Portail HAL Ensta
Not a member yet
11080 research outputs found
Sort by
WAXD investigations on the effect of loading history on strain-induced crystallization for a fully formulated filled natural rubber
No full textInternational audienceModelling crystallization under stretch is a key topic for fatigue design of rubber-like antivibration parts. Nevertheless, most of the academic studies consider unfilled natural rubber while the industrial materials are fully formulated compounds filled with carbon blacks and exhibit a highly dissipative visco-elastic behavior. This behavior is very useful for antivibration systems but complicates the characterization and modelling of the phase change, as the addition of fillers and additives brings in numerous additional dissipation sources and intricates the time effects on the thermomechanical response and on crystallization. In this study, we use well resolved WAXD synchrotron measurements to perform in situ measurements under various mechanical protocols. The objective is to characterize the evolution of the triplet {strain, stress, crystallinity index}, and their derivatives, for various time and mechanical solicitations. First, classic load/unload tension tests over a range of strain rates leading to non-equilibrium cases are achieved, to serve as a reference database on the compound studied. Then, a multi-relaxation cyclic test combining static and monotonic steps is applied in order to describe the crystallization state and kinetics around a relaxed state (sometimes called "equilibrium hysteresis"). The results provide a precious database to identify or challenge the existing thermodynamic models, for conditions seldom met in the literature: fully formulated material and various mechanical loading time histories
The Future of Whale Conservation: Harnessing Distributed Acoustic Sensing for Real-Time Monitoring
No full textInternational audienceWhale distribution is a key indicator of ocean health and ecosystem stability, highlighting the need for informed conservation efforts. Yet, current monitoring methods such as visual surveys or autonomous passive acoustic monitoring, lack the spatial and temporal resolution necessary for detailed and continuous surveys at scale.Distributed Acoustic Sensing (DAS) is a game-changing innovative technology that transforms optical fibers into a dense network of sensors for geophysical and acoustic measurements. By emitting laser pulses propagating along the fiber, DAS detects signals reflected back from various points along the cable, providing high-resolution spatial data, over tens of kms. At the cost of massive amounts of data that require fast, optimized algorithms, this technology can operate on any optical fiber, including the hundreds of telecom cables already crisscrossing the world’s oceans.In this presentation, we will illustrate how DAS implemented on seafloor cables along the French Mediterranean coast enables whale detection and localization. This innovative project aims to track the presence, migration patterns, and locations of whales, contributing valuable insights into their behavior and movement patterns. Moreover, with the ability to collect and process data in real-time opens up the possibility of sharing the cetacean locations with relevant stakeholders, including nearby vessels.This information can enable vessel operators to adjust their speed, thereby reducing the risk of collisions and protecting marine mammals from the impacts of human activities
Performances des protocoles QKD, BB84 et SARG04 pour les communications quantiques sous-marines
No full textInternational audienceIn the context of secure underwater communications, particularly the transmission of encryption keys via a quantumprocess (Underwater Quantum Key Distribution - UQKD), this study aims to highlight the influence of the propagation channelon transmission efficiency. We compare two discrete-variable QKD protocols: BB84 and SARG04, as a function of propagationdistance, seawater characteristics (clear, coastal and turbid) and atmospheric conditions. The main comparison criterion is aquantitative analysis of the Quantum Bit Error Rate (QBER). To this end, we rely on an existing analytical expression and provide anumerical simulation as a function of the sea model parameters.Dans le contexte des communications sous-marines sécurisées et plus précisément des transmissions de clés de cryptage par un processus quantique (Underwater Quantum Key Distribution -UQKD), cette étude cherche à mettre en évidence l'influence du canal de propagation sur l'efficacité de la transmission. Nous comparons ainsi deux protocoles de QKD à variables discrètes : BB84 et SARG04, en fonction de la distance de propagation, des caractéristiques d'eau de mer (claire, côtière et trouble) et des conditions atmosphériques. Le critère principal de comparaison consiste en une analyse quantitative du taux d'erreur de bit quantique (Quantum Bit Error Rate -QBER). Nous nous basons pour cela sur une expression analytique déjà établie et connue de la littérature et proposons une simulation numérique en fonction des paramètres du modèle de mer
A Wavelet Frames Based Split-Step Method to Model the Tropospheric Long-Range Propagation
