Guillaume Flood-Page, PhD student of the Geomechanics group of Laboratoire Navier, is defending his PhD thesis entitled “Assessing seismic liquefaction potential in gassy soils" on the 16th of January, 2024, at la Ruche room of ENPC, located in the Carnot building of the Descartes campus.

Summary

It is well established in geotechnical literature that the presence of a relatively small gas phase in nearly saturated soils can significantly improve their resistance to liquefaction. Meanwhile, the presence of such a gas phase has regularly been attested in soils containing organic matter and in sediments near the water table. Therefore, accounting for the presence of these small amounts of gas could substantially optimise the design of liquefaction mitigation structures. However, as things stand, two main difficulties remain from an engineering perspective. The first obstacle is the common absence of any reliable in-situ data on the saturation state of soils. To address this, we have looked into the feasibility of using P and S wave measurements from cross-hole tests to estimate the degree of water saturation. Through an experimental investigation on Fontainebleau sand, we show this approach would likely be highly unreliable due to poor resolution and scattering of compression waves in gassy soils. Nevertheless, the silver lining to this disappointing conclusion is that this very same scattering effect could potentially serve as an alternative marker for the distribution of water in soils. The second issue is the absence of any standard engineering tool for modelling gassy soils. In this light, the initial approach was to implement a fluid compressibility law in FLAC3D and combine it with established constitutive models of the Sanisand family (Dafalias et Manzari, 2004; Cheng et Detournay, 2021). Theoretical considerations showed it was impossible to model both gassy and saturated soils with a single set of parameters unless great care was given to the hydromechanical coupling. Therefore, a first piece of work consists in choosing appropriate constitutive parameters to achieve this. Then, simulating cyclic triaxial tests on gassy Hostun sand, we illustrate the considerable difficulty of fitting the constitutive parameters required for modelling cyclic behaviour, even when laboratory tests are available. Altogether, the results of this study show this approach would likely be unreliable for engineering purposes. For this reason, a more empirical method is tested. Inspired of the works of Martin et al. (1975) and Byrne (1991), il allows for a cycle by cycle estimation of excess pore pressure generation. Associated with a fluid compressibility model, this method is shown to have the potential for modelling the cyclic behaviour of gassy soils by reproducing a number of cyclic laboratory tests preformed on gassy soils.