The relationship between volumetric water content of soils and their effective dielectric permittivity is the theoretical basis of time domain reflectometry (TDR) technique. In this study

effective dielectric permittivity of a reconstituted expansive soil sampled from Nanning is measured by time domain reflectometry at different volumetric water contents

and three approaches including empirical equation method

theoretical bound method and theoretical model

are used to simulate the variation of the effective dielectric permittivity. The measured values of dielectric permittivity are found to be located within bounds of the Hashin-Shtrikman and Wiener

but the range of the solutions from the Hashin-Shtrikman bound is narrower than the Wiener bound’s. The effective dielectric permittivity of Nanning expansive soil is overestimated by Topp’s empirical equation

since the bound water in the expansive soil leads to a lower effective dielectric permittivity. The effect of bound water can be considered in Looyenga empirical equation in which three phases are separated and better agreements between predictions and tested results are obtained. Theoretical models have advantage in conceptualizing the effects of internal structure

but for the Maxwell-Garnett model and differential effective medium model

the predicting results of taking liquid phase or gaseous phase as a continuous phase are not satisfactory

since the three- phase configuration is greatly changed during water content increasing. The performance of these two theoretical models can be improved if the saturation are introduced as dependent weight functions to incorporate the calculated results of liquid phase and gaseous phase.