Nonlinear Modeling and Least Squares Optimization on Consolidation Property of Composite Soil Samples

Abstract

Nonlinear least square estimation is proposed to determine the physical index and self-weight consolidation properties of diverse composite samples being homogenized out of tube samples, which are taken from various borings in the State of Louisiana. Due to the high sensitivity of soil, a low pressure loading system is applied to perform self-weight consolidation tests, with stresses ranging from 10, 50, 100, 200 to 600 psf and one unloading segment at 200 psf. Typical data analysis using the Casagrande (Log time) method or Taylor (Root time) method could be implemented to determine the primary parameter of the coefficient of consolidation (Cv), which would require long-lasting multiple procedures to locate a few consolidation points on testing data curves manually. A simple hyperbolic tangent function model instead is proposed in this study. Parametric optimization is conducted via nonlinear least square estimation, where data points from 0 to 100% deformation will all be used to optimize the result and compute the coefficient of consolidation (Cv), in order to better predict the actual time rate of the settlement. It indicates that those composite samples are in fact the well degraded clays with the high in-situ water contents, high plasticity, low consolidation coefficients and low hydraulic conductivities. For testing results, coefficients of consolidation obtained from the Taylor method is slightly higher than the Casagrande method. Least square curve fitting has the potential to provide more accurate results than the two existing methods.