The deposition of the ternary compound Ti-N-Si using the cold-wall CVD technique from TiCl4-NH3-SiH2Cl2 presents difficulties because of undesired powder formation in the gas-phase. For a better understanding and control of the mechanism, both simulation and experiments have been carried out to determine favorable conditions to minimize the undesired adducts. First, an original technique has been developed for the reactive injection with a double-passage nozzle specially designed. Then powders have been deliberately grown at different conditions, with their chemical compositions analyzed using X-ray diffraction spectra. Further, the flow has been modeled and compared with the results provided by means of laser sheet visualization. It is shown that the powders can be avoided with some special reactive injection schemes and by controlling the nozzle exit-substrate distance. Some useful information is also delivered on controlling the color and the composition of powders during their synthesis.

The present study deals with the investigation of electrically active damage induced by direct and through protecting oxide layer implantation of 11B+ ions. The residual defects have been determined by means of Deep Level Transient Spectroscopy (DLTS). It has been found that the number of defects is practically reduced to one centre when the implantation is performed through an oxide layer. The defect spectrum evolution, under the effect of the implant mass and the RTA treatments, has been also investigated. The defect generation kinetics, under annealing treatments, is found strongly depending on 11B+ ionic number reaching the substrate.

A linear relationship is found between peak position and peak width for the X-ray diffraction peaks of MgO very small crystallites. The sign of the peak shift is explained. The peak shape is asymmetrical, the more asymmetrical the broader it is. The second Heisenberg's uncertainty relation is reinterpreted.

This paper concerns momentum and heat transfer to a nonevaporating oxidized steel spherical particle (diameter 6.5 mm − 8 mm) levitated at high Reynolds number in argon turbulent plasma jet flowing into cold air. The rapid rotation of the particle in the plasma flow ensures the stability of the levitation phenomenon over a wide range of plasma torch parameters. We use a dimensional analysis to describe the momentum transfer and the equilibrium levitation height data. The convective heat transfer coefficient was further estimated from gas and particle equilibrium temperature at the front stagnation point or from the analysis of the transient temperature signal for a particle abruptly immersed and levitated in the plasma jet. Particle rotation is shown to determine the heat transfer rate in a nonuniform plasma flow.

Heat and mass transfers occuring in a counterflow direct contact liquid-gas exchanger determine the performance of a new greenhouse air dehumidifier designed at INRA. This prototype uses triethylene glycol (TEG) as the desiccant fluid which extracts water vapor from the air. The regeneration of the TEG desiccant fluid is then performed by direct contact with combustion gas from a high efficiency boiler equipped with a condensor. The heat and mass transfers between the thin film of diluted TEG and the hot gas were simulated by a model which uses correlation formula from the literature specifically relevant to the present cross-corrugated plates geometry. A simple set of analytical solutions is first derived, which explains why some possible processes can clearly be far from optimal. Then, more exact numerical calculations confirm that some undesirable water recondensations on the upper part of the exchanger were limiting the performance of this prototype. More suitable conditions were defined for the process, which lead to a new design of the apparatus. In this second prototype, a gas-gas exchanger provides dryer and cooler gas to the basis of the regenerators, while a warmer TEG is fed on the top. A whole range of operating conditions was experimented and measured parameters were compared with numerical simulations of this new configuration: recondensation did not occur any more. As a consequence, this second prototype was able to concentrate the desiccant fluid at the desired rate of 20 kg H2O/hour, under temperature and humidity conditions which correspond to the dehumidification of a 1000 m2 greenhouse heated at night during the winter season.

Le grand intérêt suscité par les matériaux à structure fractale ces dernières décennies nous a conduits à étudier tout particulièrement le comportement électrique d'échantillons métalliques dont les motifs sont du type “Arbre fractal”. Les caractéristiques singulières de ces structures telles l'autosimilarité et la dimension fractionnaire Df confèrent des propriétés physiques particulières à ces matériaux. L'analyse de ces structures est envisagée en réalisant un condensateur de type MIS (Métal-Isolant-Semi-conducteur) dont une des armatures présente une dimension fractionnaire, l'autre est plane et l'isolant est un oxyde de silicium (SiO2). Les simulations et les mesures effectuées ont permis de mettre en évidence sur l'impédance d'une telle structure, une évolution fréquentielle étroitement corrélée à la structure même de l'échantillon. Ce comportement que l'on qualifiera par la suite de “fractal”, peut se résumer ainsi : soumis à une excitation électrique à fréquence variable, cet échantillon se caractérise par une impédance complexe qui, à partir d'une fréquence caractéristique, présente une phase constante c'est-à-dire que ses parties réelle et imaginaire ont la même loi de dépendance fréquentielle. L'étude que nous présentons concerne d'une part, la mise en évidence de ce comportement et d'autre part, les conditions permettant d'ajuster la plage de fréquence sur laquelle ce comportement apparaît.

