Internal erosion is a primary cause of degradation and failure in hydraulic structures and natural deposits. While the finite element method (FEM) is widely adopted in field-scale investigations, most existing studies are predominantly limited to two-dimensional (2D) cross-sectional analyses. This simplification proves particularly inadequate when examining complex stratigraphic configurations or specific erosion phenomena including concentrated leaks and sand-boils. This study comprehensively investigates a river dike in southern France. By integrating geophysical imaging data, we developed both a large-scale three-dimensional (3D) geometry model and a series of representative 2D cross-sectional models. A transient FEM that couples multi-field and multi-phase suffusion processes was employed to simulate seepage and fines migration within the dike foundation under realistic flood conditions. Comparative results demonstrate non-negligible discrepancies between 2D and 3D simulations. The 2D analysis demonstrates significant underestimation of suffusion effects by neglecting transverse seepage, while also failing to capture complex flow dynamics inherent in heterogeneous stratigraphic conditions. In contrast, the 3D simulation results aligned closely with field observations, intuitively elucidating the formation mechanisms of sinkholes, leaks, and sand boils while providing clear insights into their complex spatial distribution patterns. The 3D analysis further reveals the complex seepage and suffusion evolution governed by stratigraphic morphology and highlights the interplay between subsurface suffusion and surface signatures of internal erosion. These findings underscore the superiority and necessity of 3D modeling in internal erosion studies for sites featuring complex geological stratification.

The subsoils of river dikes are often composed of highly permeable and low-density river sediments. Thus, erosion signatures (leaks, sand boils, sinkholes) can appear in the protected floodplain during floods, highlighting the development of hydromorphodynamic phenomena below the surface, which may harm the safety of the dike system. A multiscale methodology is deployed to understand and analyze the influence of floodplain architecture in terms of geological formations on the appearance of local erosion signatures. Particular attention is paid to the morphology of paleovalleys and paleochannels in order to image the subsurface in terms of substrate types and interfaces using geophysical methods. This information makes it possible to propose internal erosion scenarios. Application to a study area in the South of France (the Agly dike system) leads to new results. The classical backward erosion piping scheme is not relevant to explain the observed sand boils, as they are mainly caused by the suffusion-type internal erosion process. Suffusion and contact erosion appear to be the origin of sinkholes. The distribution of these signatures appears to be directly related to the shape and dimensions of the paleovalley and paleochannels, as well as to the presence of a low-permeability topsoil.

The angle of repose (AoR) is a widely employed measurement for granular materials that is currently used by various industries to characterise the behaviour of powders, soils and grains. This study proposes a new measurement technique for the AoR, and applies it on numerical models of two recently developed devices to measure the AoR by means of Discrete Element Simulations, for plane-strain and axisymmetric repose states. 3D-printed rounded tetrahedral particles are simulated using two numerical modelling strategies, namely concave multi-spheres (clumps) and convex potential particles, to explore the effect of particle shape on the AoR, and comparisons are drawn against experimental evidence. Then, a campaign of drained triaxial compression tests is carried out for a wide range of confining pressures to investigate if any correlations exist between the AoR and constitutive (peak and critical-state) friction angles.

Internal erosion poses a significant threat to the safety of the water-retaining structures. Due to its subterranean nature, the process of internal erosion is difficult to observe directly and evaluate properly, even when typical erosion signatures are captured at the surface. Therefore, predicting internal erosion or elucidating the underlying mechanisms of superficial erosion indicators remain considerable challenges in geotechnical engineering. In this study, a numerical investigation was conducted to explore the mechanism responsible for the numerous erosion signatures (sinkholes, sand-boils, and leaks) observed at the Agly dike, in France, through a cooperated geophysical-numerical approach. Simulated configurations were mapped according to the actual stratigraphic structures identified by geophysical results (EMI and ERT). A multispecies transport finite element method was employed to examine the seepage and suffusion dynamics under periodic flooding conditions. Systematic comparative analysis demonstrated that the unique stratigraphic structures and geometries played a governing role in the occurrence of erosion signatures in situ. Specifically, the paleo-channel throat promotes sinkhole formation via suffusion and contact erosion, while confined pore pressure acting on the low-permeable surface layer induced the emergence of sand-boil and leak. Our findings underscore the crucial role of stratigraphic features on internal erosion processes and demonstrate the effectiveness of integrating geophysical investigations with numerical modeling for assessing internal erosion risks in engineering applications.

