Mesure de la conductivite hydraulique du depot d'argile Champlain de Lachenaie, Quebec: Theorie et applications.

摘要

The thesis comprises theoretical and experimental elements. For the theoretical part of the project, the finite element codes COMSOL and SEEP/W were used to model field permeability tests. With COMSOL, a displacement-pressure ( u-p) model based on the assumption of a linearly elastic material was used to combine the analyses of clay skeleton deformation and water flow in the porous media. For the experimental part of the project, the Lachenaie clay deposit, near Montreal, Quebec, was characterized in detail. Five types of permeability test were used. In the field, pulse tests and variable-head tests conducted in riser pipes with 52.5 and 12.6 mm inner diameters were conducted. In the laboratory, variable-head tests were conducted in oedometer and triaxial cells.;Results from the numerical simulations were analyzed using non-dimensional velocity graphs. This representation enables comparing the curvature of velocity graphs obtained from different types of test. The numerical results demonstrate that the velocity graph curvature depends on the α parameter, the product of the clay compressibility, the sand filter volume and the inverse of the real or virtual inner section of the observation well riser pipe. Smaller riser pipe diameters and softer soils result in velocity graph curvatures which are more pronounced. The theoretical velocity graph curvature also depends on the chosen hypothesis regarding the displacement of the clay-sand filter interface. The curvature is less pronounced if the displacements are assumed to be free at the clay-sand filter interface.;For pulse tests, the u-p model results clearly demonstrate that the deformation of the clay skeleton is two-fold. First, the cavity holding the sand filter changes volume. For a cavity with a length to diameter ratio greater than 4, a u-p model based on the hypothesis of a linearly elastic material results in a linear relationship between cavity volume and pressure. This cavity expansion follows approximately the Lamé (1852) relationship. Secondly, volume changes occur when pore pressures vary in the soil. If the two types of deformation are modeled with the same elastic parameters (Young's modulus and Poisson's ratio), the α value and the velocity graph curvature for pulse tests are fully determined by the Poisson ratio.;For variable-head tests, the two hypotheses regarding the displacements at the clay-sand filter interface (fixed or free) produce non-dimensional velocity graphs which are equivalent. However, for simulations with free displacements at the interface, the α parameter must take into account the expansion of the sand filter cavity. The Lamé relationship can be used to calculate an effective riser pipe section which takes into account this phenomenon. When this effective section is used, or when the displacements at the interface are fixed, the u-p model results for the case of radial flow correspond to the analytical solution used by Bredehoeft & Papadopulos (1980).;Numerical results obtained with SEEP/W and COMSOL allow the apparent shape factor values obtained by assuming a compressible soil skeleton to be compared with those used with the velocity graph method when a perfectly rigid soil skeleton is assumed. When α is less than 10-2, for the central portion of the velocity graph, the apparent shape factor values agree with those given by the Hvorslev ellipsoid formula multiplied by a factor 1.11. When α is greater than 10-2, the apparent shape factor values increase. For α greater than 1, the apparent shape factors values increase proportionally to α.;When the data of in situ variable-head tests (52.5 mm riser pipe) and triaxial tests are analyzed using standard interpretation methods based on the hypothesis of a perfectly rigid soil skeleton, they produce K values which are equivalent and representative of the large scale clay permeability (clay volume larger than 1 m3). For the upper part of the clay deposit, over an elevation of 5 m, the geometric mean of K is 2.1×10-9 m/s. For elevations lower than 5 m, the geometric mean of K is 1.3×10-9 m/s. Variable-head tests in oedometer cells result in lower K values which are representative of the clay matrix permeability. (Abstract shortened by UMI.).