Speaker: Mariam Joundi

Date: Thursay 30th of January 2025, 1:15pm.

Abstract:

Dans le contexte de l'ambition climatique de l'Union Européenne visant la neutralité carbone à l'horizon 2050, l'hydrogène vert émerge comme un vecteur énergétique prometteur. Cette orientation stratégique nécessite l'élaboration de solutions innovantes pour le transport sécurisé de l'hydrogène (European Hydrogen Backbone, 2020). En France, une réorientation des infrastructures est envisagée, avec la proposition de réutiliser les réseaux de gaz existants pour le transport de l'hydrogène (GRTgaz, 2019). Cependant, cette substitution induit des modifications significatives des caractéristiques mécaniques des réseaux, augmentant ainsi leur vulnérabilité vis-à-vis les sollicitations externes (Boots et al., 2021). Dans le cadre d'une thèse, nous explorons l'interaction entre ces réseaux et les mouvements de terrain, dans le but de minimiser les risques associés.

Dans ce contexte, cette étude vise à développer un méta-modèle basé sur une modélisation numérique tridimensionnelle pour évaluer l'impact des affaissements de terrain sur les réseaux de gaz enterrés. Le méta-modèle développé cherche à quantifier le taux de transmission des affaissements de sol aux conduites, en fonction d’un coefficient de rigidité relative sol-conduite. Ce travail s'inscrit dans une démarche novatrice par son approche tridimensionnelle, offrant une perspective plus complète par rapport aux méta-modèles similaires développés dans la littérature, se reposant sur des modèles de base analytiques ou numériques (Joundi et al., 2023).

Une analyse numérique paramétrique, intégrant un total de 230 simulations par éléments finis (EF), a été développée, aboutissant à l'élaboration d'un premier méta-modèle. Ce méta-modèle établit une relation entre la courbure relative des pipelines et un coefficient de rigidité relative du système sol-pipeline, constituant ainsi une base robuste pour des évaluations probabilistes futures. Ce modèle a été comparé au méta-modèle de Wang et al. 2011 afin de positionner nos résultats dans le contexte des recherches existantes.

Des premières simulations de propagation d’incertitudes ont été menées pour évaluer la probabilité de défaillance des conduites exposées à un affaissement, aboutissant à l’établissement de premières courbes de vulnérabilité.

Details

Category: Seminar

Speaker: Bastien Morin

Date: Thursay 23rd of January 2025, 1:15pm.

Abstract:

Les réservoirs miniers, formés par l’ennoyage des travaux après l’arrêt de l’exploitation du charbon, sont le siège de processus géochimiques complexes. Dans la mine de lignite de Gardanne, la roche exploitée se mêle à des formations calcaires, entraînant des interactions eau-roche spécifiques qui influencent les équilibres géochimiques. Les concentrations en fer, issues du drainage minier neutre, y posent d’importants défis environnementaux. Cette présentation décrit la campagne de terrain menée au puits Gérard à Mimet fin octobre 2024 dans le cadre de mes travaux de thèse. Au cours de cette campagne, des mesures physico-chimiques détaillées ont été réalisées, accompagnées de prélèvements d’eau à différentes profondeurs, suivis d’analyses chimiques, isotopiques et microbiologiques.

Details

Category: Seminar

Speaker: Michael Brun

Date: Friday 17th of January 2025, 1:15pm.

Abstract:

The increasing complexity of numerical models for engineering systems, taking into account ever finer mesh, accurate material models and multiphysics phenomena, occurring at very different space and time scales, boosted the development of partitioning approaches. In the field of structural dynamics, the main advantage of partitioning approaches is to gain in computational efficiency by using different time integrators with their own time step, depending on the non-overlapping subdomains composing the global mesh. It leads to the proposition of Hybrid Asynchronous Time Integrator (HATI). A large variety of HATI applications is then presented in the fields of structural dynamics and earthquake engineering, including smooth dynamics without contact as well as non-smooth dynamics with frictional contact phenomena.