No full textInternational audienceRapid and accurate methods for modeling longrange propagation in the troposphere are of major interest for various applications, including communication, navigation, surveillance, and observation. In this case, the parabolic wave equation model is widely used and commonly solved using split-step methods. In particular, the recent split-step wavelet method has shown promising results in terms of time and memory efficiency. Nonetheless, using a wavelet basis renders the propagation step tricky since the translation invariance property is lost. In this paper, we propose to use wavelet frames in place of the wavelet basis to recover the translation invariance property. This leads to a direct and efficient implementation of the method in the wavelet domain, which can be easily passed on to a GPU. Furthermore, we show that it comes at almost no cost in time or memory efficiency. Numerical experiments are performed to validate the method and highlight its advantages
Equivalent Beam Finite Element Model to Assess the Mechanical Behavior of a Composite Hydrofoil
No full textInternational audienceThis paper presents and validates a method to model the static structural behavior of composite hydrofoils with 1D beam finite elements. Classically, 3D solid and 2D shell finite elements are employed to predict the anisotropic and heterogeneous behaviors of composite hydrofoils. Modeling a hydrofoil with 1D beam elements drastically reduces its number of structural parameters and the computational time required for the finite element analysis. The present study investigates the static structural behavior of four hydrofoils, constructed in the same mold with different mechanical properties, leading to a specific bend-twist coupling for each foil. The static deformations of the foils are evaluated experimentally and numerically for different load cases. Two numerical models are considered, one with 3D solid and 2D shell elements and a second with 1D beam elements. Then, the results are compared to assess the validity of the 1D finite element model. The comparisons shows that both numerical models are in good agreement with the experimental results. The results also confirm that the 1D finite element model is able to correctly describe the impact of the fiber orientations on the structural responses of the hydrofoils. The computational time savings allowed by the 1D model are also quantified
Modélisation et simulation de l'effet Micro-Doppler en théorie du radar : application aux aéronefs
No full textInternational audienceLa détection radar de petits aéronefs tels que des drones représente un enjeu de sécurité majeur à cause de leur faible écho radar. Ce travail de recherche constitue une étude préliminaire de la signature Micro-Doppler des drones se concentrant sur les quadricoptères pour mettre en exergue les paramètres critiques pour la détection radar de drones. Nous utilisons un modèle d'optique physique utilisant des réflecteurs ponctuels pour générer le signal qui est interprété avec une analyse en temps-fréquence
Digital Twin and Control for Emulating Autonomous Sailboats with Conventional USVs: The Digital Sibling Concept
No full textInternational audienceAs environmental concerns and advancements in robotics continue to shape the future of marine technology, Autonomous Sailboats emerge as a promising solution for both scientific and commercial applications. Yet, testing of sensors, control algorithms, or obstacle avoidance solutions remains complex because of the fragility, cost, and difficulty of access to these systems. This work introduces the concept of Digital Sibling, which facilitates early Hardware-In-the-Loop testing. This technique enables the development of an experimental scheme where the complex system is simulated and its key behaviors are emulated by a simpler physical system via tracking. This novel concept, stemming from conventional Digital Twin techniques, is introduced and experimentally validated in this work. Notably, this work presents experiments where an Uncrewed Surface Vessel is used to emulate the behavior of an Autonomous Sailboat in line following, tacking and obstacle avoidance tasks
Analyse résolvante de champs moyens turbulents : une application à l’aéroacoustique
No full textInternational audienc
FABME: File-level placement policy with control admission, burst prefetching, and multi-criteria eviction for HPC multi-tier storage
No full textInternational audienc
Recovering Dense Metric Depth in Indoor Scenes from Monocular Depth Foundation Models and 2D LiDARs
No full textInternational audienceRecently, the first foundation models for monocular depth estimation such as Depth Anything have emerged. However, by being trained to make affineinvariant predictions, these methods rely on fine-tuning for making metric depth predictions and therefore perform poorly on zero-shot metric depth estimation. In a real use case, the fine-tuning stage is costly because a dedicated dataset with ground truth depth must be created and used as a training set. Additionally, finetuning can compromise the model's generalization ability. This paper proposes to leverage 2D LiDARs to rescale Depth Anything's predictions in the context of indoor scenes so as to prevent expensive fine-tuning or harming the model capacity. Our experiments demonstrate similar performance with fine-tuned approaches and enhanced results over zero-shot metric depth estimation methods.</div