Portail HAL Ensta
Not a member yet
11080 research outputs found
Sort by
Contribution à la modélisation et à la simulation numérique de l’usure par fretting
No full textWhen two contacting bodies are subjected to fretting motion, namely a cyclic tangential displacement with a very small displacement amplitude, they may experience different forms of degradation. Wear, which is defined as a progressive surface material removal, generally prevails under gross slip conditions. Wear of materials is a very complex phenomenon: it is the long-term, macroscopic result of various microscopic physical and chemical processes. This multi-physical and multi-scale nature makes its modeling difficult, such that there exists no widely adopted unified wear model. The most widespread wear models are empirical, and they sometimes fail to accurately predict wear evolution. As a result, studies on the subject are generally conducted through experimental approaches.The work presented here proposes to tackle the problem through the angle of modeling and numerical simulation. In a first part, an original modeling approach is proposed based on a thermodynamic framework. In this model, wear is accounted for by means of a damage-like variable whose role is to quantify the progressive accumulation of degradation ultimately leading to material detachment. This model uses the thick level set approach to govern the evolution of the surface geometry following wear. A numerical simulation strategy using the finite element method is defined to compute wear evolution using this model.In a second part, focus is on numerical aspects. The numerical simulation of wear problems is especially challenging, on the one hand because it involves several non-linearities (frictional contact, surface geometry evolution, material behavior), and on the other hand because it requires to simulate high amounts of time steps. In order to keep reasonable computational costs, an implicit cycle jump method is implemented. It proves to be more stable and efficient than the usually used explicit method. In addition, a simulation strategy is proposed to integrate the use of elastoplastic material behavior models within the wear simulation frameworks defined.Lorsque deux corps en contact sont soumis à une sollicitation de fretting, c'est à dire un mouvement cyclique de déplacement relatif de faible amplitude, de l'usure peut se produire. Il s'agit d'un phénomène complexe qui trouve son origine à l'échelle microscopique dans les interactions entre aspérités de surface et se traduit sur le long terme, à l'échelle macroscopique, par une perte progressive de matière en surface. La diversité des mécanismes microscopiques qui sous-tendent l'usure, en plus de sa nature multi-physique et multi-échelles, en temps et en espace, rend le phénomène très complexe à modéliser. Les modèles les plus répandus sont empiriques et, s'ils parviennent dans un certain nombre de cas à corréler les résultats expérimentaux, ils ne permettent pas de prédire efficacement l'évolution de l'usure de façon générale. Pour ces raisons, l'étude des phénomènes d'usure se fait généralement par des approches expérimentales.Le travail présenté s'attache à traiter le problème de la modélisation et de la simulation numérique de l'usure sous sollicitations de fretting. Le premier volet de ce travail consiste en une approche de modélisation originale, dans le cadre de la thermodynamique. L'accumulation progressive de dégradations conduisant à l'usure y est quantifiée au moyen d'une variable d'endommagement, et le modèle repose sur l'utilisation de l'approche thick level set. Une procédure de calcul utilisant la méthode des éléments finis est définie pour simuler l'usure en utilisant ce modèle.Dans une seconde partie, l'accent est mis sur la simulation numérique. La simulation de l'usure présente un certain nombre de difficultés, d'une part du fait des non-linéarités inhérentes au problème (contact, frottement, évolution de la géométrie des surfaces, non-linéarité du comportement des matériaux), et d'autre part en raison de la nécessité de simuler de très grands nombres de pas de temps. Afin de conserver un temps de calcul raisonnable, une méthode de saut de cycle implicite est proposée. À l'inverse de la méthode explicite classiquement employée, elle permet d'éliminer les instabilités qui peuvent apparaître lorsque le saut de cycle est trop grand. Par ailleurs, une procédure de simulation permettant d'intégrer des lois de comportement élastoplastique dans les calculs d'usure est définie
Diffraction électromagnétique par une couche mince de nanoparticules réparties aléatoirement : développement asymptotique, conditions effectives et simulations.
No full textWe consider the scattering problem, in harmonic regime, of an electromagnetic plane wave by an inhomogeneous object covered with a very thin layer of small randomly distributed perfectly conductive particles. We seek to quantify the effect of this layer on the reflection coefficient. The size of the particles, their spacing and the thickness of the layer are of the same order but small compared to the incident wavelength and the dimensions of the object. Two difficulties appear: (1) Solving numerically Maxwell’s equations in this context is extremely expensive in terms of memory size and computation time; (2) the particle distribution is not known for a given object. We will assume that it is a realization ofa supposedly random distribution.To overcome these difficulties, we propose a deterministic effective model, using a multiscale asymptotic expansion of the solution, where the particle layer is replaced by an effective boundary condition, prescribed on a surface above the particles. The coefficients involved in the condition require the solution of cell problems posed in a half-space covered by a layer ofrandomly distributed unit-sized particles.Nous considérons le problème de diffraction, en régime harmonique, d’une onde plane électromagnétique par un objet inhomogène recouvert d’une couche très fine de petites particules parfaitement conductrices distribuées aléatoirement. Nous cherchons à quantifier l’effet de cette couche sur le coefficient de réflexion. La taille des particules, leur espacement et l’épaisseur de la couche sont du même ordre mais petites par rapport à la longueur d’onde incidente et les dimensions de l’objet. Deux difficultés apparaissent : (1) Résoudre numériquement les équations de Maxwell dans ce contexte est extrêmement coûteux en terme de taille mémoire et de temps calcul; (2) la répartition des particules n'est pas connue pour un objet donné. Nous allons supposer que c'est une réalisation d'une répartition supposée aléatoire.Pour contourner ces difficultés, nous proposons alors un modèle effectif, à l’aide d’un développement asymptotique multi-échelle de la solution, où la couche de particules est remplacée par une condition aux bords effective, prescrite sur une surface située au-dessus des particules. Les coefficients qui interviennent dans la condition nécessite la résolution de problèmes, dits de cellule, posés un demi-espace recouvert d'une couche de particules, de taille unitaire, réparties aléatoirement
