Search

Dans le cadre du choix d’un site de stockage de déchets radioactifs de haute et moyenne activité à vie longue dans les argilites du Callovo-Oxfordien de Meuse/Haute-Marne, l’Andra (Agence nationale pour la gestion des déchets radioactifs) a réalisé 3 forages (EST431, EST432 et EST433) à partir d’une plate-forme implantée sur la commune de Montiers-sur-Saulx (Meuse) en position centrale d’une zone d’environ 250 km2 dite ≪ zone de transposition ≫.Les objectifs principaux des investigations ont été d’apporter des éléments pour la localisation du stockage potentiel et compléter les connaissances sur les formations encaissantes de la couche hôte (Dogger, Oxfordien carbonaté et Kimméridgien) mais également sur les formations profondes du Lias et du Trias. Plus précisément, ces derniers objectifs sont destinés à (i) acquérir une meilleure compréhension du fonctionnement global du système hydrogéologique et des échanges verticaux entre les formations et (ii) d’évaluer les ressources géothermiques potentielles sur la zone.Au-delà des objectifs liés aux problématiques de l’Andra, le forage le plus profond (EST433) a été échantillonné et étudié par plusieurs équipes de recherche associées dans le programme TAPSS (Transferts actuels et passés dans un système sédimentaire aquifère – aquitard) afin de répondre à des problématiques qui leur étaient propre. Cet article présente les données nécessaires à l’interprétation des données acquises par ces équipes et à la publication de leurs travaux.Les trois forages ont majoritairement été forés en destructif à l’exception du Kimméridgien marneux et du toit de l’Oxfordien carottés de 141 à 264 m de profondeur, du Callovo-Oxfordien et du sommet du Dogger carottés de 526 m à 770 m de profondeur. Certains horizons du Lias et du Trias ont également été carottés ponctuellement. De nombreuses diagraphies ont été réalisées entre les phases de forage ainsi que deux diagraphies thermiques 9 mois après le forage. La concaténation des données issues des trois forages a permis d’établir le log stratigraphique de la série du Tithonien au toit de l’Olénékien.Les informations du forage profond EST433 améliorent la connaissance de l’épaississement progressif de la couche du Callovo-Oxfordien vers le nord-est de la zone de transposition ainsi que la constance lithologique de celle-ci tant en vertical qu’en horizontal. Les perméabilités mesurées dans cette formation sont du même ordre de grandeur que celles observées dans les autres forages de la zone de transposition. Dans le Dogger, les variations de la perméabilité de certains niveaux sont liées aux environnements de dépôt. Les formations du Lias et du Trias sont conformes à celles dé-duites des forages antérieurs à l’exception des Marnes irisées inférieures (Keuper inférieur) plus épaisses que prévues. Les profils sismiques et les corrélations entre forages montrent l’existence d’une gouttière à l’aplomb du forage EST433 où les formations du Keuper (Carnien et Norien) sont plus épaisses. Les résultats du forage EST433 montrent également que les faciès du Lias et du Trias supérieur et moyen sont très peu poreux et que seule la base du Trias moyen (toit du Buntsandstein) contient des niveaux aquifères. Mille mètres de série sans porosité notable séparent l’aquifère du Dogger de celui du Trias inférieur.L’aquifère supérieur du Buntsandstein présente en test, entre 1862 et 1887 m de profondeur, une production d’eau à 66°C avec une salinité d’environ 180 g/l. La transmissivité calculée pour cet intervalle est de 1,1 10−3 m2/s. En comparaison avec les installations géothermiques exploitant l’aquifère du Dogger en région parisienne, la température mesurée sur EST433 est de l’ordre des températures exploitées les plus basses, la salinité nettement plus élevée (6,5 à 35 g/l en région parisienne) et la transmissivité du même ordre. Ces caractéristiques ne présentant pas un caractère exceptionnel (en termes de ressource potentielle pour une exploitation géothermique) associées au caractère argilo-gréseux de l’aquifère pouvant influer négativement sur la productivité d’un ouvrage ne font pas du Buntsandstein une ressource géothermique attractive dans les conditions technologiques et économiques actuelles.

International audience; Among diversified industrial uses, see for instance Koch [Koch, Appl. Clay Sci., (21) 2002], and following positive in situ experiments, compacted bentonite blocks are potential candidates for sealing nuclear waste repositories, thanks to their swelling ability in a wet environment. As requested by Andra (French Agency for Nuclear Waste Management) and complementarily to in situ experiments, an original experimental laboratory set-up was designed in order to reproduce the introduction and swelling of bentonite plugs inside an argillite host rock. Once the argillite/bentonite interface is established, an increase in storage tunnel gas pressure is simulated and the interface gas migration pressure (or gas critical pressure) is evaluated. More precisely, a first experimental set-up provides bentonite swelling pressure and kinetics (i.e. mainly hydraulic cut-off, time to reach asymptotic swelling pressure and value of asymptotic swelling pressure) at given initial compaction and saturation rate. This phase is preparatory to reproducing the introduction and subsequent swelling of a bentonite plug inside the argillite host rock, which uses a similar test rig. Experimental results of water permeability and gas critical pressure are provided for MX80 compacted bentonite associated to Bure Callovo-Oxfordian argillite.