This paper presents a signal based approach to detect PWM inverter faults and motor unbalances in electrical drives. A diagnostic system which uses the control variables of the drive and the current measurements is presented. It is based on frequency analysis and on the energetic characterization of these variables in the PWM frequency domain. These method very simple to implement in real time have been successfully applied to an experimental system.

This paper deals with the study of shunt active filters used for suppressing particular harmonics generated by nonlinear loads in utility distribution power systems. Both structures of shunt active filter, voltage source active filter (VSAF) and current source active filter (CSAF), are considered. The analytical study of specific harmonics identification in a given spectrum is first presented. For simulation as well as experimentation the nonlinear load is a conventional three phase thyristor rectifier and harmonics 5 and 7 are selected to be eliminated by active filter. The whole system consisting of the ac power supply network, the SCR rectifier and the shunt active filter (VSAF/CSAF) is then simulated. The simulation results are discussed and the efficiency of the two kinds of active filter are compared. Finally, for the first structure, VSAF, the simulation results are confirmed by experimental test realized by means of a fully digital control active power filter developed in our laboratory.

A new structural model of shear-thickening “dilatancy” is proposed for (strongly) stabilized disperse systems. This model is based on the effective volume fraction (EVF) concept, developed in Part I of this series and previously used for the rheological modeling of complex fluids. In such a description, the latter are considered as concentrated dispersions of basic structural units (SUs) — either small, compact clusters or primary particles —, forming large structures at low shear. As shear rate increases, rupturing of these large structures leads to the shear thinning observed prior to dilatancy. The novelty of this model lies in assuming that, beyond the onset of dilatancy, hydrodynamic forces promote, at the expense of basic-SUs, the formation of hydrodynamic clusters as both rheo-optical experiments and numerical simulations recently demonstrated, in contradiction with the (classical) theory based on a shear induced disruption of particle layering. Dilatancy directly results from the increase of the EVF of the dispersion, closely related to the increasing volume of continuous phase imprisoned inside hydroclusters whose size grows as the shear rate increases. Predictions of the model are discussed in comparison with the Mayn features observed in a large number of dilatant dispersions, especially the volume fraction dependences of viscosity and critical shear rates (onset of dilatancy, maximum and discontinuity in viscosity) also the effects of particle size, polydispersity and suspending fluid viscosity.

Les systèmes électriques ou électromécaniques doivent satisfaire à des spécifications de plus en plus contraignantes qui nécessitent la mise au point de structures de commande non linéaires. Parmi celles-ci, la commande par logique floue constitue une alternative intéressante et performante. Son principal handicap réside dans le nombre très important de paramètres à régler. Dans cet article, nous nous proposons de systématiser ces réglages dans deux cas de figure. Tout d'abord nous utiliserons la méthodologie des plans d'expérimentations pour effectuer un réglage sur site d'un contrôleur flou de type proportionnel-intégral. Ce réglage sera obtenu en ne réalisant qu'un nombre limité d'essais expérimentaux en boucle fermée avec des combinaisons prédéfinies des paramètres à régler. La combinaison optimale de ces paramètres au sens d'un critère de type IAE (Intégrale de la valeur Absolue de l'Erreur) sera déduite de l'exploitation des résultats des essais. Dans un deuxième temps, nous proposerons des réglages prédéfinis et optimisés (au sens du même critère) d'un contrôleur flou de type proportionnel-intégral-dérivé. Ces réglages préétablis ne nécessiteront qu'un seul essai d'identification du système à contrôler en boucle ouverte et peuvent donc se rapprocher des méthodologies classiques et éprouvées de réglage sur site que constituent les réglages de Ziegler-Nichols ou de Broïda pour des contrôleurs conventionnels. Dans cet article, les jeux de paramètres préétablis que nous fournirons seront valables pour des systèmes dont la réponse indicielle en boucle ouverte est modélisable sous la forme d'une fonction de transfert du premier ordre plus un retard pur. Les limites de validité de cette méthode seront précisées.

