La thèse se présente comme un bilan de plus de 25 ans de fréquentation de la craie, dans divers domaines, principalement dans le domaine pétrolier. La phénoménologie du comportement de ce matériau atypique craie est présentée dans ses divers contextes, dans une optique pluridisciplinaire. Le mémoire comprend quatre parties principales : -la première partie concerne la craie en elle-même: sa constitution, sa répartition stratigraphique et géographique, sa structure, ses propriétés physiques et pétrophysiques générales; -la deuxième partie pose les problèmes des réservoirs pétroliers de Mer du Nord; -la troisième partie, la plus volumineuse, aborde le comportement mécanique de la craie comme roche réservoir; -la quatrième partie expose certains développements en cours et considère les perspectives de développements futurs.

How can the suitability of a clay to act as a barrier to the flow of a specified fluid be determined? This question is directly related to the different mechanical and chemical stresses to which a clay barrier will be exposed. In spite of these mechanical and chemical stresses it must be guaranteed that the clay will fulfil its barrier function during the entire required containment period. This required technical life could be very long in engineering terms: 100-10000 years. During this period the clay barrier can neither be repaired nor maintained. Therefore it must be known which chemical or physical reactions will occur and how these reactions will influence the geomechanical properties of the clay. Because there was no standard approach to test the suitability of natural clays as barrier on the long-term, this had to be developed. Based on literature it was shown that the reactions between clays and fluids could be decomposed in reactions on the particle level, the interlayer level and the TOT/TO level of clay minerals: - Micrometer: Reactions on the particle level are the most frequent, the fastest to accomplish (instantaneous when leachate arrives) and have the least impact on the geomechanical properties of clays. It was shown that the double layer theory presents a valuable framework to analyse the changes in geomechanical properties upon clay-leachate contact. The properties of the fluid that are taken into account are the concentration of cations and the relative dielectric constant. Other processes on the particle level not captured by the double layer theory are e.g. the dissolution of calcitic cement and the oxidation of pyrites. The acids produced by the latter process influence reactions on the lower interlayer and TOT/TO level as well. It was shown that the natural clays possess themselves a rich variety of cations. These concentrations must be included in the analysis. New tools developed on the particle level were: - Integration of the chemical composition of the natural fluid contained in the clay in further analyses. - The discretisation of clay samples into a discontinuous but homogeneous assembly of discrete clay particles (finite element mesh) with the use of information from petrographical studies of thin sections and oedometer tests. - The implementation of a constitutive law into a numerical code to simulate the interparticle distance to interparticle fluid chemistry and mechanical stress. - Nanometer: Reactions on the interlayer level include clay mineral alteration processes. To link these processes to geomechanical properties, the clay mineral sample preparation was modified to include all clay minerals and not only the fraction smaller than two micrometers. Next a method was developed to link clay mineralogy to geomechanical properties (equivalent basal spacing). New tools developed on the interlayer level were: - The equivalent basal spacing (EBS) - Relation between the equivalent basal spacing and the liquid limit With these tools a link can be made between the clay mineralogy and geomechanical properties. Leachate - clay interactions can be analysed as well as other processes like the mixing of clays and the reactions of clays upon heating etc. - Ångström: Reactions on the TO/TOT level include the disintegration of TO arrangements, which will result in a complete destruction of a clay mineral. Of all three levels considered, changes on the TO/TOT level will cause the greatest change in geomechanical properties. Fortunately the processes on this TOT/TO level take a long history of subsequent physical and chemical reactions (hundreds to thousands of years in situ). Because changes on this level fail to be reproduced in the laboratory one must rely on natural analogues. New tools developed on the TO/TOT level were: - The link between the clay leached in the laboratory to natural analogues using thin sections and XRD diffraction analysis. Examples are shown that the aforementioned approach can be applied in any geomechanical problem involving clays.

Les sols non saturés jouent un rôle important dans de nombreux domaines. En Génie Civil (stabilité des talus, fondations superficielles), en Ingénierie de l'environnement (transport de polluants par les eaux souterraines, enfouissement de déchets radioactifs), en Ingénierie des Réservoirs (subsidence suite aux pompages dans les champs pétroliers ou dans les nappes aquifères), etc. L'objet de la thèse est l'étude du comportement hydromécanique des sols fin non saturés, tels les limons et les argiles. Les phénomènes typiques du comportement de ces sols sont d'abord présentés avant d'aborder l'étude de leurs lois constitutives. Divers modèles sont passés en revue et un modèle unifié original est développé. Ensuite, une vaste gamme d'essais de laboratoire est analysée. Ils portent sur des sols saturés, quasi saturés et non saturés et concernent quatre matériaux: l'argile de Kruibeke, le limon de Gembloux, l'argile de Boom et la bentonite d'Almeria. Cette partie de la recherche conduit à une critique détaillée des procédures classiques d'essais de laboratoires sur les sols supposés saturés. Elle fournit également une base de données importante pour la calibration de la loi constitutive unifiée développée plus haut. Enfin, des simulations numériques, réalisées en aveugle, d'expériences n'ayant pas servi à la calibration permettent de valider ce modèle.

This thesis is dedicated to the study of structural and petroleum geology problems. To this purpose, a frictional elastoplastic law based on the Van Eekelen criterion is formulated, which avoids the classical drawbacks of the Drücker Prager criterion. Also, a 2D automatic adaptive re-meshing algorithm is developed for complex multidomains configurations, in order to overcome the limitation of the Lagrangian mesh. Details of the hydromechanical formulation implemented in the LAGAMINE FE code in a large strain context are presented. Application cases (reproduction of sandbox simulation, study of a hydrocarbon trap) concentrate on the study of the strain localisation and potential fracturation using different criteria. In the first case, re-meshing technique allowed to reproduce successfully analogue experiment of thrusting propagation. In the second case, a detailed study based on different initial conditions has brought new insight to the reactivated origin of some faults and has allowed to obtain information on the potential fracturing of the hydrocarbon reservoir unit. As an academic case, the study of anorthosite diapirism is carried out, which confirms the validity of the petrological model of diapirism. Eventually, the hydromechanical coupling effects between a layered porous medium and a fault are illustrated on a simple case.