21549 research outputs found
Sort by
Unexpected enhanced reactivity of aluminized nanothermites by accelerated aging
Get PDFWe examine the thermal aging of four Al based thermites chosen among the most commonly used in microenergetic systems: Al/CuO, Al/Fe2O3, Al/Fe3O4 and Al/Co3O4. For each nanothermite system, we applied the modified Friedmann isoconversional method from DSC signals to calculate the kinetic parameters of reactions and the function of the reaction progress in air and argon. As result, it is found that all thermites should be thermally stable for one century when stored at temperatures below 200 °C, except for Al/Fe3O4 which readily converts into Al/Fe2O3 before the thermite reaction onset. Then, we designed annealing experiments to a fixed reaction progress : (i) as simulated after 100 years storage at the ambient temperature, (ii) up to 5%. The pressure development and burning rate of aged thermites are compared with as-prepared ones. After annealing at 200 °C in air and 400 °C in argon, an unforeseen faster reaction and pressurization rates are observed for Al/CuO and Al/Fe2O3 thermites. STEM and EDX show that the annealing provokes a modification of the aluminum particles structure accompanied with a softening of the alumina which explains the enhanced pressure performances and burn rates. Whereas the same nanothermites, Al/CuO and Al/Fe2O3, annealed at 400 °C in air, feature a reduced reactivity due to the over-oxidations of the aluminum core and the formation of crystalline alumina shell. These results reveal the high complexity of aging processes in powdered nanothermites as it provokes, not only a consumption of the heat reservoir, but also a modification of the Al/oxidizer interface morphology and chemistry which has a greater impact on the material reactivity than cumulative heat release and reaction progress evolution
Evaluating BDP FRAME extension for QUIC
Get PDFThe first version of QUIC has recently been standardized by the IETF. The framework of QUIC enables the proposition, negociation and exploitation of extensions to adapt
some of its mechanisms. As one example, the DATAGRAM
extension enables the unreliable transmission of data.
The BDP FRAME extension is a method that can improve
traffic delivery by allowing a QUIC connection to remember
the knowledge of path characteristics and exploit them when
resuming a session.
This technical report presents the rationale behind fast convergence in SATCOM systems and evaluate the BDP FRAME
extension in emulated and live environments
Effects of welding parameters on the microstructure and mechanical properties of the AA6061 aluminium alloy joined by a Yb: YAG laser beam
Get PDFIn this study, the effects of Yb: YAG laser welding parameters on the microstructure and mechanical properties of AA6061-T4 joints were analysed. Samples without any welding defects such as porosity, melt pool collapse, and hot cracking were produced with different welding parameters. Energetic processing parameters had a significant influence on the fusion zone (FZ) and heat-affected zone (HAZ) microstructure and dimensions, in addition to local and global mechanical properties. The weld bead width increased with increasing power and energy densities and reduced the weld bead tensile properties. High welding travel speed produced a more elongated weld pool and ripples resulted in a ‘V’ shape. A significant quantity of axial heterogeneous nucleation was observed in the FZ centre of these weld beads because of the low-energy density. In contrast, low welding travel speed produced C-shaped ripples and a few axial grain nucleation sites in the FZ. The latter was evidence of a slower solidification rate. Dendrite secondary arm spacing measurements confirmed this hypothesis. It was observed that axial grains in the FZ centre improve the weld bead tensile properties. Compared to the base material (BM), the hardness in the FZ was reduced and identical in the HAZ. The FZ hardness depended on the welding parameters. Digital image correlation (DIC) strain measurements indicated higher deformation near the FZ, but when geometrical defects were removed, the FZ deformed more homogeneously
High temperature translaminar fracture of woven-ply thermoplastic laminates in tension and in compression
Get PDFPEEK reinforced by a woven carbon fibers obtained by consolidation process, has been studied by
means of Compact Tension (CT) and Compact Compression (CC) tests. An orthotropic laminate
and two quasi-isotropic laminates, have been tested at room temperature (RT) and at 150 ◦C (i.e.
for a temperature slightly higher than the glass transition temperature Tg), with the purpose of
evaluation the influence of the stacking sequence and the temperature on the damage mechanisms.
