Thèse Étude Multi-Échelle de la Déformation Localisée Autour des Forages et Galeries dans les Roches Hétérogènes et Anisotropes H/F - Doctorat.Gouv.Fr

Établissement : Université Grenoble Alpes École doctorale : I-MEP² - Ingénierie - Matériaux, Mécanique, Environnement, Energétique, Procédés, Production Laboratoire de recherche : Laboratoire Sols, Solides, Structures et Risques Direction de la thèse : Pierre BESUELLE ORCID 0000000195864888 Début de la thèse : 2026-10-01 Date limite de candidature : 2026-05-18T23:59:59 Ce projet de doctorat vise à étudier l'initiation et l'évolution de la déformation localisée autour des excavations souterraines dans des roches poreuses hétérogènes et anisotropes, un enjeu clé pour la stabilité et l'intégrité des forages profonds et des galeries dans le contexte de la transition énergétique. Les excavations profondes dans les formations naturelles induisent des redistributions de contraintes tridimensionnelles complexes, et les modes de localisation de la déformation qui en résultent dépendent fortement des chemins de chargement multiaxiaux, ainsi que de l'anisotropie et de l'hétérogénéité naturelles de la roche. Malgré leur importance pour les applications d'ingénierie, ces conditions in situ restent insuffisamment représentées dans les ensembles de données expérimentales existants et ne sont que partiellement prises en compte dans les approches de modélisation actuelles.
Sur le plan expérimental, le projet proposé combinera des essais triaxiaux vrais particulièrement innovants sur des éprouvettes prismatiques de grès creuses, avec un dispositif d'imagerie in operando permettant de suivre l'apparition et la propagation de la déformation localisée lors d'une dépressurisation contrôlée de la cavité. Une évaluation systématique de la polarisation du champ de contraintes, de l'endommagement près de la cavité, de la transition vers la localisation, ainsi que des effets de l'anisotropie et de l'hétérogénéité du matériau sous des chemins de contraintes représentatifs des conditions d'excavation, permettra d'obtenir un ensemble de données expérimentales inédit.
Du point de vue numérique, l'étude commencera par comparer des modèles classiques d'éléments finis élastoplastiques aux résultats expérimentaux et évaluera la capacité des lois de comportement classiques à reproduire la réponse de la roche lors d'une dépressurisation de cavité sous chargement multiaxial. Une seconde composante, multi-échelle, s'appuiera sur un modèle hiérarchique FEM×DEM intégrant les mécanismes à l'échelle du grain (perte de cohésion, fragmentation des grains, réarrangement du réseau granulaire) et des distributions d'hétérogénéité contrôlées. L'étude numérique apportera de nouvelles perspectives sur la capacité des modèles multi-échelles avancés à reproduire les mécanismes gouvernant le développement des dommages autour des excavations profondes. Cette étude apportera également une méthodologie alternative afin réaliser des modèles prédictifs capables de prendre en compte des chemins de contraintes réalistes et la complexité microstructurale.
The characterization of localized deformation and its evolution in porous rocks surrounding underground cavities is a major challenge for assessing the risks and feasibility of deep boreholes and galleries associated with the energy transition (e.g., CO sequestration, hydrogen storage, geothermal systems, nuclear waste management). The excavation of such cavities within rock masses that have been mechanically stable over geological timescales induces significant perturbations in the near field stress distribution, potentially generating damaged zones in the surrounding rock and compromising the stability and functionality of the storage repositories, including a perturbation of the permeability field. The initiation and development of localized deformation in these zones are strongly influenced by 1/ the evolution of three-dimensional stress states-often following complex loading paths- 2/ the material anisotropy of the surrounding rock and 3/ the spatial variability of the mechanical properties inherent to natural sedimentary formations. Depending on the in situ loading conditions and rock properties, damage in localized areas can increase permeability in the excavation near field or, on the contrary, seal off the vicinity of the excavation by pore closure and connectivity reduction induced by compaction.
Current improvements in our understanding of the mechanisms involved and in modeling tools are hampered by the very limited experimental data available in the literature and, consequently, by the lack of validation of modeling tools, despite all the efforts made to develop predictive models.
