Soutenance

The cold compaction process of powders made up of ductile particles is widely used in the industry. During this forming phase, interparticle adhesion develops during densification, building up the mechanical strength of the manufactured parts. One of the main challenges of this process is the occasional occurence of failure during densification, due to highly deviatoric stresses in the vicinity of geometric singularities. These defects lead to the rejection of the manufactured parts. Despite widespread recognition of this problem, there are few studies in the literature focusing on the micro-mechanical understanding of these failure phenomena within this type of material involving highly deformable particles. Currently, there is no numerical model capable of predicting the formation of such defects. Only time-consuming trial-and-error procedures efficiently mitigate them. Given the often expensive and challenging nature of experimental campaigns along complex loading paths, the role of particle adhesion and deformability in rupture phenomena remains poorly understood.

Given these limitations, a numerical approach based on the multi-particle finite element method appears as a promising alternative. Such a method allows for the explicit modelling of the microstructure of an idealized granular medium, considered as a representative assembly of particles. The method corresponds to a coupling between the finite element method and the discrete element method\!: the particles are meshed in volume so as to fully account for their deformations through constitutive laws based on continuum mechanics; and the interactions between particle surfaces are handled using finite-element contact formulations. The main drawback of this method lies in its high computational cost, limiting the number of particles that can be modelled. However, by employing a multi-scale approach, it is possible to derive mesoscopic mechanical properties associated with an equivalent Cauchy volume element from the simulation of a relatively small number of particles, through homogenization techniques and formulation of appropriate boundary conditions. A study focusing on the analysis of the influence of boundary conditions on the simulated mechanical response of a numerical granular sample subjected to quasi-static loading is thus proposed.

Such a micro-mechanical approach requires an accurate description of contact interactions. Specifically, the mechanical strength of parts obtained through powder compression is significantly influenced by the development of adhesion at the contact level. Therefore, a multi-scale adhesive contact model, based on the weighting of the surface energy by a roughness model, is implemented in the multi-particle finite element code. This contact model allows for the local prediction (at each node of the finite element mesh) of the level of adhesion developed by external mechanical loads, which is consistent with the powder compaction process. The constructed numerical model is finally used to predict the mesoscopic properties associated with the equivalent Cauchy volume element. This numerical model enables the exploration of highly deviated loading paths that are inaccessible experimentally. Its purpose is to support the future development of continuum mechanics-based model for the modeling of a powder volume during compression.

Jury

• En savoir plus sur Thèse de Nils Audry

The shear behavior of rock-rock and concrete-rock interfaces is investigated for the design and structural stability assessment of geotechnical structures (tunnels, concrete-gravity dams). Most of the studies focused on the static shear behavior of rock-rock interfaces. These studies show that the shear resistance of rock-rock interfaces depends mainly on normal stress and roughness. However, the investigation of the static shear behavior of concrete-rock interfaces under low normal loadings shows that the concrete-rock bonds significantly influence the shear resistance beyond the normal stress and roughness. Despite this progress, more experimental and numerical work is required to further understand the shear behavior of concrete-rock interfaces. In particular, it is relevant to investigate the dynamic shear behavior of concrete-rock interfaces.

This thesis investigates the influence of roughness and concrete-rock bonds on the shear behavior of concrete-rock interfaces subjected to low normal loading. A two-fold approach combining experimental and numerical modeling is adopted. The experimental work includes a pioneering experimental campaign to investigate the dynamic shear behavior of concrete-rock interfaces focusing on impact shear loading.

The interpretation of recent experimental works, particularly concrete-granite interfaces, has distinguished the influence of the roughness scales on the shear behavior of concrete-rock interfaces. The unevenness (micro-roughness) is responsible for forming concrete-rock bonds. The waviness (macro-roughness) contributes to the shear resistance of concrete-rock interfaces through surface interlocking. The numerical modeling of these interfaces is based on this separation. The cohesive-frictional model simulates the influence of the unevenness and, therefore, the concrete-rock bonds. In contrast, the influence of the waviness is taken into account by the explicit representation of the 3D morphology of the interface through surface interlocking. This model is validated by performing a comparative evaluation of the results of experimental tests and the results of simulations of direct shear tests.