La prédiction de l’érosion par surverse des digues de protection contre les crues et des barrages en remblai constitue l’une des difficultés majeures dans l’analyse de la sûreté de ces ouvrages. Bien qu’identifié comme étant le mécanisme de rupture dominant de près de la moitié des ruptures de digues fluviales, les méthodes et outils dont disposent les ingénieurs pour le caractériser, de manière déterministe, restent encore très limités. Notamment, la caractérisation du processus d’érosion, en termes géométriques et temporels, prenant en compte le comportement des matériaux constitutifs du remblai et de sa fondation et les spécificités des sollicitations hydrauliques sur les digues, n’est encore à la portée d’aucun outil de modélisation numérique validé au niveau requis. Les ingénieurs doivent donc, pour l’instant, se contenter d’approches recourant largement à l’empirisme : critères dits « de jugement d’expert » ou formules empiriques. Ces approches très simplifiées sont connues pour présenter de fortes incertitudes et ne pas garantir des résultats systématiquement du côté de la sécurité. C’est dans le but final d’améliorer significativement la robustesse et la précision des outils de prédiction de l’érosion par surverse pour les ingénieurs, qu’EDF et CNR ont décidé de lancer le projet de recherche Overcome. L’objectif de ce projet est d’inclure au sein de la plateforme numérique open source TELEMAC-2D des modules représentant différents types de processus physiques en lien avec les ruptures par surverse de digues, basés sur une approche expérimentale multi-échelles très poussée. Cette communication présente les tous premiers résultats de l’un des premiers sujets investigués par ce projet : la description des mécanismes physiques d’érosion par surverse des remblais constitués de sol grossiers à granulométrie étalés. Ces sols sont présents dans une grande partie des digues situées en zones montagneuses ou dans les vallées situées juste en aval. Une première série d’essais de surverse à petite échelle a été menée au laboratoire de l’Université Polytechnique de Madrid, en réalisant des remblais homogènes de 55 cm de hauteur à partir de trois matériaux différents : un sable assez uniforme, un gravier de granite et un sol alluvionnaire à granulométrie étalée, issu du site CNR de Montfaucon. Si les essais avec le remblai en sable ont confirmé le mécanisme attendu d’érosion de surface, il a été assez surprenant de voir des différences significatives de mécanisme d’érosion entre les matériaux de Montfaucon et graviers de granite. Dans le matériau de Montfaucon, le mécanisme s’apparente à de l’érosion de surface, tandis qu’avec le gravier de granite, le mécanisme d’érosion débute avec du « Headcut migration » puis évolue à la fin en érosion de surface. Par ailleurs, le sol de Montfaucon s’est avéré quatre fois moins érodable que le gravier de granite. Ces premiers résultats appellent à être confirmés par des essais de répétabilité à petite échelle et des essais à plus grande échelle.

To simulate large, history-dependent material displacements, the Material Point Method (MPM) solves for the kinematics of Lagrangian material points being embedded with mechanical variables while moving freely within a fixed mesh. The MPM procedure makes use of the latter mesh as a computational grid, where the momentum balance equation with the acceleration field are first projected onto nodes, before material points can be moved. During that process, a number of different choices have been adopted in the literature for what concerns the computational definition of time increments of velocity and position, from the knowledge of nodal acceleration. An overview of these different motion integration strategies is herein proposed, with a particular emphasis on their impact onto the MPM conservative properties. Original results illustrate the discussion, considering either simple configurations of solid translation and rotation or a more complex collapse of a frictional mass. These analyses furthermore reveal hidden properties of some motion integration strategies regarding conservation, namely a direct influence of the time step value during a time integration being inspired by the Particle In Cell (PIC) ancestor of the MPM. The spatial, resp. temporal (in comparison with vorticity), discretizations are also shown to affect the angular momentum conservation of the FLIP method, resp. an affine extension of PIC (APIC).

