Statistical Assessment of Hydraulic Properties of Unsaturated Soils

Gilson De F N Gitirana,Delwyn G Fredlund

doi:10.28927/sr.391081

Abstract

The availability of statistical values for soil parameters is essential in reliability-based geotechnical design and sensitivity analysis. Unfortunately, there are few statistical studies available about unsaturated soil parameters. The primary objective of this paper is to present a methodology for the statistical assessment of hydraulic properties of unsaturated soil and to present the results of a statistical study carried out using a large database of soil properties. Two fundamental unsaturated soil properties are considered; namely, the soil-water characteristic curve (SWCC) and the hydraulic conductivity function. Appropriate nonlinear functions and fitting parameters with well-defined and unique physical and/or geometrical meanings were adopted. The main contribution of this article is the establishment of central tendency measures, standard deviations, and correlation coefficients for the unsaturated soil parameters, considering soil datasets grouped according to soil texture. It was determined based on the analyses results that the air-entry value, primary SWCC slope, residual SWCC slope, saturated hydraulic conductivity, and hydraulic conductivity function slope could be well described using lognormal probability density functions. Finally, general guidelines are provided regarding the statistical values to be adopted for the unsaturated soil properties studied.

Highlights

The analysis of unsaturated soil problems can be significantly enhanced and rationalized if the uncertainty associated with the soil properties is addressed through probabilistic analysis
The primary objective of this paper is to present the results of a statistical study carried out using 186 datasets of soil-water characteristic curves and hydraulic conductivity functions
Normality tests were performed on the soil parameters and on the natural logarithm of most soil parameters in order to determine whether normal or lognormal density functions reasonably represent each unsaturated soil parameter

Summary

Introduction

The analysis of unsaturated soil problems can be significantly enhanced and rationalized if the uncertainty associated with the soil properties is addressed through probabilistic analysis. Probabilistic analyses require the knowledge of statistical measures for the input properties involved. The main statistical measures required in probabilistic analyses are the central tendency values, standard deviations (or coefficients of variation), and the correlation matrix for the input variables. Representative values for the coefficients of variation and correlation coefficients are often used for the assessment of soil properties in engineering practice. Typical values for correlation coefficients and central tendencies can be found for saturated soil parameters. Most statistical assessment studies found in the geotechnical literature, such as the study by Ladd (1983), deal with the uncertainty of saturated soil parameters. Contrary to the standard deviation, the coefficient of variation can often be assumed to be independent of the magnitude of

Objectives

Methods

Results

Conclusion