A new approach has been proposed to investigate the influence of roughness on the shear resistance of concrete-rock interfaces based on the validated numerical modeling. The approach consists of generating synthetic rough rock surfaces through random field simulations and performing simulations of direct shear tests of concrete-rock interfaces comprising the generated surfaces. The analysis of the results of these tests leads to the proposition of two failure criteria. The first is an analytical relationship, while the second is a neural network model.

The dynamic shear behavior of concrete-rock interfaces has been investigated as well. The dynamic shear loading is applied using a modified split Hopkinson pressure bar system with one input bar and two output bars. A modified confinement ring is used to generate low normal stresses. The outcomes of this study show that, like the static shear behavior, the dynamic shear behavior of concrete-rock interfaces depends on the concrete-rock bonds and friction. However, there is a double mobilization of concrete-rock bonds and friction in the pre-peak and the post-peak stages. The dynamic shear resistance is not strongly dependent on the normal stress. Furthermore, the dynamic shear resistance is three to four times the static shear resistance of interfaces.

The interpretation of the recent static experimental studies, the numerical simulations of rough concrete-rock interfaces, the analytical models proposed to estimate the shear resistance of concrete-rock interfaces, and the pioneering dynamic testing of concrete-rock interfaces have all provided new insights into the understanding of the shear behavior of concrete-rock interfaces.

Jury

• En savoir plus sur Thèse de Menes Badika Kiyedi

This work falls within the context of the aging of large concrete containment structures, such as reactor buildings in nuclear power plants. It aims at devising a numerical strategy for forecasting the long-term physical behavior of such structures, in order to better anticipate their maintenance.

Recent improvements of the understanding of physical phenomena behind aging and the increase of computational resources have enabled the development of numerical models aiming at simulating the Thermo-Hydro-Mechanical and Leakage (THML) behavior of large aging concrete structures. Nevertheless, the input parameters of such models are tainted with uncertainties, due to a lack of knowledge or to a natural randomness. Consequently, this thesis is mainly based on the general framework of Uncertainty Quantification (UQ), aiming at explicitly modeling uncertainties in numerical simulation. In this framework, the uncertainties tainting input parameters are typically modeled by probability laws, and are subsequently propagated through the model in order to study the variability of its response, or to estimate specific quantities of interest, such as moments or quantiles. However, THML computational models typically involve a large amount of uncertain parameters, most of them being not measurable directly. For this reason, the input probability law modeling their uncertainties is usually chosen in a subjective way, based on expert judgement. Then, this thesis is placed in the framework of Bayesian inference, in order to update a prior level of knowledge on input parameters from noisy observational data of the response of the structure under study.

Firstly, this work aims at coupling Bayesian inference with numerical techniques adapted with THML computational models, these last most often lying on costly finite element codes. In this perspective, a recent Bayesian computational framework is studied as an alternative to classical MCMC sampling techniques, and a coupling with surrogate modeling techniques is proposed in order to efficiently draw samples from posterior distributions with a reduced computational cost. Next, a general methodology aiming at performing probabilistic forecasts of the long-term THML behavior of containment structures is presented. In this context, the uncertainties of input parameters of the adopted THML model are quantified through Bayesian inference, from in-situ monitoring data. The proposed methodology is illustrated through a study of the VeRCoRs mock-up (1:3 scale nuclear containment building).

Lastly, in the framework of a reliability analysis, this thesis aims at estimating risks of exceeding regulatory leakage thresholds, while modeling the effect of eventual maintenance operations. In this framework, the impact of the choice of the inputs' probability law on some quantity of interest (including probabilities of failure) is assessed, through a robustness analysis. Next, a Bayesian approach aiming at updating both probabilities of failure and input parameters from monitoring data is presented. The overall methodology is applied to the realistic case of an operating 1:1 scale nuclear containment building.