Les algorithmes génétiques sont des méthodes adaptatives de plus en plus utilisée pour la résolution de certains problèmes d'optimisation. Le présent travail consiste d'une part, à mettre en œuvre un A.G sur des problèmes d'identification des machines électriques, et d'autre part à comparer ses performances avec les méthodes classiques tels que la méthode du maximum de vraisemblance et la méthode électrotechnique basée sur des essais à vides et en court-circuit. Ces méthodes sont appliquées sur des machines asynchrones de différentes puissances. Les résultats obtenus sont comparés selon certains critères, permettant de conclure sur la validité et la performance de chaque méthode.

Shock wave generation from laser-produced plasma in water confinement regime for an incident 25–30 ns / 40 J / λ = 1.064 µm laser beam has been studied. Transmitivity measurement of the parasitic plasma breakdown at the surface of water confinement has been performed with a continuous laser probe. Above 6 GW/cm2 incident laser intensity, the critical electronic density for the laser wavelength is achieved inside the plasma by cascade ionization during the pulse duration. Above 10 GW/cm2, peak power density transmitted through the plasma saturates. The laser pulse transmitted through the breakdown plasma corresponds to the part of incident laser pulse preceding the transmission cut-off.

The use of rheometers for the evaluation of rheological properties and the establishment of behaviour laws for fluids requires the knowledge of the shear rate at any moment and everywhere in the region between the cone and the plate, referred to as “the gap” throughout this paper. However, the accurate determination of the shear rate supposes that the constitutive equation of the fluid is known beforehand. In order to avoid this paradox, rheometers are generally built such that the shear rate is supposed to be approximately constant throughout the gap. This approximation is realistic for steady flow but may be crude for other types of fluid motion. The aim of the present work is to determine the limits of validity of such an approximation when testing complex fluids in a cone and plate geometry. In this paper, only purely viscous properties are taken into account. The numerical solution is based on the control-volume method. When non-linear and time-dependent effects occur, it is shown that the flow cannot be represented by simple shearing conical surfaces. This result is especially important for the characterisation of time-dependent fluids (as thixotropic fluids), typical of unsteady flow. Finally the abilities of the proposed model which is named “RHEOUTIL” are highlighted by comparison between numerical simulations and experimental results. The analysis of non Newtonian fluids emphasises the limits of the code “RHEOUTIL”. Indeed, the model has to evolve in order to take into account the whole complexity of the fluid, which may also exhibit viscoelastic properties, yield stress and so on.

The temperature profiles in wall-stabilized atmospheric pressure arcs in argon and Ar-CO2 mixture have been obtained at currents of 5 A to 40 A using spectroscopy techniques. The temperature profiles were measured using the absolute intensity of spectral lines. The monatomic excitation temperatures are determined experimentally from three chosen spectral lines of argon (415.86, 427.22, 696.54 nm), spectral lines of carbon (538.03, 477.175 nm) and spectral lines of triple oxygen (777.19, 777.42, 777.54 nm) in wall-stabilized arc. The results show the plasma had not the same composition locally as the gas injected in to the middle of the arc column. The results obtained are in good agreement with previously published data for pure argon and show the gradient formation of an arc core and the increase in axis temperature as the gas mixture is varied from pure argon to Ar-CO2 mixture. Another result of these investigations is to show the importance of high purity gas when using monatomic gases, the addition of even a few per cent of CO2 to argon having a drastic effect both upon extinction of the arc according to current values and upon the temperature profiles.