First, HATI is employed for Reinforced Concrete structures under earthquake loading, without contact, using multi-fibre beam elements whose cross-section is divided into steel and concrete fibres associated with cyclic and nonlinear behaviors. Then, the frictional contact is explored by developing an explicit time integrator in Moreau’s framework, called CD-Lagrange scheme, based on the velocity-impulse formulation with the contact constraints enforced using Lagrange multipliers.  The CD-Lagrange scheme is adopted for reproducing the frictional contact phenomena experiencing by crane bridge under earthquake loading following the HATI approach. It enables us to focus on high frequency phenomena with frictional contacts in small subdomains including the rails of the girder beams and the trolley wheels, using small time steps satisfying the CFL condition, whereas the main part of the complex problem is dealt with a classical implicit time integration scheme with large time steps. In this way, we benefit from advantages of the two time integrators, leading to computation time gains as well as versatility in applying appropriate damping formulations depending on explicit or implicit subdomains. CD-Lagrange scheme is also adopted for simulating the eccentric pounding phenomena between real scale two-storey steel-concrete structures, tested on the shaking table AZALEE (CEA Saclay, France), enabling us to obtain very satisfactory results with respect to experimental data, in terms of floor displacements and acceleration spectra in a large frequency range.

Soil Structure Interaction (SSI) problems are also considered by developing Multi Time Step Hybrid Perfectly Matched Layers (PML) in the context of the simulation of wave propagation in the time domain for 2D and 3D unbounded domains. Indeed, HATI approach is useful to integrate in time the complex space semi-discrete equations in the 2D/3D PMLs, independently from the time integration of the classical semi-discrete equations of motion in the domain of interest. Hybrid PMLs are also employed in research developed within LSMS laboratory, EPFL, concerning the simulation of an unbounded elastic block on a rigid flat plane: It involves rate and state friction laws at the interface, compared to classical Coulomb’s law with constant friction coefficient.  

On the basis of the HATI approach, multi time step explicit/implicit co-simulation strategies are developed in order to couple a seismology software (EFISPEC3D, BRGM) based on 3D Hexahedral Spectral Elements with FE software (Akantu, EPFL) in the context of SSI applications: A seismic non-linear analysis of a concrete gravity dam is carried out using the HATI approach combined with mortar coupling to deal with non-matching meshes at the interface between the subdomains. The seismology software simulates the seismic source as well as the large medium concerned by the wave propagation, using an explicit time integrator, whereas the non-linear behavior of the dam is simulated by the FE software with an implicit time integrator.  

To conclude, the presented works aim at developing numerical coupling approaches in order to tackle complex problems in structural dynamics. The conducted research has deep roots in Computational Mechanics while focusing on Earthquake Engineering problems in a constant effort to apply the developed innovative tools to engineering systems.

Details

Category: Seminar

Speaker: Ayoub Belhachmi

Date: Thursay 09th of January 2025, 1:15pm.

Abstract:

Constructing a geological numerical model is a key step in studying and exploring the subsurface. These models are constructed from seismic or well data, which consist of data points associated with values corresponding to their geological ages. This task involves constructing an implicit function, also known as the stratigraphic function, which interpolates this set of data points. Often, the available data are sparse and noisy, making this task difficult, especially for reservoirs where the geological structures are complex, with distinct discontinuities and unconformities.

In this seminar, we present a novel method for computing the stratigraphic function, which represents geological layers, using piecewise quadratic
𝐶1 splines on triangulations, specifically Powell-Sabin splines. This method enables better handling of geological layers with strong curvatures and reduces mesh resolution while maintaining high smoothness and regularity.

A key focus of our work is regularization, the most challenging component of mesh-based implicit modeling approaches. Classical methods often fail for data with high thickness variations, producing inconsistent models with bubble effects. To address this, we propose two innovative regularization energies inspired by fundamental PDEs: the anisotropic diffusion equation and the bending equation of an anisotropic thin plate. In both approaches, tensors of diffusion or rigidity are iteratively adapted to the data's variations and anisotropy.

Details

Category: Seminar

Speaker: Bahaa Abou Chakra

Date: Thursday 19th of December 2024, 1:15pm.

Abstract:

Les argilites du Callovo-Oxfordien, en tant que barrières géologiques potentielles pour le stockage des déchets radioactifs en France, sont soumises à des sollicitations couplées, comme les variations de contraintes pendant l'excavation, les changements de saturation, les variations thermiques dues aux déchets exothermiques et des processus chimiques. Pour étudier la faisabilité du stockage géologique à court et long terme, des modèles couplés et des caractérisations expérimentales THMC (Thermo-Hydro-Mécanique-Chimique) doivent être développés. Cette étude évalue l'effet thermique sur le comportement mécanique de l'argilite du Callovo-Oxfordien à travers des essais triaxiaux, simulant les conditions de stockage in situ, avec différentes pressions de confinement et différentes orientations, parallèle et perpendiculaire au plan de litage. L'objectif aussi est de caractériser et quantifier les variations morphologiques et volumétriques des fissures, mesurées par tomographie 3D aux rayons X, pour évaluer l'impact de l'endommagement sur l'intégrité structurelle des échantillons.

