A preliminary study for numerical representation of resonant column experiments in sand

Abstract

This paper represents a collaborative effort utilizing the finite element method to simulate the resonant column (RC) apparatus. The aim is to explore how soil dynamic properties change with varying strain levels. The RC test, renowned for its ability to analyze soil behaviour under dynamic loads, is the focus of our study. However, accurate measurement of dynamic properties using the RC can be influenced by several factors, necessitating further investigation. These factors, including strain uniformity, base fixity, strain localization, topplaten coupling, sample shape, and soil uniformity, are the key areas of our research. To gain a deeper understanding of the effects of these variables on measured shear wave velocity and damping ratio, we performed finite element analysis using Abaqus/Explicit, a commercial finite element package based on continuum mechanics. The model was based on the specific RC setup configuration at the University of Waterloo. Initial parameters included the lowstrain properties of sand (shear modulus, Poisson’s ratio, damping ratio) with shear strain adjusted as a loading variable. Torsional loads were applied across shear strains from 10-5 to 10-4. The element size of the soil specimen mesh was varied to 25 mm, 10 mm, 7.5 mm, and 5 mm to observe its effect on the outcomes of the RC test. The finite element model analyzed the free vibration of the cylindrical sand sample post-forced vibration, assessing dynamic properties. Modal analysis of the RC configuration was performed to verify the primary influence of the first torsional mode. A Z-factor has been proposed as a multiplier of experimentally obtained damping ratio. Comparisons of damping ratios and resonant frequencies at various shear strains between finite element modelling and laboratory data demonstrate a strong correlation in the case of nonlinear shear strain, with differences firmly ranging from 0.50% to 3.5%.