En microélectronique, plusieurs tendances telles que l'empilement 3D et l'amincissement de puces amènent des défis thermiques grandissants. Ces défis sont exacerbés lorsqu'appliqués aux appareils mobiles où l'espace et la puissance disponibles pour le refroidissement sont limités. Le but de cette thèse est de développer des outils de conception et méthodes d'implémentation de microcanaux pour le refroidissement microfluidique de puces 2D et 3D avec points chauds destinés aux appareils mobiles.Une méthode de conception pour optimiser la configuration des microcanaux refroidissant une puce est développée utilisant un plan d'expériences numériques. La configuration optimisée propose le refroidissement à une température maximale de 89 °C d'un point chaud de 2 W par un écoulement où la perte de charge est plus petit que 1 kPa. Des prototypes avec différents empilements et distributions de microcanaux sont fabriqués par gravure profonde et apposés par pick-and-place. Un banc de caractérisation et une puce thermique test sont fabriqués pour caractériser expérimentalement les prototypes de refroidissement avec différentes configurations. Un prototype avec microcanaux limités aux alentours des points chauds et reportés sur la face arrière de la puce test atteint une résistance thermique de 2.8 °C/W. Cela est réalisé avec un débit de 9.4 ml/min et des pertes de charges de 19.2 kPa, soit une puissance hydraulique de 3 mW. Ce refroidissement extrait 7.3 W générés sur un seul serpentin à un flux thermique de 1 185 W/cm² pour un coefficient de performance de 2 430. Les résultats de l'optimisation suggèrent que la dissipation thermique soit exploitée en ajoutant des microcanaux en parallèle, plutôt qu'en allongeant les microcanaux. On observe expérimentalement comme numériquement que la résistance liée à la hausse de température du fluide domine la résistance totale. Enfin, il apparaît que les différents empilements ont un effet plus important sur la résistance thermique que les distributions de microcanaux dans les plages observées. / In microelectronics, trends such as 3D stacking and die thinning bring major thermal challenges. Those challenges are exacerbated when applied to mobile devices where the available space and power for cooling are limited. This thesis aims at developing design tools and implementation techniques for microchannels cooling on 2D and 3D chips with hot spots for mobile devices. A design technique to optimize the microchannel configuration for chip cooling is developed using numerical experimentation plans. The optimized configuration suggests a cooling configuration reaching a maximum temperature of 89 °C on a 2 W hot spot, using a flow at a pressure drop plus petit que 1 kPa. Prototypes with different stacking and microchannel distributions are fabricated using deep reactive ion etching process and stacked using pick-and-place technique. A characterization bench and a thermal test chip are fabricated for experimental characterization of the cooling prototypes from various configurations. A prototype with microchannel zones limited to the hot spot vicinity and installed on the backside of the test chip reached a thermal resistance of 2.8 °C/W. This performance is achieved using a flow rate of 9.4 ml/min with a pressure drop of 19.2 kPa, representing a hydraulic power of 3 mW. Such cooling removes 7.3 W generated on a single heat source, representing a heat flux of 1 185 W/cm² for a coefficient of performance of 2 430. The optimization results suggest that the heat spreading is better exploited using parallel microchannels, rather than lengthen microchannels. It is both observed experimentally and numerically that the thermal resistance related to the fluid temperature rise is the major contribution to the total thermal resistance. Finally, it appears that the different stacking effects on thermal resistance are more important than the microchannels distributions in the observed ranges.

Use of geosynthetics for reinforcing soil beds supporting shallow foundations has gained tremendous popularity in recent times. In this thesis, to study and understand the behaviour of geosynthetics reinforced soil foundations, model load tests are carried out on square footings resting on sand beds reinforced with geosynthetics. The effects of various parameters like type and tensile strength of geosynthetic material, depth of reinforced zone, spacing of reinforcement layers, width of reinforcement and form of reinforcement on the performance of square footings on reinforced sand beds are studied. Results from these tests are analyzed to understand the effect of various parameters in improving the bearing capacity and reducing the settlement of footings. An equation is developed to estimate the ultimate bearing capacity of square footings resting on geosynthetic reinforced sand beds by multiple regression analysis of the experimental data. The model loading tests on reinforced soil foundations are simulated in the numerical model using the computer program FLAC3D (Fast Lagrangian Analysis of Continua in 3D). Finally parametric studies on a full scale reinforced soil foundation are conducted. From the experimental, analytical and numerical investigations carried out in this thesis, some important conclusions are drawn regarding the effective depth of reinforced zone, optimum spacing and quantity of reinforcement layers. Relative efficiency of various forms of reinforcement is discussed. Validity of the regression and numerical models developed is verified through experimental data from present study and also for data from other researchers.