Laboratory tests on excavation stability are generally conducted under axisymmetric loading conditions, on hollow cylinder specimens (e.g., Cuss et al., 2003, Haimson, 2007, Meier et al., 2013). Although this type of loading is relatively simple to implement, it does not take into account the important role of the multi-axial external loading (far-field stress anisotropy) on both the initiation location and development of strain localization in the excavation near-field. In addition, it applies loads to the material along paths with fixed principal stress directions, which again is not representative of real conditions. Tests carried out on hollow (prismatic) specimens under true triaxial loading conditions are of higher interest but remain the exception (Papamichos et al., 2010b, Lu et al., 2021). However, as for the tests on hollow cylinders, the observations of failure pattern are only made post mortem, after the specimens have been dismantled. It has now been identified that the final fracture surface on hollow prismatic specimens can be strongly affected by significant edge effects and can distort the understanding of damage initiation and strain localization evolution before the full specimen failure (Mukerhjee et al., 2025). In order to overcome this problem and to provide a greater interest in modeling improvement, in operando full field measurements on hollow prismatic specimens are a good alternative to this problem. They begin to appear in the literature, but with loading conditions that are still simplistic, generally with plane stress state (Yao et al., 2025) and not representative of real in situ stress conditions. To the best of our knowledge, there is no experimental configuration with two parallel galleries, although these are interesting configurations for analyzing the interaction of cavities' near-fields, although numerical studies have been carried out on this subject (Patutin et al., 2022).
In terms of modeling, the analysis of damage and strain localization in the excavations near-field has been the subject of numerous studies and benchmarks, as it corresponds to major application challenges. Outside the choice of a constitutive law to describe the rock behavior, numerous parameters are taken into account, such as, to name the main ones, the anisotropy of the far-field stress (Hu et al., 2017, Papamichos, 2010a), the material anisotropy of the rock (Gerolymatou et al., 2025), the thermo-hydro-mechanical couplings (Pardoen et al., 2014, François et al., 2012) and loading rates. In most cases, the constitutive models are calibrated from laboratory tests such as axisymmetric triaxial tests, which are relatively far removed from the loading paths followed during modeling. A regularization is added to objectively model strain localization. Modeling results, such as hole deformations or failure thresholds, are sometimes compared with experimental data obtained from observations or measurements inside the hole. As for strain localization patterns in the hole near-field, they are not validated due to the lack of available experimental data. At best, the numerical non-uniqueness of strain localization pattern is discussed (Marinelli et al., 2015, Gerolymatou et al., 2025). There is a clear lack of data from structural tests (as opposed to elementary tests) to validate these modeling studies and compare model results with experimental tests. Such tools could and should complement the identification of constitutive laws based on basic elementary laboratory tests, they would enable the models to be enriched by the fact that the stress paths imposed during these tests are more complex (and more realistic) than the paths followed on the elementary tests.
The role of material heterogeneities in the initiation and development of strain localization is also generally overlooked. Heterogeneities in natural rocks are very often multi-scale, from the grain scale to the metric scale, via centimetric variability. Areas of weakness (e.g., zones of higher porosity) can act as nucleation points for strain localization, and thus influence the localization pattern (Doré-Ossipyan et al., 2025). This leads to a certain non-reproducibility of tests at the experimental level, which should be taken into account at the numerical level to study the impact on damage and strain localization around excavations (Marimos et al., 2019).
In this context, it is essential to deepen our understanding of the role played by rock heterogeneity and anisotropy in the mechanisms of strain localization and damage propagation around deep excavations. The proposed study aims to enhance design and stability assessment strategies for such underground structures.
This PhD research will investigate, through combined experimental and numerical approaches, the mechanisms of localized deformation around boreholes and galleries in heterogeneous and anisotropic sedimentary rocks, under true triaxial stress paths representative of in-situ conditions.