Jury

• En savoir plus sur Thèse de Donatien Rossat

Jury

• En savoir plus sur Thèse d'Alberto Terzolo

Jury

• En savoir plus sur Thèse de Alejandro Ortega Laborin

• En savoir plus sur Thèse de Marielle Dargaud

Grenoble - Presqu'île

 

• En savoir plus sur Thèse de Maria Blasone

The drying of cement-based materials affects directly their durability, which has a major economic, societal, and environmental impact.
The conventional experimental techniques such as gravimetric and sensor-based measurements, which are employed to study the drying-driven processes, provide only bulk-averaged or point-wise measurements which are not sufficient to characterize these processes. However, the significant advances in full-field techniques have allowed unprecedented insight into these local processes. Notably, for cement-based materials, x-ray and neutron tomography lend themselves as highly complementary tools for the study of their THM behavior. In fact, the high sensitivity to density variations of x-ray imaging gives access to the developments of fractures, in 4D (3D+time). On the other hand, neutron tomography allows the study of the evolution of the moisture field in 4D, thanks to its high hydrogen sensitivity.
This Ph.D. takes advantage of these two highly complementary techniques, which, together with advanced numerical modelling tools, allowed for a novel experimental/numerical insight to study the drying-driven physical processes in cement-based materials. In this work, two experimental campaigns were performed at the NeXT instrument located at the ILL. In the first campaign, neutron tomography was employed to characterize the moisture distribution of a set of cylindrical mortar samples which were set to dry sequentially in a TH controlled environment (T=20 ºC, RH=35 %) to represent different hydric states. The main phases of the mortar (aggregates, cement paste, and voids were separated, and saturation profiles were deduced and validated against the weight loss measurements. In the second experimental campaign, more complexity was added to the experimental drying conditions by heating concrete and cement paste samples up to moderate temperature which has led to the appearance of cracks in the later sample. The analysis of the acquired simultaneous neutron/x-ray data-set (once aligned in time and across modalities) allowed for the quantification of the 4D moisture profiles which were found to predict an overall water loss at hydric equilibrium coherent with the corresponding analytical analysis. In the cement paste sample, the x-ray dataset captures the evolution of an extensive cracking network, opening, and propagation toward the core of the sample. Then, a novel analysis procedure was proposed which allowed the extraction of these fractures and the analysis of their interplay with local drying as captured through neutron imaging.
On the other hand, the numerical approach employed in this study consisted of improving the numerical model predictive capacity by assessing the implications of common simplifications on the modelling response. These common simplifications can be divided in three main categories which regard the consideration of gaseous transport modes, the TH and HM couplings, and the morphological description of the material. Quantification of these simplifications effects regarding the used model and the choice of TH coupling laws was done by comparing mass loss response surfaces in relative humidity and temperature space for multiple configurations. The results show relative error maps at early, mid, and late drying stages for every compared case. On the other hand, the simplification regarding the HM coupling was evaluated in a 2D mesoscopic simulation framework where an artificial concrete mesostructure had to be generated for cracking localization purposes. The cracking impact was then assessed both locally on the saturation fields and on the global mass loss response. Finally, a CT-FE mapping scheme was proposed which consisted of extracting the mesoscale morphology of concrete (aggregates and pores) from x-ray/neutron attenuation fields and presenting it explicitly on a FE mesh. This has permitted to perform 3D multiphase THM simulation of concrete at the mesoscale.

JURY

• En savoir plus sur Thèse de Hani Cheikh Sleiman

Saint-Martin-d'Hères - Domaine universitaire

Some ceramic grades, such as silicon carbide (SiC) or alumina (Al2O3), are used as ballistic materials thanks to their excellent mechanical performances, such as their hardness, while being light, where weight gain is a major issue for the design of military equipment for personal and vehicle protection. Since the Vietnam War, ceramics have been largely used and integrated as front face in bilayer shielding to stop the threat of AP (Armour Piercing)-type projectiles during a ballistic impact. Nevertheless, the projectile leads to an intense damage in the ceramic due to, amongst other phenomena, a dynamic tensile loading that manifests by multiple cracking, called fragmentation, particularly unfavourable for the integrity of the ballistic protection and its capacity to deal with a second impact. In order to develop a more performing shielding material, it is essential to understand the link between the microstructure of ceramics, the damage generated under impact and their ballistic performances.

