Variations of the optical detection path-length in image correlation based flow-field measurements result in strong errors in position allocation and thus lead to a strong enhancement of the measurement uncertainty of the velocity. In this contribution we use digital holography to measure the wavefront distortion induced by fluctuating phase boundary, employing spatially extended guide stars. The measured phase information is used to correct the influence of the phase boundary in the detection path employing a spatial light modulator. We analyze the potential of guide stars that are reflected by the phase boundary, i.e. the Fresnel reflex, and transmitted. Our results show, that the usage of wavefront shaping enables to strongly reduce the measurement uncertainty and to strongly improve the quality of image correlation based flow-field measurements. The approaches presented here are not limited to application in flow measurement, but could be useful for a variety of applications.

This thesis presents performance evaluation and a field validation study of a time and temperature indexed autoregressive with exogenous (4-3-5 ARX) building thermal load prediction model with an aim to integrate the model with actual predictive control systems. The 4-3-5 ARX model is very simple and computationally efficient with relatively high prediction accuracy compared to the existing sophisticated prediction models, such as artificial neural network prediction models. However, performance evaluation and field validation of the model are essential steps before implementing the model in actual practice. The performance of the model was evaluated under different climate conditions as well as under modeling uncertainty. A field validation study was carried out for three buildings at Mississippi State University. The results demonstrate that the 4-3-5 ARX model can predict building thermal loads in an accurate manner most of the times, indicating that the model can be readily implemented in predictive control systems.

Det byggs mer och mer med prefabricerade håldäcksbjälklag vilka har stora dimensionerings och produktionsfördelar. Långa spännvidder och lätta bjälklag gör denna typ av konstruktion känslig för vibrationer och därför ställs konstruktörer idag inför en problematik som inte tidigare vanligtvis varit en dimensionerande faktor. Som en naturlig följd av detta är kunskapen inom området begränsad och förståelsen för denna typ av konstruktion vid dynamiska problem eftersatt. Hur håldäcksbjälklag beter sig vid små dynamiska laster, från mänsklig aktivitet på bjälklaget, och hur denna typ av problematik ska hanteras beräkningsmässigt behöver studeras. Vid statiska beräkningar antas bjälklaget vara enkelspänt men vid dynamiska beräkningar kan bland annat inte detta antagande göras då effekten från elementens sammankopplingar och eventuella pågjutningar påverkar bjälklagets styvheter och därmed dess dynamiska beteende. Dynamikproblem kan inte behandlas genom att studera enskilda element då bjälkaget måste hanteras som en sammanhängande enhet där kringliggande struktur påverkar hela bjälklagets dynamiska beteende. I detta arbete har fältmätningar i en befintlig byggnad, med excitering av en mänsklig analyser för att studera hur bjälklaget kan hanteras och dynamiska responsberäkningar. Fältmätningar har genomförts på tre olika bjälklagstyper där accelerationer har studerats i en mängd olika punkter. Accelerationer i fältmitt, intill upplagsväggar, parallella väggar och balkar har studerats för att leda till slutsatser kring hur bjälklagets styvheter påverkar dess dynamiska beteende. Dessa slutsatser kring uppmätta accelerationer har sedan legat till grund modeller kan representera verkligheten på ett godtagbart sätt beträffande dynamiska problem. Finita element-beräkningar för de tre olika bjälklagen har genomförts med varierande resultat, där det kunnat konstateras att komplicerade bjälklagsgeometrier med inslag av pelare, balkar och flera fack håldäckselement är svårt att beräkningsmässigt representera på ett bra sätt. Däremot har det visat sig att enkla bjälklagsgeometrier med endast ett fack håldäckselement upplagt på betongväggar kan representeras relativt väl. Parallella väggar har kunnat konstateras fungera som upplag med accelerationer av samma storleksordning som intill upplagsväggar. Dessutom har det visat sig genom mätningar och beräkningar att en pågjutning rimligen kan antas ha en uppstyvande effekt för bjälklaget. Däremot hur stor denna uppstyvande effekt är har inte kunnat konstateras, då detta skulle kräva andra typer av mätningar.