In the present thesis, studies on the response of soil nail walls subjected to static and seismic conditions using finite element based numerical simulations and the principle of reliability analysis have been performed. The basic methodology constitutes the study of various aspects of soil nail walls such as analyses of important external, internal and facing failure modes, development of axial forces, and displacement observations by considering various typical and prototype cases. For better understanding and presentation, subject matter of the thesis is organised in the following ten chapters. Chapter 1 of the thesis provides an introduction to the soil nailing technique and highlights some of its applications, advantages, and limitations. Chapter 2 provides a detailed review of existing literature on the soil nailing technique. Chapter 3 provides a detailed overview the various methodologies adopted in the thesis for the analyses and response study of the soil nail walls. Chapter 4 deals with the important aspects related to the plane strain finite element based numerical simulations of soil nail walls. In particular, addresses the implications of the use of advanced soil models and the consideration of bending stiffness of soil nails on the overall response of the soil nail walls. Chapter 5 presents finite element simulations based appraisal of the conventional design methodology of soil nail walls, and studies the response of typical soil nail walls under static and seismic conditions. Chapter 6 presents a reliability based study of the important failure modes of soil nail walls subjected to the variability in in-situ soil parameters, and highlights the importance of reliability analysis in context of soil nail walls. Chapter 7 proposes load and resistance factor design (LRFD) methodology in context of soil nail walls, and highlights the need in advancement of the existing conventional design methodology for soil nail walls. Chapter 8 illustrates the use of factorial design of experiment methodology in developing regression models for stability criteria analysis of soil nail walls. Chapter 9 proposes methods for assessing the adequacy of field pullout tests performed in accordance with the prevalent soil nailing guidelines. Further, a reliability based methodology is proposed for the evaluation and various applications of field pullout tests results have been illustrated. Chapter 10 summarises the various studies reported in the thesis and provides a few important conclusions. It is believed that the various studies reported in the thesis contribute to the enhancement of the existing knowledge on soil nailing technique, advancement in the analysis and design methods, and in general, are useful to the soil nailing practice.

Joints are critical parts of timber structures, transmitting static and dynamic forces between structural members. The ultimate behavior of e.g. a building depends strongly on the structural configuration and the capacity of its joints. The complete collapse of a building or other less extensive accidents that may occur usually start as a local failure inside or in the vicinity of a joint. Such serious failures have recently occurred in the Nordic countries. Especially the collapses of two large glued-laminated timber (glulam) structures clearly indicate the need of an improved joint design. The trend toward larger and more complex structures even further increases the importance of a safer design of the joints. One aim of this partly experimentally and partly numerically based work has been to investigate if the short term capacity of steel-timber dowel joints loaded parallel to the grain is affected by an initial drying exposure. The experimental results showed that the load-bearing capacity of the joints is indeed reduced by such moisture changes. Moisture induced stresses was mentioned to be the explanation. The key point is that the climates chosen in the present work (20°C / 65% RH and 20°C / 20% RH) are equivalent to service class 1 according to EC5 (Eurocode 5 2004). Thus, EC5 predicts no decrease in load-bearing capacity, in relation to the standard climate used during testing. A decrease in load-bearing capacity in the range of 5-20%, which was found in the present work, is of course not negligible and, therefore, there could be a need to introduce the effect of drying in design codes. Because similar results were also observed for a double-tapered glulam beam, further work should consider timber structures in general. Two numerical methods in order to predict the capacity of multiple steel-timber dowel joints loaded parallel to the grain were tested in the thesis. For the first method, where fracture mechanics (LEFM) concepts were implemented, a good correlation with the experimental results was seen. Also for the second method, where the capacity for a single dowel-type joint as given in EC5 was used as a failure criterion, a good correlation to traditional EC5 calculations of multiple dowel-type joints was seen. One advantage of using numerical methods in design is that the capacity of the joint can be calculated also for cases when the dowels are placed in more complex patterns. From both a structural and an architectural point of view this can be very important. In addition, such numerical methods are effective tools for the structural engineer when considering complicated loading situations in joints, i.e. eccentric loading giving moments in the joint.

