Faire avancer la sûreté nucléaire

La Recherchev2


Extensions récentes de l'utilisation des modèles de fracturation discrète pour l'étude d'implantation d'un stockage souterrain de déchets radioactifs.



Email :

Mot de passe :

Billaux, D. Guerin, F. Riss, J. Dewiere, L. Fillion, E. hydrogéologie, (2000) (no.2) p. 41-52.

Type de document > *Article de revue

Mots clés > déchets radioactifs, faille/fracture, stockage profond

Unité de recherche > IRSN/DEI/SARG/BEHRIG

Auteurs > DEWIERE Lionel

Date de publication > 01/01/2000


The sitting of a nuclear waste repository in a geological medium involves, among other aspects, predicting water inflows in the shafts and drifts, and evaluating possible geometries for the waste handling and storage galleries. In sedimentary host rocks, porous medium hydrogeology can be used easily to provide water inflow estimates, while geology will describe the geometry of the various layers, as well as the limited number of faults that may cut them. However, crystalline rocks such as the Vienne site, may be cut by numerous faults and fractures. To deal with such host rocks, we need new concepts - which have been under development during the last 15 years - in order to describe properly the spatial arrangement of discontinuities, its consequences in terms of the site hydrogeology, and in terms of the geometry of volumes available between faults for designing the underground storage cavities. A starting point is building a model of the fractures, using the statistical description of the investigated fracture field, including dips, dip directions, sizes, and intensities noted in boreholes or on outcrops. Such a model can then be used to compute flows. It is based on idealizing fractures as planar objects, often disks, with statistical geometrical properties inferred from available data. The model realism can be improved by conditioning the geometry on data, either directly observed - by fixing in space observed fractures - or indirectly inferred - by integrating the results of hydraulic, or even tracer tests. Discrete fracture models can then be used for many treatments, well beyond simple flow and transport computations. We illustrate this through two studies applied to the crystalline Vienne massif. First, image analysis techniques that were first developed for two dimensions, and have been recently extended to three dimensions, help with describing the space available between discontinuities, in order to define the sound rock blocks available for the waste emplacement. The statistical distribution of the block properties, i.e. volume, sizes, shape, is computed first by a procedure using standard image analysis algorithms called 'erosion-dilatation', and then by a morphological study. This phase accounts for special constraints on block shapes, such as the need for a minimum distance between fractures and the boundaries of the blocks.