La capacité opérationnelle axiale en service de pieux battus reste une zone d'incertitude, en particulier pour les structures offshore. La recherche sur le terrain a montré que le frottement latéral peut augmenter au cours des mois ou des années après le battage. Si des tendances similaires se retrouvent dans des ouvrages offshore, les avantages en terme d'ingénierie de réalisation peuvent être très importants. D'autre part, les fondations sur pieux de plates-formes de gaz, de pétrole sont soumises à des chargements cycliques à long terme qui peuvent influencer leur capacité à l'arrachement. Les pieux battus en eau profonde connaissent un grand nombre de cycles complets de charge-décharge pouvant contribuer à la dégradation du frottement latéral lors de l'installation. Cette thèse vise à mieux comprendre les principaux résultats obtenus avec des pieux réels en sable siliceux, par le biais d'une recherche à échelle de laboratoire sous conditions environnementales contrôlées. Ce travail fait partie d'un programme de recherche commun entre le Laboratoire 3SR de Grenoble, l'Imperial College London, et le projet français de recherche ANR- SOLCYP. La réponse de l'interface sol-pieu lors de l'installation ainsi que les périodes de vieillissement et de chargements cycliques axiaux ont été étudiés au laboratoire en utilisant des pieux-modèles installés dans la chambre d'étalonnage de Grenoble. Plusieurs essais avec le pieu modèle Mini-ICP (instrumenté avec des capteurs de tension totale à la surface du pieu (SST) pour les mesures de contraintes radiales de cisaillement à 3 sections) ont permis l'analyse de chemin de contrainte locale à l'interface du pieu. Des capteurs miniatures ont en outre été installés dans le massif de sable pour une mesure de contrainte lors de l'installation du pieu et son chargement ultérieur. Les effets des méthodes d'installation, de la taille des particules de sable, ou de la saturation du sable et du chargement de l'environnement, ont été pris en compte pour le vieillissement de la capacité. Les évolutions locales de l'interface radiale et du cisaillement sont en accord avec les prédictions des méthodes de conception modernes basées sur le CPT. Des preuves d'effets d'échelle soulignent l'importance des conditions aux limites appliquées à la modélisation physique. Des séries d'essais non-alternés purement en traction, ainsi que des essais alternés ont été réalisés sous contrôle en charge ou en déplacement. Les mesures locales effectuées dans les chemins de contraintes effectives montrent une contraction radiale de la masse de sable au voisinage du pieu. Les incréments de l'amplitude de charge et du déplacement imposé accélèrent les taux de dégradation cyclique. Un nouveau diagramme de stabilité cyclique a été réalisé, en résumant les essais de chargement cycliques axiaux pour les pieux foncés et battus dans du sable siliceux moyennement dense. Des mécanismes complexes comme la rupture des grains et des changements de densité locale à l'interface du pieu peuvent affecter la réponse des pieux. La cinématique derrière leur installation et l'interaction avec le sol environnant reste encore très limitée. Comprendre comment un matériau granulaire interagit avec le pieu est important pour étudier la réponse globale du pieu. Les observations globales du comportement des pieux dans la chambre d'étalonnage ont été modélisées à une échelle micro en utilisant la tomographie aux rayons X du Laboratoire 3SR à Grenoble. Le programme expérimental comprenait des essais sur une chambre d'étalonnage modèle afin d'analyser le champ de déplacement lors de l'installation d'un pieu modèle, à l'aide des techniques de corrélation d'images (DIC) en trois dimensions. Des analyses micromécaniques d'échantillons «intacts» récupérés post mortem à l'interface du pieu ont été également effectuées pour mettre en évidence de possibles changements radiaux de densité ainsi que la rupture des grains. / The operational in-service axial capacity of driven piles remains an area of uncertainty, especially for offshore structures. Field research has demonstrated that axial shaft capacities may increase over the months or years after driving. If similar trends apply offshore, the realisable engineering benefits are very significant. On the other hand, the piled foundations of oil/gas platforms and wind/water turbines are subject to long term environmental and in service cyclic loading due for example to waves, vibrations and storms that may also affect their shaft capacity. Deep driven piles experience large numbers of full load-unload cycles that contribute to shaft capacity degradation during installation. This thesis aims to improve understanding of the main results obtained with full-scale piles in silica sand through a laboratory scale investigation performed under controlled environmental conditions. This work was part of a joint research programme between the Grenoble Laboratory 3SR and Imperial College London, and the French National SOLCYP research project. The response of the soil-pile interface during installation, ageing periods and cyclic loading tests have been studied using laboratory model piles installed in the large Grenoble Calibration Chamber. Several tests with the Mini-ICP pile allow the analyses of local stress path at the pile's interface. This model pile is instrumented with surface stress transducers (SST) for local measurements of total and radial shear stresses at 3 different sections along the pile's shaft. In addition, miniature soil stress transducers