In this thesis direct numerical simulations are used to investigate two phenomenain shear flows: laminar-turbulent transition over a flat plate and periodicvortex shedding induced by a jet in cross flow. The emphasis is on understanding and controlling the flow dynamics using tools from dynamical systems and control theory. In particular, the global behavior of complex flows is describedand low-dimensional models suitable for control design are developed; this isdone by decomposing the flow into global modes determined from spectral analysisof various linear operators associated with the Navier–Stokes equations.Two distinct self-sustained global oscillations, associated with the sheddingof vortices, are identified from direct numerical simulations of the jet incrossflow. The investigation is split into a linear stability analysis of the steadyflow and a nonlinear analysis of the unsteady flow. The eigenmodes of theNavier–Stokes equations, linearized about an unstable steady solution revealthe presence of elliptic, Kelvin-Helmholtz and von K´arm´an type instabilities.The unsteady nonlinear dynamics is decomposed into a sequence of Koopmanmodes, determined from the spectral analysis of the Koopman operator. Thesemodes represent spatial structures with periodic behavior in time. A shearlayermode and a wall mode are identified, corresponding to high-frequency andlow-frequency self-sustained oscillations in the jet in crossflow, respectively.The knowledge of global modes is also useful for transition control, wherethe objective is to reduce the growth of small-amplitude disturbances to delaythe transition to turbulence. Using a particular basis of global modes, knownas balanced modes, low-dimensional models that capture the behavior betweenactuator and sensor signals in a flat-plate boundary layer are constructed andused to design optimal feedback controllers. It is shown that by using controltheory in combination with sensing/actuation in small, localized, regionsnear the rigid wall, the energy of disturbances may be reduced by an order of magnitude.