In studies related to deep geological disposal of radioactive waste, it is current practice to transfer external information (e.g. from other sites, from underground rock laboratories or from natural analogues) to safety cases for specific projects. Transferable information most commonly includes parameters, investigation techniques, process understanding, conceptual models and high-level conclusions on system behaviour. Prior to transfer, the basis of transferability needs to be established. In argillaceous rocks, the most relevant common feature is the microstructure of the rocks, essentially determined by the properties of clay–minerals. Examples are shown from the Swiss and French programmes how transfer of information was handled and justified. These examples illustrate how transferability depends on the stage of development of a repository safety case and highlight the need for adequate system understanding at all sites involved to support the transfer.

La surete d'un stockage en formation geologique est basee sur des fonctions allouees aux differentes barrieres (colis de dechets, barrieres ouvragees, formation geologique), pour assurer la protection des dechets vis-a-vis de l'eau et le confinement des radionucleides sur le long terme. Les recherches ont porte sur les proprietes de transfert dans chacune des barrieres dans les conditions d'environnement d'un stockage. En s'appuyant sur des progres metrologiques importants, elles ont abouti a la determination de tres faibles permeabilites et diffusivites avec une incertitude reduite. Les grands axes de developpement visent d'une part la comprehension des mecanismes a l'echelle microscopique pour mieux evaluer l'evolution dans le temps de ces proprietes, d'autre part le role des interactions entre barrieres dans cette evolution.

Clay formations in their natural state exhibit very favourable conditions for disposal of radioactive waste. One concern regarding waste disposal is that due to the necessary underground excavations and the associated disturbance and damage in the area close to these excavations, the favourable properties of such formations could change and the host rock could lose part of its barrier function. Stress redistribution will lead to the creation of a so-called excavation damaged zone (EDZ) which will be controlled by the initial stress field, the material properties (e.g., material anisotropy), the existence of natural fracture zones or local inhomogeneities of the rock mass and the geometry of the tunnel. Comprehensive investigations at different sites (e.g., HADES, Belgium, Mont Terri, Switzerland, Tournemire, France) have shown that an EDZ occurs in soft or plastic clays as well as in indurated and more brittle claystones. The short-term excavation-induced reaction of the rock during tunnelling, which leads to the initial EDZ, cannot be avoided but is reasonably well understood and the associated processes can be adequately modelled. The long-term behaviour of the tunnel near-field can be significantly influenced by adequate support measures and the time-dependent evolution of the EDZ before the emplacement of the waste and the backfilling of the tunnel can be controlled. The properties of the initial EDZ alter significantly during the transient phase, when the buffer and rock mass are heated by the heat-producing waste and become saturated due to the flow of formation water from the host rock. Experimental results in the laboratory and in-situ clearly show that (self-)sealing leads to a significant reduction in the effective hydraulic conductivity of the EDZ with time, thus reducing the potential flow along underground excavations. Expected long-term conductivities within the EDZ are in the range of 10 −10 to 10 −12 m/s. Performance assessment calculations for different repository designs in different clay host rock formations show that the influence of the EDZ on radionuclide release is quite limited. It has been shown that even for very conservative, so-called “what if?” cases the very stringent regulatory guidelines can be met.

International audience; This work benefited from a comprehensive set of logging and mineralogical data that was acquired from the 2-km-deep EST433 borehole of Montiers-sur-Saulx (Meuse, France). A multiple linear regression analysis was performed on three sets of data: (a) sonic data (P- and S-wave velocities), (b) total combinable magnetic resonance (TCMR) porosity data, and (c) X-ray diffraction (XRD) data. The analysis was performed to estimate the clay content in the Callovo-Oxfordian (COx) claystone and other clay-rocks of the stratigraphic column. The inferred statistical relationships were compared to results obtained from a conventional spectral gamma ray (GR) approach.The results indicate that simple linear regressions using only elastic wave velocities (a) accounted for more than 81% of the vertical variability of the clay content in the COx formation (i.e., coefficient regression R-2 values greater than 81%) and (b) enable the estimation of clay content with the same accuracy as the spectral GR approach. Both approaches, our acoustic regressions and spectral GR methodology, provide estimate errors that are within the range of uncertainties associated with the XRD analysis applied. Moreover, using the whole set of XRD data showed that the elastic wave velocities and the TCMR porosity as explanatory variables are better predictors of total clay content than swelling clay mineral content.Blind estimations of clay content were performed using data acquired in clay-rocks from the same stratigraphic column and in the COx formation located a few kilometers distant in the EST423 borehole. The results demonstrated that the established regressions must be used with caution when vertically and horizontally extrapolating the estimated clay content. This difficulty is partly explained by the influence of the microstructure of the studied clayrocks, which was insufficiently taken into account via the petrophysical variables considered in this work.

