Numerical Back-Analysis of In-Situ Constant Head Tests in Partially Saturated Soil Cover to Determine the Permeability Function

Abstract

The determination of the coefficient of saturated permeability, k_sat, of a porous medium is a complex task to carry out, since k_sat highly depends on several factors, e.g. nature of the soil, size and the shape of soil particles, void ratio, soil structure, etc., and it also depends on the testing procedure adopted.

Since accurate determinations of the k_sat are necessary for numerous applications in the engineering practice, methods for both direct and indirect measurements of the permeability are commonly adopted, mainly by inducing and monitoring either steady-state or transient seepage processes in the soil deposit to investigate.

Typically, in-situ determinations are preferred to those in the laboratory since it is generally believed that in-situ measurements allow to account for any hydraulic heterogeneities in the soil, resulting in more representative testing results; however, during in-situ tests, the initial soil state and the boundary conditions of the induced seepage in the soil are not imposed and controlled, and this poses uncertainties which lead to potentially inaccurate k_sat determinations.

In this paper, coupled hydro-mechanical numerical back-analyses of constant head permeability tests, i.e. Guelph’s tests, are reported aimed at better estimating the k_sat, by computing also the wetting path towards full saturation occurring in the soil.

The comparison between the k_sat determinations, obtained from numerical back-analyses, and the results from the semi-empirical relationships commonly adopted for processing the in-situ tests, allow to confirm that the latter always tend to overestimate the saturated permeability values of a non-negligible quantity.