Knowledge of flow separation is very essential for proper understanding of both external and internal aerothermodynamics of bodies. Because of unique flow features such as thick boundary layers, merged shock layers, strong entropy layers, flow separation in the flow field of bodies at hypersonic speeds, is both complex as well as interesting. The problem of flow separation is further complicated at very high stagnation enthalpies because of the real gas effects. Notwithstanding the plethora of information available in open literature even for simple geometric configurations the experimentally determined locations of flow separation and re-attachment points do not match well with the results from the computational studies even at hypersonic laminar flow conditions. In this backdrop the main aim of the present study is to generate a reliable experimental database of classical separated flow features around generic configurations at hypersonic laminar flow conditions. In the present study, flow visualization using high speed camera, surface convective heat transfer rate measurements using platinum thin film sensors, and direct skin friction measurements using PZT crystals have been carried out for characterizing the separated flow field around backward facing step, double cone and double wedge models. The numerical simulations by solving the Navier-Stokes equations have also been carried out to complement the experimental studies. The generic models selected in the present study are simple configurations, where most of the classical hypersonic separated flow features of two-dimensional, axi-symmetric and three dimensional flow fields can be observed. All the experiments are carried out in IISc hypersonic shock tunnel (HST2) at Mach 5.75 and 7.6. For present study, helium and air have been used as the driver and test gases respectively. The high speed schlieren flow visualization is carried out on backward facing step (2 and 3 mm step height), double cone (semi-apex angles of 150/350 and 250/680) and double wedge (semi-apex angles of 150/350) models by using high speed camera (Phantom 7.1). From the visualized shockwave structure in the flow field the flow reattachment point after separation has been clearly identified for backward facing step, double cone and double wedge models at hypersonic Mach numbers while the separation point could not be clearly identified because of the low free stream density in shock tunnels. However the flow visualization studies helped clearly identifying the region of flow separation on the model. Based on the results from the flow visualization studies both the physical location and distribution of platinum thin film gauges was finalized for the heat transfer rate measurements. Surface heat transfer rates along the length of two backward facing step (2 and 3 mm step height) models have been measured using platinum thin film gauges deposited on Macor substrate. The Eckert reference temperature method is used along the flat plate for predicting the heat flux distribution. Theoretical analysis of heat flux distribution down stream of the backward facing step model has been carried out using Gai’s dimensional analysis. The study reveals for the first time that at moderate stagnation enthalpy levels (~2 MJ/kg) the hypersonic separated flow around a backward facing step reattaches rather smoothly without any sudden spikes in the measured values of surface heat transfer rates. Based on the measured surface heating rates on the backward facing step, the reattachment distance was estimated to be approximately 10 and 8 step heights downstream of 2 and 3 mm step respectively at nominal Mach number of 7.6. Convective surface heat transfer experiments have also been carried out on axi-symmetric double cone models (semi-apex angles of 15/35 and 25/68), which is analogous to the Edney’s shock interactions of Type VI and Type IV respectively. The flow is unsteady on the double cone model of 25/68 and measured heat flux is not constant. The heat transfer experiments were also carried out on the three-dimensional double wedge model (semi-apex angles of 15/35). The separation and reattachment points have been clearly identified from the experimental heat transfer measurements. It has been observed that the measured heat transfer rates on the double wedge model is less than the double cone model (semi-apex angles of 150/350) for the identical experimental conditions at the same gauge locations. This difference could be due to the three-dimensional entropy relieving effects of double wedge model. PZT-5H piezoelectric based skin friction gauge is developed and used for direct skin friction measurements in hypersonic shock tunnel (HST2). The bare piezoelectric PZT-5H elements (5 mm × 5 mm with thickness of 0.75 mm) polarized in the shear mode have been used as a skin friction gauge by operating the sensor in the parallel shear mode direction. The natural frequency of the skin friction sensor is ~80 kHz, which is suitable for impulse facilities. The direct skin friction measurements are carried out on flat plate, backward facing step (2 mm step height) and double wedge models. The measured value of skin friction coefficient (integrated over an area of 25 sq. mm; sensor surface area) at a distance of 23 mm from the leading edge of the sharp leading edge backward facing step model is found to be ~ 0.0043 while it decreases to ~ 0.003 at a distance of 43 mm from the leading edge at a stagnation enthalpy of ~ 2MJ/kg. The measured skin friction matches with the Eckert reference temperature within ± 10%. The skin friction coefficient is also measured on the double wedge at a distance of 73 mm from the tip of the first wedge along the surface and is found to be 4.56 × 10-3. Viscous flow numerical simulations are carried out on two-dimensional backward facing step, axi-symmetric double cone and three-dimensional double wedge models using ANSYS-CFX 5.7 package. Navier-Stokes Simulations are carried out at Mach 5.75 and 7.6 using second order accurate (both in time and space) high resolution scheme. The flow is assumed to be laminar and steady throughout the model length except on the double cone (semi-apex angles of 250/680) model configuration, which represents the unsteady flow geometry. Analogous Edney Type VI and Type IV shock interactions are observed on double cone, double wedge (semi-apex angles of 150/350) and double cone (semi-apex angles of 250/680) models respectively from the CFD results. Experimentally measured convective heat transfer rates on the above models are compared with the numerical simulation results. The numerical simulation results matches well with the experimental heat transfer data in the attached flow regions. Considerable differences are observed between the measured surface heat transfer rates and numerical simulations both in the separated flow region and on the second cone/wedge surfaces. The separation and reattachment points can be clearly identified from both experimental measurements and numerical simulations. The results from the numerical simulations are also compared with results from the high speed flow visualization experiments. The experimental database of surface convective heating rates, direct skin friction coefficient and shockwave structure in laminar hypersonic flow conditions will be very useful for validating CFD codes

No description available.

Le travail présenté dans ce mémoire avait pour objectif d'étudier le processus de la pénétration d'eau dans les composites en carbone/époxyde dans un premier temps, et dans un deuxième temps, d'étudier l'effet de la prise en eau par ces matériaux sur les performances mécaniques des composites et leur joints collés. L'intégration de ces phénomènes physiques dans la modélisation numérique est d'une grande importance dans la prédiction de la durabilité d'une structure en composite subissant un vieillissement hygrothermique. Par conséquent, ce travail consiste non seulement en des observations expérimentales, mais aussi en des simulations numériques. Des corrélations entre les résultats obtenus permettent d'une part de mieux comprendre ce qui se passe dans un système composite avec l'assemblage collé soumis à des charges mécaniques, de l'initiation d'endommagement jusqu'à la rupture finale ; d'autre part, de valider un modèle numérique robuste dans le but de la conception et de l'optimisation. Les originalités de ce travail se situent à différents niveaux en proposant : 1. un nouveau modèle de diffusion à deux-phases permettant de mieux décrire l'effet de l'épaisseur des stratifiés sur la pénétration de l'eau; 2. un nouveau modèle RPM " Residual Property Model " afin de prévoir la dégradation des propriétés mécaniques due à la prise en eau ; 3. une nouvelle loi de traction-séparation linéaire-exponentiel pour décrire la courbe-R observée dans les essais DCB en mode I pur sur les composites stratifiés afin de les intégrer plus facilement dans les modèles numériques

