With the rise of energetic demand and the growing scarcity of proved reserves of oil, the oil industry explores areas of extreme conditions and geologically complex basins. To estimate the key parameters for exploration-production, simulation softwares are used to reconstitute the history of the basin. The physical modelling and the numerical formulation on which they lean must be enriched to understand (even predict) overpressures. This thesis work contains two items: the improvement of the modelling of geomaterials using micromechanics and the development of a hydromechanical simulation tool handling the specificities of our problem. A new micromechanical modelling of geomaterials is designed to take into account the mechanism of pressure-dissolution (chemical compaction). The asset of micromechanics is to produce a macroscopic law calibrated with microscopic data which can be measured in laboratory. Since the work of Athy (1930), modelling the evolution of porosity over geological time-scale remains a major challenge. Today it is estimated by means of empirical curves of porosity-depth, which regrettably present a big variability. Our approach consists in evaluating the porosity during burial relatively to the strain of the skeleton and the pore pressure - these two coupled variables resulting from fundamental mechanical equations. An original formulation is designed according to this approach to treat the sedimentation and the desequilibrium compaction, tensorially, in large strains, following a mechanism of mechanical compaction, with a time-dependent behavior. Its numerical encoding is almost finished and has already been partially validated with analytical solution. Once finished, this simulation tool should allow to treat not only oedometrical situations but more general situations (contrary to current simulators) and study basins with complex tectonic history