Details

Category: Seminar

Speaker: Amandine Fratani

Date: Thursday 28th of November 2024, 1:15pm.

Abstract:

When creating a geological model from borehole data or 2D sections, the interpretation of 3D faults is often ambiguous and uncertain. This work focuses on the problem of associating partial fault observations, which has recently been formalized using a graph in which each fault observation is represented as a graph node, and graph edges carry the potential of pairwise associations. The likelihood of an association is computed using selected expert geological rules. However, fault observations are not pairwise independent, which prevents the consideration of higher-order effects such as the distribution of the throw or the length along several aligned nodes. To complement this approach, we propose a mathematical formalism for the use of high-order associations. The definition of expert rules in a multiple-point problem is challenging because of the very high dimensionality of the problem. To alleviate this, we propose to supplement expert rules by supervised machine learning using analog or incomplete interpretations. This work uses a Random Forest learner trained from a set of selected fault features computed from fault traces extracted from known 3D geological models (e.g., the length of the fault trace, the throw value, etc.). The association likelihood inference is formulated as a classification problem to determine the probability that fault observations belong to the same fault object based on the variables computed from the features of the k observations. To prevent overfitting, we propose to mimic a partly interpreted case: we split the 3D domain in two disjoint, spatially contiguous sectors A and B, and use sector A as training and sector B for testing. Preliminary results demonstrate the ability of Random Forest to retrieve probabilities of triplets that complete the pairwise representation.

Details

Category: Seminar

Speaker: Pauline Collon

Date: Thursday 21tst of November 2024, 1:15pm.

Abstract:

L'activité minière a des conséquences évidentes sur les paysages et la stabilité des sols sus-jacents, mais aussi sur la qualité de l'eau du fait d'un phénomène nommé "Drainage Minier". Dans ce séminaire  je reviendrais sur les travaux réalisés en 2000-2005 dans le cadre du GISOS pour caractériser le drainage minier neutre à l'oeuvre pour les mines de fer lorraines, en décrivant les moyens expérimentaux et numériques qui avaient alors été utilisés pour prédire l'évolution de la qualité de l'eau en sortie de bassin. Ces travaux servent aujourd'hui de "référence" méthodologique pour l'étude d'un autre drainage minier, celui de la mine de lignite de Gardanne (Marseille), sujet de la thèse de Bastien Morin (BRGM - GeoRessources/RING).

Details

Category: Seminar

Speaker: Marius Rapenne

Date: Thursday 14th of November 2024, 1:15pm.

Abstract:

Docker is an open-source platform as service that provide OS-level virtualization of packages to allow the distribution of software in containers. It allows for a quick and easy deployment of any application in isolation in different environment. The goal of this seminar is to provide a quick overview of virtualization and its uses, and then focus on docker, the creation of docker images and containers, through some exercise.

Details

Category: Seminar

Speaker: Mohammad Mahdi Rajabi

Date: Friday 8th of November 2024, 1:15pm.

Abstract:

To address common challenges in neural networks—such as large data requirements, poor generalization, overfitting, lack of transparency, and physically unrealistic outputs—incorporating physical intuition into different stages of neural network design has proven to be highly effective. This approach leverages the strengths of neural networks while ensuring their outputs adhere, fully or partially, to established physical laws. As a result, it improves the reliability, interpretability, and practicality of neural networks and can reduce the need for vast training datasets. This methodology is particularly useful for modeling physical systems, such as those in solid and fluid mechanics, as well as cyber-physical systems like smart grids and autonomous vehicles. With the increasing number of techniques and publications in this area, a clear and structured review of these methods has become essential. I will present an overview of current methods, terminology, and best practices for integrating physics into neural networks, providing a detailed classification of approaches including pre-training integration, in-training integration, and architecture-level embedding. I will also discuss the limitations of existing methods and highlight promising directions for future research.

Details

Category: Seminar