Oscillatory flows and capillary effects in partially saturated and unsaturated porous media : applications to beach hydrodynamics

Dans cette thèse, on étudie les écoulements oscillatoires en milieux poreux (non saturés ou partiellement saturés) dus à des oscillations tidales des niveaux d'eau dans des milieux ouverts adjacents aux milieux poreux. L'étude est centrée sur le cas des plages de sable en hydrodynamique côtière, mais les applications concernent, potentiellement et plus généralement, les problèmes d'oscillation et de variation temporelle des niveaux d'eau dans des systèmes couplés, lorsque ceux-ci mettent en jeu des interactions entre les écoulements de sub-surface (milieux poreux) et les eaux de surface (milieux ouverts) : plages naturelles et artificielles; digues portuaires; barrages en terre; berges de fleuves; estuaires. Le forçage tidal des écoulements souterrains est représenté et modélisé ici, tant expérimentalement que numériquement, par une oscillation quasi-statique du niveau d'eau dans un réservoir externe ouvert, connecté au domaine poreux. On s'intéresse plus particulièrement aux écoulements verticaux forcés par une pression oscillatoire imposée au bas d'une colonne de sol. Sur le plan expérimental, ce type de forçage est obtenu par une machine à marée équipée d'un arbre rotatif. Au total, on utilise dans ce travail trois types d'approches (expérimentale, numérique, analytique), l'objectif étant d'étudier le mouvement vertical de la surface "libre" et l'écoulement non saturé sus-jacent, de façon à prendre en compte aussi bien les pertes de charge dans la zone saturée que les gradients de pression capillaire dans la zone non saturée. […]In this thesis, we study hydrodynamic oscillations in porous bodies (unsaturated or partially saturated), due to tidal oscillations of water levels in adjacent open water bodies. The focus is on beach hydrodynamics, but potential applications concern, more generally, time varying and oscillating water levels in coupled systems involving subsurface / open water interactions (natural and artificial beaches, harbor dykes, earth dams, river banks, estuaries). The tidal forcing of groundwater is represented and modeled (both experimentally and numerically) by quasi-static oscillations of water levels in an open water reservoir connected to the porous medium. Specifically, we focus on vertical water movements forced by an oscillating pressure imposed at the bottom of a soil column. Experimentally, a rotating tide machine is used to achieve this forcing. Overall, we use three types of methods (experimental, numerical, analytical) to study the vertical motion of the groundwater table and the unsaturated flow above it, taking into account the vertical head drop in the saturated zone as well as capillary pressure gradients in the unsaturated zone. Laboratory experiments are conducted on vertical sand columns, with a tide machine to force water table oscillations, and with porous cup tensiometers to measure both positive pressures and suctions along the column (among other measurement methods). Numerical simulations of oscillatory water flow are implemented with the BIGFLOW 3D code (implicit finite volumes, with conjugate gradients for the matrix solver and modified Picard iterations for the nonlinear problem). In addition, an automatic calibration based on a genetic optimization algorithm is implemented for a given tidal frequency, to obtain the hydrodynamic parameters of the experimental soil. Calibrated simulations are then compared to experimental results for other non calibrated frequencies. Finally, a family of quasi-analytical multi-front solutions is developed for the tidal oscillation problem, as an extension of the Green-Ampt piston flow approximation, leading to nonlinear, non-autonomous systems of Ordinary Differential Equations with initial conditions (dynamical systems). The multi-front solutions are tested by comparing them with a refined finite volume solution of the Richards equation. Multi-front solutions are at least 100 times faster, and the match is quite good even for a loamy soil with strong capillary effects (the number of fronts required is small, no more than N≈ to 20 at most). A large set of multi-front simulations is then produced in order to analyze water table and flux fluctuations for a broad range of forcing frequencies. The results, analyzed in terms of means and amplitudes of hydrodynamic variables, indicate the existence, for each soil, of a characteristic frequency separating low frequency / high frequency flow regimes in the porous system