Research on strength characteristics and prediction model of composite solid waste-modified expansive soil based on multiscale analysis
摘要
In current research on improving expansive soils with composite solid wastes, most studies focus on a single type of solid waste or lack coupled analyses of pore multi-scale structures and macroscopic mechanical strength. In this study, four typical industrial solid wastes—GGBS, CS, SF, and DG—were selected as stabilizing materials to investigate the strength evolution and structural stability of modified expansive soils under different dry densities and multi-stage wetting–drying cycles. The results show that the composite solid waste system destroys the montmorillonite structure through synergistic effects and generates C–S–H and AFt, significantly enhancing soil compactness and compressive strength. The optimal solid waste proportions were determined to be 10.4%, 5.5%, 5.0%, and 4.3%, respectively. Strength increased with dry density and reached its optimum at 1.83 g·cm−3. During the early stages of wetting–drying cycles, secondary hydration enhanced strength; subsequently, strength decreased due to pore expansion and structural deterioration and eventually stabilized with a retention ratio of approximately 0.8. Microstructural analyses revealed that the addition of solid wastes weakened the layered structure of montmorillonite, and that pore evolution followed a “shrink–fill → expand–fracture → stabilize” pattern, consistent with macroscopic strength evolution. Based on this, quadratic relationships (R2 > 0.98) between cycle number–strength and cycle number–fine-pore fraction (FPF), together with the regular variation of model coefficients with dry density, were used to construct an FPF–ρ–UCS joint prediction model. This model can accurately predict strength without requiring cycle number input, demonstrating stability, reliability, and strong engineering applicability, and providing effective theoretical support for expansive soil improvement design and rapid on-site strength evaluation.