The study of internal erosion of earth dams, dikes, and levees is an essential ingredient to address their safety. Indeed, internal erosion is known to be a major cause of failure for dams and levees. Several sections of the Agly river dike (south-west France) are affected by the appearance of sand boils on the side of the protected zone. These phenomena generally occur without any significant erosion of the dike itself. In order to better understand the reasons for these phenomena, we took advantage of a drought (dry riverbed) to image, using induced polarization tomography, the subsurface from the riverbed to the plains extending inland behind the areas prone to these pathologies. Induced polarization imaging is a geophysical technique that extends the classical electrical resistivity tomography to include low-frequency polarization mechanisms. This approach shows that the river dike is partly built on a sand-filled paleochannel, allowing water to flow under the dike and its clay core during river floods. During floods, underground flow under the dike is responsible for sand boil phenomena on the side of the protected zone. Numerical hydraulic simulations show that the discharge zone has specific concentration points, particularly at the toe of the dike even when the permeability of the sand filling the paleochannel is homogeneous. These concentrated discharges are controlled by the water level in the river, the geometry of the dike and the shape of the clay-sand interface of the paleochannel. This study highlights the role of geophysical techniques, especially induced polarization, in providing key information to better understand erosion and fluidization phenomena affecting river dikes.

The repose of granular materials is investigated via two different Discrete Element Method (DEM) implementations in comparison with an experimental reference from a recently proposed benchmark setup. On a methodological standpoint, a rigorous measurement method of the angle of repose (AOR) is first proposed for plane-strain and axisymmetric conditions as encountered in the reference experiments.Additionally, two systematic procedures are designed in order to also determine the void ratio of the heap, as a fundamental property of granular matter possibly influencing the AOR. A physical discussion is then developed on the role of particle shape, considering the non-spherical nature of reference particles with a convexity value of C = 0.954 . Adoping non-convex multi-spheres aggregates (i.e. clumps), the first DEM modelling approach successfully predicts the AOR within a 8 35.95 ± 0.88^∘ to 31.26 ± 0.95^∘ . For the loading setup(s) at hand, the AOR is eventually shown to bear no constitutive nature. It is for instance independent of initial void ratio but is still different than the critical friction angle. The latter may actually serve as a lower bound for the process-dependent AOR. These conclusions are drawn from a statistical analysis of a large set of results, accounting for the random nature of the microscopic arrangement in the studied process.

Dans le Bulletin 195 Barrages en Sol Cimenté, récemment publié par la Commission Internationale des Grands Barrages (CIGB), le terme « Sol Cimenté » désigne tout sol lié, que ce soit suite à un traitement à la chaux, au ciment, ou à la chaux en association avec le ciment, le cas échéant avec addition de composants pouzzolaniques. Le bulletin fait aussi mention des liants hydrauliques routiers. En Europe, où ces produits sont normalisés et contrôlés, ils peuvent avantageusement remplacer les ciments. Le traitement des sols est une technique ancienne et éprouvée qui a connu au cours des soixante dernières années un développement spectaculaire, en particulier dans le domaine des infrastructures de transport. Des exemples d’applications existent également dans le domaine des ouvrages hydrauliques, ce qui a conduit la CIGB à publier le Bulletin 54, Sol-ciment pour barrages en remblai (1986), consacré à l’utilisation de matériaux granulaires traités au ciment pour la protection des talus amont des barrages en terre. Depuis les années 1980, la technique a considérablement évolué grâce aux progrès de la technologie d’une part, et à une utilisation plus rationnelle de la chaux aérienne calcique d’autre part. En outre, d’importants programmes de recherche et des retours d’expériences positifs ont permis de mieux explorer et caractériser les performances des sols traités et d’en tirer un meilleur profit. Le nouveau Bulletin 195 couvre l’utilisation des sols cimentés dans la réalisation d’ouvrages hydrauliques. S’il s’appuie sur le Bulletin 54 exclusivement consacré au ciment, il le complète avec l’utilisation de la chaux aérienne calcique, permettant ainsi d’élargir le champ d’application de la technique au traitement des sols limono-argileux. S’agissant de la valorisation de matériaux naturels, donc non élaborés, il insiste particulièrement sur l’importance des études préalables : reconnaissance des gisements et études en laboratoire. Profitant de l’expérience acquise dans les infrastructures de transport, il étend largement l’utilisation des sols traités à d’autres applications que les protections de talus amont, notamment aux barrages homogènes de hauteur moyenne (jusqu’à 30 m de haut), aux barrages zonés et aux digues. Il fournit des recommandations utiles pour la conception, la construction et la réparation des barrages et des digues faisant usage de sols cimentés.