The CT type tests show that the temperature influence on the overall mechanical response
remains low. In contrast, the CC type tests show a decrease in the mechanical resistance and an
augmentation in the global ductile behavior. By means of the 2D Digital Image Correlation and
the implementation of a algorithm based on the sigma indicator (confidence interval in the area of
interest), the crack propagation was measured during loading, then the G-R curves have been
obtained from the compliance method. In tension, there is an increase in the translaminar fracture
toughness with the laminate thickness and a slight improvement in this one due to the influence
of temperature. In compression, the temperature strongly decreases the translaminar fracture
toughness
Suivi Argos d'Eidolon Helvum en Afrique de l'Ouest pour comprendre le comportement de sélection des sites dortoirs à l'interface chauves-souris et hommes
Get PDFLes chauves-souris sont impliquées dans de nombreuses maladies zoonotiques. Plus précisément, les chauves-souris Eidolon helvum sont suspectées d’être un des réservoirs du virus de la maladie Ebola, qui a causé 30 épidémies depuis son identification en 1976. L’épidémiologie de la maladie et l’écologie des espèces impliquées restent cependant mal connues. Eidolon helvum parcourt plusieurs milliers de kilomètres sur l’ensemble de l’Afrique de l’Ouest au cours de mouvements migratoires annuels, mais peu de données sont disponibles sur la mobilité et la distribution de l’espèce. Ce travail porte sur les mouvements migratoires et le comportement de sélection d’habitat des chauves-souris Eidolon helvum, en cherchant à comprendre quels facteurs environnementaux influencent la sélection de leurs sites dortoirs. La télémétrie Argos et l’utilisation d’indices d’eau et de végétation issus de la télédétection satellitaire ont permis de réaliser un suivi spatio-temporel de trois individus, sur une durée d’un an. 29 sites dortoirs ont pu être identifiés. L’interprétation des résultats du modèle de distribution de l’espèce a montré que les chauves-souris sélectionnent l’habitat de type construction. Aucune influence des indices issus de la télédétection n’a pu être démontrée. L’attrait pour ces zones urbaines est donc à intégrer à la gestion du risque de transmission de pathogènes à l’homme et à la gestion du risque de propagation de pathogènes à l’échelle de l’Afrique occidentale
Tutorial on Radiation Effects on CMOS Image Sensors
No full textNowadays, CMOS Image Sensors (CIS), also called Active Pixel Sensors (APS), represent the most popular solid-state imager technology as illustrated by its ubiquity in mass consumer smartphones and cameras. In particular, CISs are used in various imaging applications in harsh radiation environments (e.g. space remote sensing, nuclear and scientific instrumentation and medical imaging). During this lecture, the particularities of the CIS technology will be briefly presented. In a second part, an overview of the most important radiation effects on these imagers will be discussed. In particular, the physical mechanisms at the origin of the most important Total Ionizing Dose (TID) and Displacement Damage (DD) effects in CISs will be detailed including: charge transfer degradation, dark current increase, hot pixel creation, and random telegraph signals (RTS)
FactoRing: Asynchronous TSN-compliant Network with low bounded Jitters for Industry 4.0
Get PDFTime-Sensitive Networking (TSN) describes a set of features extending the functionalities and Quality of Service
(QoS) of standard Ethernet to enable determinism, reliability and reconfigurability, key requirements for Industry 4.0. The TSN profile for industrial automation (IEC/IEEE 60802 standard) defines specific options to favor reliability and reconfigurability.
Nevertheless, there are multiple possible options for scheduling to guarantee determinism. Today, the scheduling standards in TSN can be categorized according to the implemented communication paradigm: asynchronous or synchronous. This paradigm is of utmost importance to quantify the synchronization need and the reconfigurability effort. The main contribution of this work is the specification of an asynchronous TSN-compliant network for Industry 4.0, FactoRing, that bridges the gap between both paradigms to guarantee low bounded jitters and latencies, without the need of synchronization and complex network planning. Moreover, FactoRing supports ring-based topologies to significantly reduce installation wiring and costs. In this paper, we first present the main industry 4.0 requirements and assess the ability of recommended TSN mechanisms for industrial automation versus such requirements. Afterwards, we detail the main features of Factoring including QoS, reliability and reconfiguration management. Finally, preliminary results on
performance metrics like jitters, latencies and buffer usage are discussed and the first conclusions on the promises of Factoring to meet Industry 4.0 requirements are derived
Prévalence et facteurs de risque d'obésité et de surpoids dans une population de chats sains présentés en consultation de médecine préventive à l’École Nationale Vétérinaire de Toulouse et à l’École Nationale Vétérinaire de Maisons-Alfort
Get PDFUne enquête visant à évaluer la composition corporelle de chats adultes, la perception de celle-ci par leur propriétaire, ainsi que des facteurs de risque d’obésité, a été conduite sur 318 chats venus en consultation de médecine préventive à l’ENVT et à l’ENVA, en 2020-2021. Les résultats montrent que 15% de la population étudiée est en surpoids modéré à obèse. Étonnamment, moins de la moitié des chats de cette étude ont présenté un statut musculaire satisfaisant, avec seulement 10% des chats ayant une musculature optimale. Cette observation soulève de nombreuses interrogations. Par rapport à des études antérieures, la prévalence d’obésité semble avoir diminué, malgré une augmentation de l’incidence du surpoids. Notre étude montre que le facteur de risque principal d’obésité est une faible activité physique, avec une influence de l’âge de l’animal. Les résultats sont globalement en accord avec ceux déjà publiés
Development of a Formal Verification Methodology for B Specifications using PERF formal toolkit. Application to safety requirements of railway systems.
