper reviews the mechanical performance, degradation mechanisms, and artificial intelligence-based prediction of GC under high temperature, carbonation, and freeze–thaw conditions. GC exhibits strength enhancement at 150-350 °C, followed by degradation above 400 °C due to microcracking; fiber and nanomodification improve residual strength. Carbonation resistance is generally lower than that of conventional concrete, with significant pH reduction and more severe degradation in high-calcium systems. Under freeze–thaw cycles, GC shows superior durability, with much lower strength loss than Portland cement concrete, and improved resistance through slag incorporation and fiber reinforcement. Performance deterioration is mainly governed by microcrack propagation, pore structure evolution, and expansive product formation. Machine learning models demonstrate high accuracy in predicting GC performance under extreme conditions (R > 0.90). This paper also conducts pioneering research on various machine learning and deep learning models in terms of fire resistance, carbonation resistance, and frost resistance of geopolymer concrete, providing an understanding of the current status of geopolymer concrete in the field of artificial intelligence.