Comparative Study on Strength Characteristics of Fiber-Reinforced Expansive Soils Stabilized with Cement and Alkaline Activators under Freeze–Thaw Cycles: A Random Forest–Based Approach

Expansive subgrade soils undergo significant swell–shrink cycles due to moisture imbalance (ingress and egress), resulting in low strength and high compressibility. This study examines the effects of freeze–thaw (F-T) cycles on fiber-reinforced expansive soils treated with an alkaline activator stabilizer (AAS). The AAS was prepared using slag and sugarcane bagasse ash (SB) with sodium silicate and sodium hydroxide at a 0.4 water-to-solid ratio. Geotechnical and microstructural properties of AAS-treated (slag/SB: 0/100 and 25/75) soils reinforced with polypropylene (PLF) and glass fibers at dosages of 0.5% and 1% were compared against those of cement-stabilized soils under cyclic temperature conditions. Results demonstrated that the fiber–AAS soil achieved a 22% improvement in geomechanical strength than cement-stabilized mixtures. Moreover, PLF reinforcement effectively minimized tensile cracking after six F-T cycles by enhancing interfacial bonding. Furthermore, an artificial intelligence model using the random forest algorithm was also developed to predict AAS strength under F-T cycles. Validated with experimental data, the model demonstrated high accuracy, reducing the need for extensive testing. This study underscores the potential of fiber-reinforced AAS and AI modeling in improving expansive soil performance under F-T conditions.