Exploring the Thermophysical Properties of the Thermal Conductivity of Pigmented Polymer Matrix Composites with Barium Titanate: A Comparative Numerical and Experimental Study
No full textInternational audienceThis research paper focuses on investigating the thermal conductivity behavior of polymer matrix composite materials, specifically those composed of PSU and BaTiO3, both experimentally and numerically. The thermal conductivity of composites has been studied using a variety of theoretical and semi-empirical methods. However, in cases where the filler concentration is minimal, these models provide a superior estimate. To numerically resolve the thermal heat transfer for an elementary cell, the finite element method is employed in this study. The impact of contact resistance, barium titanate percentage, and quenching temperature on the composite’s effective thermal conductivity and dynamic behavior is given consideration. The results demonstrate that the suggested numerical model is in good agreement with experimental measurements as well as Hatta–Taya and Hashin–Shtrikman’s analytical models. The results provide significant insight into the thermal conductivity behavior of composites, which can inform the development of more effective thermal management solutions for composite materials. Effective thermal management is critical for the successful application of polymer matrix composite materials in various engineering applications. Thermal conductivity is a key factor in thermal management and is influenced by factors such as the concentration of filler particles, their shape, size, and distribution, and the matrix material’s properties
COLA: COarse LAbel pre-training for 3D semantic segmentation of sparse LiDAR datasets
No full textInternational audienc
L’autonomie stratégique des nations cause-t-elle le développement économique ?Analyse à partir d’un panel de 34 pays (1993-2016)
No full textInternational audienc
Large Language Models as Superpositions of Cultural Perspectives
No full textPreprintLarge Language Models (LLMs) are often misleadingly recognized as having a personality or a set of values. We argue that an LLM can be seen as a superposition of perspectives with different values and personality traits. LLMs exhibit context-dependent values and personality traits that change based on the induced perspective (as opposed to humans, who tend to have more coherent values and personality traits across contexts). We introduce the concept of perspective controllability, which refers to a model's affordance to adopt various perspectives with differing values and personality traits. In our experiments, we use questionnaires from psychology (PVQ, VSM, IPIP) to study how exhibited values and personality traits change based on different perspectives. Through qualitative experiments, we show that LLMs express different values when those are (implicitly or explicitly) implied in the prompt, and that LLMs express different values even when those are not obviously implied (demonstrating their context-dependent nature). We then conduct quantitative experiments to study the controllability of different models (GPT-4, GPT-3.5, OpenAssistant, StableVicuna, StableLM), the effectiveness of various methods for inducing perspectives, and the smoothness of the models' drivability. We conclude by examining the broader implications of our work and outline a variety of associated scientific questions. The project website is available at https://sites.google.com/view/llm-superpositions
Offline and Online Use of Interval and Set-Based Approaches for Control and State Estimation: A Review of Methodological Approaches and Their Application
No full textInternational audienceControl and state estimation procedures need to be robust against imprecisely known parameters, uncertainty in initial conditions, and external disturbances. Interval methods and other set-based techniques form the basis for the implementation of powerful approaches that can be used to identify parameters of dynamic system models in the presence of the aforementioned types of uncertainty. Moreover, they are applicable to a verified feasibility and stability analysis of controllers and state estimators. In addition to these offline approaches for analysis, interval and set-based methods have also been developed in recent years, which allow to solve the associated design tasks and to implement reliable techniques that are applicable online. The latter approaches include set-based model-predictive control, online parameter adaptation techniques for nonlinear variable-structure and backstepping controllers, interval observers, and fault diagnosis techniques. This paper provides an overview of the methodological background and reviews numerous practical applications for which interval and other set-valued approaches have been employed successfully
Fusion of deep learning architectures for enhanced target recognition on SAR images
No full textInternational audienceIn various applications of radar imagery, one of the fundamental problems is mainly linked to the analysis and interpretation of the images provided, in particular the recognition of moving and/or fixed targets. This task has become more difficult due to the large volume of radar data. This led to the use of automatic processing and target recognition methods. The aim of this study is to explore data fusion in SAR (Synthetic Aperture Radar) image classifiers. To this end, we propose a new approach to combine three CNN (Convolutional Neural Networks) architectures with several fusion rules. First, we perform a training process of three deep learning architectures; namely, the basic CNN, the Xception, and the AlexNet architectures. Then, two fusion techniques are proposed. The first one deals with the majority rule and the second uses a neural networks to combine the decision outputs obtained from three elementary classifiers to achieve the final decision. To evaluate and validate the proposed approach, the MSTAR (Moving and Stationary Target Acquisition and Recognition) dataset is used. The obtained performances of the fusion techniques improve the recognition rate with a final accuracy of 99.59% for the majority rule and 99.51 for the neural network-based rule, which surpasses the accuracy of each individual CNN
Planetizing History
No full textInternational audienceThe expansion of the human sphere beyond Earth has larger repercussions for understanding the present than is usually acknowledged. The outcome of a collaboration between a historian and a philosopher, this paper takes up a term originally coined by Teilhard de Chardin in 1946 and proposes ‘planetization’ as a key analytical concept. Planetizing history amounts to situating the space of history within an extra-global horizon, emphasizing the significance of technological infrastructures positioned in outer space including robotic spacecraft and orbital infrastructures for the environmental, social, and political histories of what has been classified as globalization. To planetize history, then, is to show that the history of the globalized present cannot be written from an exclusively terrestrial point of view