This thesis seeks to better understand the dynamic fragmentation phenomenon generated at high strain rates in high fracture-toughness ceramics, including a bio-inspired alumina material mimicking nacre microstructure. This artificial nacre is, a priori, more crack resistant than conventional ceramics as it is characterised by a high static fracture-toughness due to its specific “Brick-and-Mortar” (or BM) microstructure reproduced in the material called here MAINa.

Jury

* Image caption

Some major results from the PhD: steel core extracted from API-BZ bullet with MAINa microstructure observed in Scanning Electron Microscopy (chapter 2); comparison between the experimental force-displacement response along with the numerical response using the identified experimental law of the steel core (chapter 3); numerical simulation of the penetration process of the steel core (800 m/s) in a SiC ceramic, 56 μs after impact (chapter 4); dynamic cracking of MAINa microstructure, tested in two orientations of platelets during Rockspall tests (chapter 5); multiple fragmentation of MAINa samples (0° orientation) after both Edge-On-Impact test in sarcophagus configuration and tandem test with tomographic segmentation (chapter 6).

• En savoir plus sur Thèse de Yannick Duplan

Monopile is the most common foundation system for offshore wind turbine structures, statistically about 73%, as per Wind Europe, 2020 report. A pile can be defined as flexible or rigid depending on the embedded length to diameter ratio (Le/D) and the relative stiffness of pile and soil. The existing codes for pile design are mainly developed for flexible piles, whereas monopile for new offshore wind turbines typically falls in the rigid category. The dominant complex cyclic wind and waves loads on offshore wind turbine structure and consequently on the monopile foundation, act in the lateral direction. The API design procedure, representing the lateral and vertical soil response through uncoupled non-linear springs, is developed for flexible piles and is recognised as conservative for rigid piles. The behaviour shall represent a coupled vertical and lateral soil-structure interaction because a rigid pile presents rotation deformation mode instead of deflection. The deformation mode further demands a different formulation mechanism, including the distributed moments along the pile shaft and shear & moment behaviours at the pile base as an essential part of the soil response investigations.
This work presents the numerical models aimed to address the limited understanding of coupling consideration of monopile installed in sand. The cone penetration tests performed in the calibration chamber, data treatment with ICP method and available Fontainebleau sand NE34 properties database in the literature provide the constituent parameters for model definition.
First, a PLAXIS 3D finite element model presents the model pile in the calibration chamber configuration with representative boundary conditions and the constitutive behaviour of the sand. The model pile geometry and load magnitudes are the outcomes of a similitude relationship with a representative prototype. A constant mass placed on the pile head represents the vertical load (the dead weight of the wind turbine structure). A simplified lateral point load represents the complex environmental loads, acting at a distance above foundation level, represents the lateral and moment load at mudline. Thus, vertical (V), lateral (H) and moment (M) collectively represent the combined load, investigated in both monotonic and cyclic loading cases. Different combined loading cases in the limits of horizontal and vertical load capacities represent the overall behaviour of the model pile. The observation of normal and shear stress changes close to the pile-soil interface at different depths quantify the pile-soil interaction. The response investigation at some strategic stress points in the FE model soil volume provides a basis for soil-stress transducers (SSTs) layout plan in the experimental soil volume. A methodology to formulate the lateral and shear stresses evolution close to the pile surface as representative of the coupled interaction is presented.
Second, a local-macro element (LME) model, an assembly of non-linear springs formulated using a Matlab toolbox ATL4S, presents the soil-pile interaction with inherent coupling considerations at different embedment depths. The PLAXIS model outcomes define the basis for a corresponding model scale LME model. The obtained results from both numerical investigations demonstrate the significance of vertical-lateral coupled interaction as a set of hypothesised equations for rigid monopile foundation. A similitude work provides a relation between a prototype scale and the lab-scale monopile model. It further aids in comparison and validation with the available experimental and numerical results in the literature.

Jury

 

 

• En savoir plus sur Thèse de Ritesh Gupta