Avancées dans la caractérisation et la modélisation du comportement des sols non saturés : protocoles expérimentaux, fiabilité numérique et milieux stratifiés

« RÉSUMÉ : Cette thèse présente une étude approfondie du comportement hydraulique des sols non saturés, en intégrant des protocoles expérimentaux détaillés, des modèles prédictifs et des analyses numériques avancées afin d'améliorer la précision et la fiabilité des simulations d'écoulement transitoire dans des milieux poreux homogènes et stratifiés. Ce travail repose sur le constat que la mécanique des sols saturés ne permet pas de représenter adéquatement la complexité des conditions naturelles non saturées, notamment dans la zone vadose, où les sols subissent des cycles d'humidification et de dessiccation dus à l'infiltration, à l'évaporation et aux variations climatiques. Des applications en ingénierie géotechnique, telles que la stabilité des pentes sous pluie, les performances des couvertures de décharges ou la gestion des résidus miniers, exigent des modèles robustes capables de représenter ces systèmes dynamiques et souvent hétérogènes. Un volet essentiel de cette recherche concerne la détermination expérimentale des courbes de rétention d’eau (CRE), qui décrivent la relation fondamentale entre la succion matricielle et la teneur en eau volumique. À l’aide de cellules de Tempe, des courbes de rétention ont été mesurées pour des sables silteux à des paliers de succion maintenus pendant 24, 48 et 72 heures. L’étude met en évidence une limite critique de la norme de 24 heures plus un délai d’observation pour estimer si l’équilibre a été atteint. En effet, la mesure des faibles volumes d’eau s’écoulant de la cellule Tempe, particulièrement pour les sols à faible perméabilité, présente des difficultés notables. Ces volumes sont souvent proches de ou inférieurs à la sensibilité limite de l’appareil de mesure, ce qui accroît les incertitudes de mesure et peut donner l’impression que l’équilibre hydraulique est atteint alors qu’un drainage partiel persiste. Cette situation entraîne une sous-estimation de la teneur en eau et, par conséquent, une erreur dans la détermination de la courbe de rétention, en particulier pour les sables silteux et autres matériaux à drainage lent.En effet, la mesure des faibles volumes d’eau s’écoulant de la cellule Tempe, particulièrement pour les sols à faible perméabilité, présente des difficultés notables. Ces volumes sont souvent proches de ou inférieurs à la sensibilité limite de l’appareil de mesure, ce qui accroît les incertitudes de mesure et peut donner l’impression que l’équilibre hydraulique est atteint alors qu’un drainage partiel persiste. Cette situation entraîne une sous-estimation de la teneur en eau et, par conséquent, une erreur dans la détermination de la courbe de rétention, en particulier pour les sables silteux et autres matériaux à drainage lent.»

Abstract

« ABSTRACT : This dissertation presents a comprehensive investigation into the hydraulic behavior of unsaturated soils, integrating detailed experimental procedures, predictive modeling, and advanced numerical analysis to improve the accuracy and reliability of transient flow simulations in both homogeneous and stratified porous media. The work is grounded in the recognition that traditional saturated soil mechanics fail to capture the complexities of natural unsaturated conditions, particularly in the vadose zone where soils experience cyclic variations in moisture due to infiltration, evaporation, and climate-driven changes. Applications in geotechnical engineering, such as slope stability under rainfall, landfill cover performance, and mine tailings management, demand robust models that can represent these dynamic and often heterogeneous systems. An essential component of this research concerns the experimental determination of water retention curves (WRCs), which describe the fundamental relationship between matric suction and volumetric water content. Using Tempe cells, retention curves were measured for silty sands at suction steps maintained for 24, 48, and 72 hours. The study highlights a critical limitation of the conventional 24-hour standard, along with the need for an additional observation period to verify whether hydraulic equilibrium has been reached. Indeed, measuring the very small volumes of water draining from the Tempe cell, particularly for low-permeability soils, presents significant challenges. These volumes are often close to or lower than the detection limit of the measuring apparatus, which increases measurement uncertainty and may create the false impression that hydraulic equilibrium has been achieved while partial drainage is still occurring. This situation leads to an underestimation of the water content and, consequently, to errors in determining the retention curve, especially for silty sands and other materials characterized by slow drainage.This dissertation presents a comprehensive investigation into the hydraulic behavior of unsaturated soils, integrating detailed experimental procedures, predictive modeling, and advanced numerical analysis to improve the accuracy and reliability of transient flow simulations in both homogeneous and stratified porous media. The work is grounded in the recognition that traditional saturated soil mechanics fail to capture the complexities of natural unsaturated conditions, particularly in the vadose zone where soils experience cyclic variations in moisture due to infiltration, evaporation, and climate-driven changes. Applications in geotechnical engineering, such as slope stability under rainfall, landfill cover performance, and mine tailings management, demand robust models that can represent these dynamic and often heterogeneous systems. An essential component of this research concerns the experimental determination of water retention curves (WRCs), which describe the fundamental relationship between matric suction and volumetric water content. Using Tempe cells, retention curves were measured for silty sands at suction steps maintained for 24, 48, and 72 hours. The study highlights a critical limitation of the conventional 24-hour standard, along with the need for an additional observation period to verify whether hydraulic equilibrium has been reached. Indeed, measuring the very small volumes of water draining from the Tempe cell, particularly for low-permeability soils, presents significant challenges. These volumes are often close to or lower than the detection limit of the measuring apparatus, which increases measurement uncertainty and may create the false impression that hydraulic equilibrium has been achieved while partial drainage is still occurring. This situation leads to an underestimation of the water content and, consequently, to errors in determining the retention curve, especially for silty sands and other materials characterized by slow drainage.This dissertation presents a comprehensive investigation into the hydraulic behavior of unsaturated soils, integrating detailed experimental procedures, predictive modeling, and advanced numerical analysis to improve the accuracy and reliability of transient flow simulations in both homogeneous and stratified porous media. The work is grounded in the recognition that traditional saturated soil mechanics fail to capture the complexities of natural unsaturated conditions, particularly in the vadose zone where soils experience cyclic variations in moisture due to infiltration, evaporation, and climate-driven changes. Applications in geotechnical engineering, such as slope stability under rainfall, landfill cover performance, and mine tailings management, demand robust models that can represent these dynamic and often heterogeneous systems. An essential component of this research concerns the experimental determination of water retention curves (WRCs), which describe the fundamental relationship between matric suction and volumetric water content. Using Tempe cells, retention curves were measured for silty sands at suction steps maintained for 24, 48, and 72 hours. The study highlights a critical limitation of the conventional 24-hour standard, along with the need for an additional observation period to verify whether hydraulic equilibrium has been reached. Indeed, measuring the very small volumes of water draining from the Tempe cell, particularly for low-permeability soils, presents significant challenges. These volumes are often close to or lower than the detection limit of the measuring apparatus, which increases measurement uncertainty and may create the false impression that hydraulic equilibrium has been achieved while partial drainage is still occurring. This situation leads to an underestimation of the water content and, consequently, to errors in determining the retention curve, especially for silty sands and other materials characterized by slow drainage.»