The design of the main current systems of an actively shielded and of an iron shielded MRI device for nuclear resonance imaging, is considered. The model for the analysis of the magnetic induction produced by the current system, is based on the combination of a Boundary Element technique and of the integration of two Fredholm integral equations of the first and the second kind. The equivalent current magnetization model is used for the calculation of the magnetization produced by the iron shield. High field uniformity in a spherical region inside the device, and a low stray field in the neighborhood of the device are required. In order to meet the design requirements a multi-objective global minimization problem is solved. The minimization method is based on the combination of the filled function technique and the (1+1) evolution strategy algorithm. The multi-objective problem is treated by means of a penalty method. The actively shielded MRI system results to utilize larger amount of conductor and produce higher magnetic energy than the iron shield device.

It is shown that the production of electrons by the laser-gas interaction, described in a previous paper [1], at a given position in the interelectrode gap of a DC electrical discharge, is responsible for a transient current associated to the displacement of the running point on the current-voltage characteristic. A theoretical model of the laser-induced current impulse, in good agreement with experiments, is proposed in the case of homogeneous field. Some macroscopic coefficients, like first and second ionization coefficients, or electrons and ions velocities, are examined. The estimated values are compared with ones found in the literature. The theory in homogeneous field, well supported by the experiments, allows to give a qualitative interpretation to the laser-induced current impulse in inhomogeneous field (point-to-plane geometry). As a result, depending on the location of the running point on the I−V characteristic and on the laser energy available for photon-ionization, it is shown that the laser-gas interaction is an accurate and effective technique for the investigation of the fundamental processes involved in each electrical state of the discharge (townsend discharge, self-sustained discharge, glow discharge) and in the transition from a state to another (for example, avalanche-to-streamer transition).

The possibility of characterizing immersed targets in laboratory conditions, whatever the water surface roughness is discussed in this paper. The first idea was to reduce the interface effects or better to suppress them within the analysis.We used the Mueller formalism to characterize targets in the laboratory. In optics, each element can be described by a Mueller matrix. Then, in the present work we tried to decompose the global measured matrix and to separate the interface matrix from the target matrix which is the one of interest. This paper shows it impossible to perform such a decomposition. We succeeded in reducing the interface effects without eliminating them entirely. So this paper focuses on finding a decomposition algorithm of a Mueller matrix to deduce the polarimetric type of a target whatever the interface state.

In RF cold plasma reactors, energy transfer from the generator to the discharge is not perfect and only 50 to 70 percent (or even less) of the power is effectively transmitted to the discharge, the reMayning part being mostly dissipated in the matching network. The choice of this circuit is therefore the most important factor in view to optimize the energy transfer from the power supply to the plasma. Two classical networks, Π- and L-type, are analytically studied in their interaction with the experimental device. From this study, in both cases, numerical values are deduced for the coil inductance, and a capacitance range is determined for the capacitors. An experimental study in argon plasma at pressure range of 40-200 mTorr and at 13.56 MHz, performed with the two types of matching networks leads to a much effective energy transfer with the L-type, which must be chosen in order to improve the energetic yield of surface processing (sputtering or etching for instance) with a RF plasma reactor.

Our work presents the study of bone remodeling scheme in the general case of n unit-elements model. The analytical result for the stationary states presented here as original was obtained and a condition which is necessary for the stability of the remodeling theory was derived. Forward finite-difference time-stepping is used to verify the prediction of the analytical study. This numerical study appears to confirm the analytical results.

L'objectif de cet article est de présenter la synthèse d'un modèle d'analyse pluridisciplinaire adapté à la caractérisation des performances d'un entraînement électrique alimenté à fréquence variable sous faible tension. Le modèle prend en compte le couplage entre les phénomènes électromagnétique, thermique et électrique. Il permet d'étudier le fonctionnement sous les conditions de régime thermique permanent ou transitoire qui interviennent dans la définition des cycles de travail des engins de traction à énergie embarquée. Ce modèle d'analyse constitue un élément d'une procédure de dimensionnement optimal qui sera développée dans un article séparé ; aussi l'exposé est-il conduit en explicitant les dépendances fonctionnelles entre les paramètres de dimensionnement et les performances. La procédure d'analyse est appliquée à l'étude d'un moteur à aimants permanents périphériques alimenté suivant le mode "brushless trapézoïdal" ; les performances sont évaluées pour deux types de cycle de travail : le cycle S1 en régime permanent, et le cycle S2 en régime thermique transitoire ; l'influence des paramètres de commande du convertisseur est examinée.