L’utilisation d’une méthode d’assimilation de données, associée à un modèle de convection anélastique, nous permet la reconstruction des structures physiques d’une partie de la zone convective située en dessous d’une région solaire active. Les résultats obtenus nous informent sur les processus d’émergence des tubes de champ magnétique au travers de la zone convective ainsi que sur les mécanismes de formation des régions actives. Les données solaires utilisées proviennent de l’instrument MDI à bord de l’observatoire spatial SOHO et concernent principalement la région active AR9077 lors de l’ ́évènement du “jour de la Bastille”, le 14 juillet 2000. Cet évènement a conduit à l’avènement d’une éruption solaire, suivie par une importante éjection de masse coronale. Les données assimilées (magnétogrammes, cartes de températures et de vitesses verticales) couvrent une surface de 175 méga-mètres de coté acquises au niveau photosphérique. La méthode d’assimilation de données employée est le “coup de coude direct et rétrograde”, une méthode de relaxation Newtonienne similaire à la méthode “quasi-linéaire inverse 3D”. Elle présente l’originalité de ne pas nécessiter le calcul des équations adjointes au modèle physique. Aussi, la simplicité de la méthode est un avantage numérique conséquent. Notre étude montre au travers d’un test simple l’applicabilité de cette méthode à un modèle de convection utilisé dans le cadre de l’approximation anélastique. Nous montrons ainsi l’efficacité de cette méthode et révélons son potentiel pour l’assimilation de données solaires. Afin d’assurer l’unicité mathématique de la solution obtenue nous imposons une régularisation dans tout le domaine simulé. Nous montrons enfin que l’intérêt de la méthode employée ne se limite pas à la reconstruction des structures convectives, mais qu’elle permet également l’interpolation optimale des magnétogrammes photosphériques, voir même la prédiction de leur évolution temporelle. / We use a data assimilation technique, together with an anelastic convection model, in order to reconstruct the convective patterns below a solar active region. Our results yield information about the magnetic field emergence through the convective zone and the mechanisms of active region formation. The solar data we used are taken from the instrument MDI on board the spatial observatory SOHO on July 2000 the 14th for the event called ”bastille day event”. This specific event leads to a solar flare followed by a coronal mass ejection. Assimilated data (magnetograms, temperature maps and vertical velocity maps) cover an area of 175 Mm × 175 Mm at photospheric level. The data assimilation technique we used, the ”Nudging Back and Forth”, is a Newtonian re- laxation technique similar to the ”quasi linear inverse 3D”. Such a technique does not require computation of the adjoint equations. Thus, simplicity of this method is a numerical advantage. Our study shows with a simple test case the applicability of this method to a convection model treated with the anelastic approximation. We show the efficiency of the NBF technique and we detail its potential for solar data assimi- lation. In addition, to ensure mathematical unicity of the obtained solution, a regularization has been imposed in the whole simulation domain. This is a new approach. Finally, we show that the interest of such a technique is not limited to the reconstruction of convective patterns but that it also allows optimal interpolation of photospheric magnetograms and predictions.

This work is concerned with the theoretical description of the Scanning Gate Microscopy (SGM) in general and with solving particular models of the quantum point contact (QPC) nanostructure, analytically and numerically. SGM is an experimental technique, which measures the conductance of a nanostructure, while a charged AFM tip is scanned above its surface. It gives many interesting results, such as lobed and branched images, interference fringes and a chequerboard pattern. A generally applicable theory, allowing for unambiguous interpretation of the results, is still missing. Using the Lippman-Schwinger scattering theory, we have developed a perturbative description of non-invasive SGM signal. First and second order expressions are given, pertaining to the ramp- and plateau-regions of the conductance curve. The maps of time-reversal invariant (TRI) systems, tuned to the lowest conductance plateau, are related to the Fermi-energy charge density. In a TRI system with a four-fold spatial symmetry and very wide leads, the map is also related to the current density, on any plateau. We present and discuss the maps calculated for two analytically solvable models of the QPC and maps obtained numerically, with Recursive Green Function method, pointing to the experimental features they reproduce and to the fundamental difficulties in obtaining good plateau tuning which they reveal.