Behaviour of Sandy Soils with Varying Fines Content and Saturation Under Monotonic and Cyclic Triaxial Loading: Experimental Study and Model Validation

Abstract

The behaviour of sandy soils under varying loading and saturation conditions is fundamental in geotechnical engineering, particularly for the safe and resilient design of foundations, embankments, and earth structures. While the behaviour of fully saturated soils is relatively well understood, the mechanical response of partially saturated soils, especially under cyclic loading remains a topic of active research due to its complexity and significant impact on infrastructure performance. This study investigates the influence of fines content and degree of saturation on the monotonic and cyclic behaviour of sandy soils and validates a constitutive model capable of simulating these responses within a Critical State Soil Mechanics (CSSM) framework.

An extensive triaxial testing program was conducted on sandy soils mixed with varying proportions of non-plastic feldspar fines under both saturated and partially saturated conditions. Tests included monotonic and cyclic loading to capture key behavioural characteristics. Results show that fines content and degree of saturation strongly influence soil response. Low fines content (<20%) preserved sand-like dilative behaviour, while high fines content (>30%) produced a fines-dominated compressive response. Partially saturated soils exhibited enhanced cyclic resistance due to matric suction, though very low saturation led to brittleness and strain localisation.

This study demonstrates the potential of a critical state-based model to simulate complex soil responses under different saturation and fines conditions, while also highlighting the limitations in current modelling approaches for cyclic behaviour. The findings underscore the importance of targeted experimental data for model calibration and point to the need for further model development, particularly in simulating cyclic degradation, suction hysteresis, and strain accumulation with improved computational efficiency.