Strength parameters, failure patterns and constitutive model analysis of multimethyl polyphenyl isocyanate modified waste slurry under triaxial loading conditions

The modification and application of waste slurry represent a crucial approach to mitigating environmental impact while creating valuable construction materials. This study comprehensively investigated the mechanical properties of polymethylpolyphenyl isocyanate (PM-200) modifying waste slurry (PS) to assess its potential as roadbed filler. Through systematic triaxial compression tests, the effects of confining pressure, PM-200 content, and water content on the mechanical behavior of PS were evaluated. Results revealed that decreasing water content and increasing PM-200 content significantly enhanced shear strength while intensifying brittle characteristics. Higher confining pressure and PM-200 content or lower water content transformed the failure mode from conjugate plane shear failure to single clear shear failure, accompanied by reduced fractal dimension of cracks. Based on these experimental findings, a statistical damage constitutive model incorporating residual strength was developed, demonstrating excellent agreement with experimental data across various testing conditions. The model parameters effectively quantified material brittleness, with shape parameter m positively correlating with brittleness degree. To optimize the constitutive model parameters, a novel hybrid approach combining BP neural network, genetic algorithm with three-region distribution strategy, and Bayesian optimization was implemented. This closed-loop feedback system dynamically optimized both model parameters and neural network hyperparameters, effectively preventing convergence to local optima and ensuring physical consistency through penalty function constraints. The refined model accurately captured the stress-strain relationship characteristics of PM-200 solidified soil with enhanced generalization capability. This comprehensive study provides valuable theoretical guidance for understanding PS behavior and optimizing its application in road construction materials and fillers.