Get PDFThe design of complex systems involves several design models supporting different analysis techniques for validation and verification purposes. These activities lead to the definition of heterogeneous modelling languages and analysis techniques. In this setting, meeting certification standards becomes a key issue in system engineering. Reducing heterogeneity due to the presence of different modelling languages can be addressed by providing an integrated framework in which involved modelling languages and techniques are formalised. In such a framework, checking global requirements fulfilment on heterogeneous models of a complex critical system becomes possible in many cases. The work presented in this thesis addresses the problem of integrated verification of system design models in the context of transportation systems, in particular railway systems. It has been achieved in context of the B-PERFect project of RATP (Parisian Public Transport Operator and Maintainer) aiming at applying formal verification using the PERF approach on the integrated safety-critical models of embedded software related to railway domain expressed in a single unifying modelling language: High Level Languge (HLL). We also discuss integrated verification at the system level. The proposed method for verification of safety-critical software is a bottom-up approach, starting from the source code to the high-level specification. This work addresses the particular case of the B method. It presents a certified translation of B formal models to HLL models. The proposed approach uses Isabelle/HOL as a unified logical framework to describe the formal semantics and to formalise the transformation relation between both modelling languages. The developed Isabelle/HOL models are proved in order to guarantee the correctness of our translation process. Moreover, we have also used weakbisimulation relation to check semantic preservation after transformations. In this thesis, we also present the implementation of the defined transformation syntactic rules as the B2HLL tool. Moreover, we show the model animation process we set up to validate the B2HLL translator tool with respect to the formalised transformation rules we defined in Isabelle/HOL. This approach helps us to validate definitions, lemmas and theorems of our formalised specifications. We have used the B2HLL tool to translate multiple B models, and we also show that when models are translated into this unified modelling language, HLL, it becomes possible to handle verification of properties expressed across different models
Inertial and Second-order Optimization Algorithms for Training Neural Networks
Get PDFNeural network models became highly popular during the last decade due to their efficiency in various applications. These are very large parametric models whose parameters must be set for each specific task. This crucial process of choosing the parameters, known as training, is done using large datasets. Due to the large amount of data and the size of the neural networks, the training phase is very expensive in terms of computational time and resources. From a mathematical point of view, training a neural network means solving a large-scale optimization problem. More specifically it involves the minimization of a sum of functions. The large-scale nature of the optimization problem highly restrains the types of algorithms available to minimize this sum of functions. In this context, standard algorithms almost exclusively rely on inexact gradients through the backpropagation method and mini-batch sub-sampling. As a result, firstorder methods such as stochastic gradient descent (SGD) remain the most used ones to train neural networks. Additionally, the function to minimize is non-convex and possibly nondifferentiable, resulting in limited convergence guarantees for these methods. In this thesis, we focus on building new algorithms exploiting second-order information only by means of noisy firstorder automatic differentiation. Starting from a dynamical system (an ordinary differential equation), we build INNA, an inertial and Newtonian algorithm. By analyzing together the dynamical system and INNA, we prove the convergence of the algorithm to the critical points of the function to minimize. Then, we show that the limit is actually a local minimum with overwhelming probability. Finally, we introduce Step-Tuned SGD that automatically adjusts the step-sizes of SGD. It does so by cleverly modifying the mini-batch sub-sampling allowing for an efficient discretization of second-order information. We prove the almost sure convergence of Step-Tuned SGD to critical points and provide rates of convergence. All the theoretical results are backed by promising numerical experiments on deep learning problems