Phd proposal : Prediction ability of discrete modelling for the design of complex geotechnical structures

The objective of the PhD project is to propose a discrete numerical model (DEM) simple enough to be used in engineering on geotechnical structures, and complex enough to provide a prediction capability of better quality than conventional elasto-plastic relationships (with 1 or 2 load surfaces), in particular in the case of complex loading paths (e.g. with rotation of principal stress axes and/or cyclic loading).

The work of the PhD student will be devoted to:
• the study of the local ingredients (grain shape or contact law) constituting the best compromise to enhance the prediction ability of the DEM, in particular for kinematically constrained loading paths (undrained constitutive response), for loose or very loose soils, or even when both are combined (e.g. liquefaction of loose soils under undrained cyclic loading);
• the definition of calibration methods that are both robust and simple, based on in-situ tests (pressuremeter and/or static penetrometer), and on laboratory tests;
• the definition of an initial state of the micro-structure of the numerical granular assembly representative of the initial state of the soil (porosity, connectivity, anisotropy), in order to propose methodologies that meet the best compromise between prediction quality and simplicity of implementation;
• the validation of the model from laboratory tests,
• the application and validation of the model on geotechnical structures (boundary problems), such as the different construction phases of embedded walls, or the foundation piles of wind turbines. Boundary problems will be addressed either through the FEMxDEM double scale numerical method, or through DEM only by adopting an adaptive discretization to significantly reduce the number of particles and thus the computation cost.

Deadline: send applications by Monday, May 25, 2020, but as soon as possible is better