Date of Award

Spring 1-1-2011

Document Type


Degree Name

Doctor of Philosophy (PhD)

First Advisor

John Scott McCartney

Second Advisor

Judith Wang

Third Advisor

Dobroslav Znidarcic

Fourth Advisor

Abbie Liel

Fifth Advisor

Hon-Yim Ko


A semi-empirical model was developed in this study to predict the impact of effective stress state and hydraulic hysteresis on the small strain shear modulus of unsaturated, compacted soils. Unlike previous empirical relationships for the small-strain shear modulus, this model incorporates constitutive relationships between effective stress, void ratio, stress history, hardening, and soil consistency. The model incorporates a stress-dependent hysteretic soil water retention curve relationship and a definition of mean effective stress equal to the product of the degree of saturation and matric suction.

The model is experimentally validated by considering small strain shear modulus data for a variety of soil types in the literature as well as from an independent testing program with a fixed-free resonant column device modified for suction control with the axis-translation technique. A flow pump was used to control the equilibrium matric suction and volumetric water content in a compacted silt specimen. The change in volume of the specimen was measured using a proximeter vertically mounted atop the soil specimen.

In both the model and experiments, for a constant net confining stress, the small strain shear modulus was observed to increase in a nonlinear fashion during drying, albeit at a reduced rate as the water occlusion conditions are reached. During subsequent wetting, the value of Gmax does not follow the same trend as during drying, similar to the hysteresis observed in the Soil Water Retention Curve (SWRC). Different from the SWRC, the value of Gmax remains higher than that during drying. This hysteretic trend is attributed to hardening due to the effective stress changes associated with increased suction during drying. After calibration with parameters defined from the data available in the literature, the predictive model follows the experimental data.