The experimental part of the thesis is aimed at fine characterization of rock behavior in the cavity near-field in specimens subjected to multiaxial far-field loading. In an unprecedented way, the tests will combine true triaxial loading, hollow specimens with pressurized cavity (one or even two parallel cavities) and in operando full field measurement. They will be used to build up a unique database on damage modes and strain localization during gallery excavation under well-controlled loading conditions. It will consist in laboratory testing of rock specimens with a central circular cavity, representing the underground geometry of the rock around boreholes and galleries. The specimens will be installed inside the true-triaxial device and loaded to different external stress levels, preferentially in plane strain conditions, with the internal cavity simultaneously pressurized to achieve the initial stress state. Note that the experimental device is already fully operational and does no need preliminary modification. To reproduce the excavation phase, the internal cavities will then be depressurized, modifying the stress distribution in the specimen. During the entire loading and depressurization phases, digital images will be captured of the specimen surface, perpendicular to the cavity. Digital image correlation (DIC) will provide access to incremental deformation fields on the whole specimen surface, enabling the identification of early and mature deformation modes during initial loading and subsequent cavity unloading, identification of the damage zone extension, as well as the influence material anisotropy on the deformation patterns. Particular attention will be devoted to the characterization on initial material heterogeneities, via systematic X-ray tomography of pre-test specimens, to study their impact on the initiation and development of localized failures.
For the experimental campaign, several loading paths and material configurations are planned. Regarding the stress state, one test configuration will be performed for an external stress in an isotropic condition and two test configurations in external deviatoric stress states of different magnitudes, for cylindrical cavity aligned with the intermediate or the major principal stress direction. In all cases, cavity depressurization will occur after the desired stress state will be reached. Sandstone is selected here because it is a good candidate as a porous rock model material, and for which we already have an extensive database of elementary tests. For the rock material, an isotropic and an anisotropic (bedded) sandstone will be investigated. The anisotropic sandstone will further be tested at three different orientations to the principal stresses for a single deviatoric loading condition. Prior extensive characterization of the two selected sandstone in loading experiments of homogeneous prismatic specimens will provide valuable comparative measurements. Therefore, twelve different experiments will be performed in the scope of the PhD thesis, with a plan to duplicate two experiments after all initial configurations have been performed.
The numerical part will consist of two parts. The first, relatively modest part, will consist in modeling the boundary problem of gallery excavation with an existing elasto-plastic model (e.g. Plasol in Lagamine FEM code), associated with an enriched continuum formalism. We will use this classic model for geomaterials, for which we have constitutive parameters identified on the basis of elementary tests, with particular attention to explicitly reproducing the localized post-peak stress phase (El Moustafa, 2014). The exercise will allow us to compare simulations with experiments, and to analyze the contribution of structural testing to improving model identification.
The second, more ambitious, part will use a hierarchical model approach based on computational homogeneization FEMxDEM. This advanced numerical approach, developed at laboratoire 3SR (Desrues, 2019), enables a close and concurrent link between the macroscopic boundary value problem (BVP) simulated by the Finite Element Method (FEM), and the micromechanical behavior of the material using the Discrete Element Method (DEM). While a numerical FEMxDEM implementation is fully functional and has been used to simulate the behavior of sandstone in previous work (Couture, 2020), some further improvements will be made: (1) integration of an existing particle breakage version of the DEM model (Richefeu, 2025) used in this framework - the model implementation part will be carried out mainly by the supervisors, (2) a controlled heterogeneous distribution of elementary volumes over the FEM domain (in the spirit of Marchadour's current PhD thesis work), (3) the preparation of DEM initial state based on sandstone microstructures.
The first objective of this numerical part will be to evaluate the performance of the FEMxDEM model in qualitatively reproducing experimental observations in the different experimental conditions: stress-strain evolution, emergence and propagation of localization zones and failure modes. The performance of the model will be evaluated in comparison to the more classical elasto-plastic model. In the case of the FEMxDEM model, the micro-structure data will also provide additional insight into the interplay between cohesion damage, grain breakage and grain reorganization inside localized zones and close to the cavity.

Le ou la candidat.e doit avoir une solide formation en mécanique des solides, une certaine familiarité avec la géomécanique, ainsi qu'un enthousiasme pour les méthodes expérimentales non-conventionnelles et les simulations numériques. Le ou la candidat.e idéal doit avoir une expérience en programmation ou d'analyse de code dans des langages de programmation scientifique tel que Matlab/Python ou Fortran/C/C++. De plus, une expérience dans la génération et l'analyse de données expérimentales (e.g., méthodes d'imagerie, instrumentation, préparation d'échantillons, etc.) est souhaitable. De solides compétences en communication orale et écrite en anglais sont indispensables.