From March 1994 to August 1996, ANDRA conducted a large reconnaissance work on three sites for possible underground research laboratories. These surveys highlighted for each site the presence of a geological formation, the properties of which meet the requirements of the French Fundamental Safety Rules. On the East site, the selected formation is a 130 m thick argillite layer. On the Gard site the layer is composed of close-packed siltites which are 200–400 m thick. On the Vienne site a hercynian granodiorite batholith was found under a 160 m thick jurassic overdurden.

SUMMARY Until now, observations of mechanically and thermally induced microcracks in rocks could only be carried out by indirect measurements or destructive observations on samples brought back to atmospheric pressure conditions. A special triaxial test cell was designed in order to perform direct observations during loading. The use of a cell in tomography apparatus involves new devices: (1) a movable horizontal load frame around a scanner; and (2) a test cell transparent to X-rays, able to withstand up to 28 MPa maximum confining pressure and temperatures of up to 180°C. Volumetric strains are compared with radiological density measurements. The first processed X-ray images locating microcracks during propagation are also presented. Mineralogical eVects on the crack location can be demonstrated. Strain inferred from CT density measurement is clearly correlated with the strain usually measured by a strain gauge. DiVerent phases of mechanical behaviour are described: contracted phase and failure by macrocrack formation. The principal results obtained with this tool are the description of the porosity formation and macrocracking. Results show two principal factors localizing the porosity. First, the diVused porosity volume is controlled by mineralogical parameters, quartz and plagioclase grains, and boundaries of biotite grains during the thermal and mechanical loading. Second, macrocracking begins at the perimeter of the central section of core and grows towards the sample/piston interface. It seems that the first macrocracking is not located in the high-porosity zone formed during the loading phase, but in a relatively low-porosity zone.

This paper presents the different studies realized or launched by ANDRA in collaboration with different contractors, including laboratory and in situ experiments, as well as physical and numerical modelizations, related to the thermo-hydromechanical behavior of clays and clayey materials. Clays are considered as both potential host rocks and sealing materials, among other geological formations and materials, respectively. The study of a high-level or long-lived nuclear waste disposal concept is necessarily a step-by-step procedure, with two main objectives: on one hand, ensuring a good disposal design, including feasibility of the vaults and their stability during the operating phase with the eventual continuation of the phase of waste retrievability, i.e. on a long period of time not shorter than one hundred years. On the other hand, the objective is to preserve the long-term clay properties with regard to radionuclide behavior, in particular those properties which influence their containment capacity. Swelling clays as an engineered barrier material have been considered by ANDRA for a long time. Buffer material as an interface between the nuclear waste packages and the geological barrier and tight core of dams in drifts and shafts are the two types of constructions for which the clay conditioning and performance have been investigated. In both of these applications, the understanding of the THM behavior of the swelling clay is required. In the first case the THM load parameters are much more severe and the geochemical imbalances are of greater consequences. R & D programmes are necessarily integrated into the design process of the disposal concept. The results will allow selection between technical solutions estimated as valuable candidates for validation within the project duration, i.e. 13 years, and those solutions which cannot be. In return the design process indicates what are the more profitable axis for the future R & D efforts.

The structural deformation in clays results from microscopic phenomena involving the mechanical contact-stress change, the physico-chemical variation of repulsive forces in expansive clays, and thermal dilatancy of macropores. These textural strains are associated to three plastic mechanisms represented by respectively the yield surfaces f Tm , f R-A and f T . Under a thermal cycle, the sizes of interlamellar spaces between clay platelets are not modified, hence the temperature cycle is expected to have no effect on repulsive forces and thus the second mechanism is not affected by temperature changes. This paper suggests a formulation of a model of thermo-elasto-plastic behaviour of non-expansive saturated clays characterised by two plastic mechanisms. The mechanical yield surface f Tm of the contact-stress mechanism is based on a modified cam-clay model; the thermal softening yield surface f T is a plane separating two thermal domains. In normally consolidated conditions, the resulting response to an increase of temperature is compressive. However, in highly overconsolidated conditions, a small irreversible dilative volumetric strain is observed when the temperature is above a threshold value. In intermediate conditions, the material starts with an expansion and tends to a compression. The constitutive model combines thermo-mechanical hardening, predominant in normally consolidated states (NCS) and absent in overconsolidated states (OCS) where the thermal softening occurs. The characterisation of the model requires information about rheological parameters obtained from oedometric and triaxial paths. Lastly, some numerical simulations of thermo-mechanical tests on remoulded Boom, ‘Bassin Parisien’ and Pontida clays are presented, which show satisfactory agreement between experiments and model predictions.

Cet article presente le site Est (bordure Est du bassin de Paris, au sud de Bar-le-Duc, entre la vallee de la Marne a l'Ouest et le fosse de Gondrecourt a l'Est), le site Gard (triangle constitue des vallees de la Ceze au sud-ouest et du Rhone a l'est, et limite au nord-ouest par une ligne Bagnols-sur-Cese/Pont-Saint-Esprit) et le site Vienne (cantons de Civray et Charroux, situes sur le seuil du Poitou)