Ce travail porte sur les mécanismes physiques à l’origine des processus d’érosion interne dans les sols de fondation des digues de protection contre les inondations. La présence de matériaux perméables est le plus souvent associée à la présence d’une paléo-vallée comblée de sédiments alluviaux sous le lit de la rivière et sous les digues ou de paléo-chenaux sableux pouvant s’étendre au niveau de la zone protégée. Le cas des digues de l’Agly montre que plusieurs processus d’érosion interne doivent être pris en compte pour décrire ces phénomènes : l’érosion régressive, l’érosion de contact et la suffusion. L’utilisation combinée des méthodes d’induction électromagnétique (EMI) et de tomographie de résistivité électrique (ERT) est une solution rapide et peu coûteuse qui permet d’imager le sol et de fournir la géométrie des différentes couches. Combinées à des sondages carottés, les résultats obtenus permettent de localiser la profondeur des interfaces et de mettre en évidence plusieurs scenarii possibles d’apparition de résurgences, de sand-boils et de fontis en bordure de digue, ou au niveau du val protégé.

This work focuses on the mechanisms that trigger internal erosion of the pervious foundation of flood protection dikes. The origin of these permeable layers is generally attributed to the presence of a paleo-valley and paleo-channels filled with gravelly-sandy sediments beneath the river bed and dikes. These layers may extend into the protected area. Visual observations of leaks, sand boils and sinkholes in the protected area testify to internal erosion processes in the underground soil. Local geological conditions are part of the information to be sought to explain these processes: presence of permeable soils and position of interfaces. Results obtained on Agly dikes (France), using two classical geophysical methods (EMI and ERT), were analyzed using cored soils and showed that it is not enough to simply conclude to the presence of backward erosion piping. The possibility of internal erosion, such as suffusion or contact erosion, must also be considered as the cause of leaks, sand boils and sinkholes. As the results obtained are explained by the presence of a paleo-valley and paleo-channels beneath the river bed and dikes—commonly encountered in this context—the methodology presented and the results obtained are likely to be relevant for many dikes.

This paper presents an erosion interpretation of cohesive granular materials stressed by an impinging jet based on the results of a micromechanical simulation model. The numerical techniques are briefly described, relying on a two-dimensional Lattice Boltzmann Method coupled with a Discrete Element Methods including a simple model of solid intergranular cohesion. These are then used to perform a parametric study of a planar jet in the laminar regime impinging the surface of granular samples with different degrees of cohesive strength. The results show the pertinence of using a generalized form of the Shields criterion for the quantification of the erosion threshold, which is valid for cohesionless samples, through empirical calibration, and also for cohesive ones. Furthermore, the scouring kinetics are analysed here from the perspective of a self-similar expansion of the eroded crater leading to the identification of a characteristic erosion time and the quantification of the classical erosion coefficient. However, the presented results also challenge the postulate of a local erosion law including erodibility parameters as intrinsic material properties. The paper then reviews the main limitations of the simulation and current interpretation models, and discusses the potential causes for the observed discrepancies, questioning the pertinence of using time-averaged macroscopic relations to correctly describe soil erosion. The paper concludes addressing this question with a complementary study of the presented simulations re-assessed at the particle-scale. The resulting local critical shear stress of single grains reveals a very wide dispersion of the data but nevertheless appears to confirm the general macroscopic trend derived for the cohesionless samples, while the introduction of cohesion implies a significant but systematic quantitative deviation between the microscopic and macroscopic estimates. Nevertheless, the micro data still shows consistently that the critical shear stress does actually vary approximately in linear proportion of the adhesive force.