A simple model for binder-stabilised soils considering nonlinear stiffness characteristics and strain softening

Abstract

Binder-stabilised soils exhibit significant nonlinear small-strain stiffness characteristics and post-peak strain-softening behaviour. It is crucial to properly model these behaviours to optimise design, enhance safety, and reduce construction costs in land reclamation projects using the deep mixing technique. Existing models for stabilised soils either fail to capture these important behaviours or require too many parameters that are difficult to calibrate. This study developed a new, simple model for binder-stabilised soils within the bounding surface plasticity framework for analysing deep mixing projects. The proposed model requires only seven parameters, all of which can be easily calibrated through unconfined compression, triaxial, and tensile tests. The model was validated by simulating laboratory triaxial, physical model tests, and field tests. It was also compared with two other models: the Mohr-Coulomb (MC) model, which has a comparable number of parameters, and an advanced but complicated bounding surface model with significantly more parameters. The new model effectively captures key aspects of binder-stabilised soil behaviour, such as strain-softening and nonlinear small-strain stiffness. It shows comparable performance to the advanced bounding surface model, except in the modelling of dilatancy. However, this limitation is unlikely to significantly impact the model performance in engineering analysis because the dilatancy of strongly bonded soil plays a relatively minor role compared to other modelling aspects (e.g., the dilatancy of untreated soils and bonding degradation of binder-stabilised soils). Both the new, simple model and the existing, complex model perform significantly better than the MC model, particularly in capturing important strain-softening behaviour and nonlinear small-strain stiffness characteristics, which are crucial for ultimate and serviceability limit state analyses, respectively.