were installed into the sand mass for total stress measurements during pile installation and loading. Possible ageing effects as installation methods, sand particle size, sand saturation and environmental loading were studied. Local evolution of interface radial and shear stresses agree with predictions from modern CPT based design methods. Evidence of possible scale effects remark the importance of the boundary conditions applied in physical modelling. Series of one-way purely tensile and two-way axial cyclic loading tests were performed under load and displacement control. Local measurements made of the effective stress paths shows radial contraction of the sand mass in the vicinity of the pile. Increments in loading amplitude and imposed displacements accelerate cyclic degradation rates. A new interactive shaft stability chart was produced as a summary of axial cyclic loading tests for both jacked and driven piles in medium dense silica sand. Laboratory tests confirm findings from field tests where one-way low amplitude cycles lead to beneficial increases in tensile pile capacity of up to 20%. Complex mechanisms as grain breakage and local density changes at the pile's interface. The kinematics behind the installation of piles and its interaction with the surrounding soil is still limited. Understanding how granular material interacts with the pile may reveal important to understand the global pile response. The global observations of the pile behaviour from calibration chamber tests were modelled at a micro scale using Micro Computed Tomography at the Grenoble Laboratory 3SR. The experimental campaign included tests on a model calibration chamber devoted to the displacement field analyses during the installation of a model piles using three dimensional (3D) digital image correlation (DIC). Micromechanical analysis of « intact » post-mortem samples recovered at the pille's interface were also conducted for evidences of radial density gradient and grain breakage.

The objective of the present work is to determinate the pressure drop and the heat transfer coefficients, normally applied to analytical calculations of hydrogenerators thermal design, obtained by applying numerical calculation (Computational Fluid Dynamics - CFD) and validated by experimental results and field measurements. The object of study is limited to the most important region of the ventilation system (the cooling air ducts of stator core) to get numerical results of heat transfer and pressure drop coefficients, which are impacted mostly by the entrance of air ducts. The numerical calculations considered three-dimensional, steady-state, incompressible and turbulent flow; and were based on the Finite Volume methodology. The turbulent flow computations were carried out with procedures based on RANS equations by selecting k-omega SST (Shear-Stress Transport) as turbulence model. Grid quality metrics were monitored and the uncertainties due to discretization errors were evaluated by means of a grid independence study and application of an uncertainty estimation procedure based on Richardson extrapolation. The validation of numerical method developed by the present work (specifically to simulate the flow dynamics behavior and to obtain numerically the pressure drop coefficient of the airflow to enter and pass through the Stator Core Air Duct in a hydrogenerator) is performed by comparing the numerical results to experimental data published by Wustmann (2005). The reference experimental data were obtained by a model test. The comparison between numerical and experimental results shows that the difference of pressure drop for Reynolds numbers higher than 5000 is 2% at maximum, while for lower Reynolds numbers, the difference increases significantly and reaches 10%. It is presented that the most reasonable hypothesis for higher discrepancy at lower Reynolds numbers can be assigned to the experiment\'s non-steady-state condition. It is to conclude that the proposed numerical method is validated for the upper region of the analyzed range. Additionally to the model test validation, field measurements were executed in order to confirm numerical results. Measurements of pressure drop in the stator core of a real hydrogenerator were a challenge. Nevertheless, despite all the difficulties and considerable high field measuring uncertainties, trend curves behavior are similar to numerical results. Finally, series of numerical calculation, varying geometrical parameters of the air-duct inlet design and operational data, were done in order to obtain pressure drop coefficients trend curves to be directly applied to analytical calculation routines of whole hydrogenerator ventilation systems. Parallel to it, thermal numerical calculation was executed in the prototype simulation in order to define the convective heat transfer coefficient. / O objetivo do presente trabalho é determinar os coeficientes de perda de carga e transferência de calor, normalmente aplicados nos cálculos analíticos de design térmico de hidrogeradores, obtido pela aplicação de cálculo numérico (Computacional Fluid Dynamics - CFD) e validado por resultados experimentais e medições de campo. O objeto de estudo é limitado à região mais importante do sistema de ventilação (os dutos de ar de arrefecimento do núcleo do estator) para obter resultados numéricos dos coeficientes de transferência de calor e de perda de carga, que são impactados principalmente pela entrada de dutos de ar. Os cálculos numéricos consideraram escoamentos tridimensionais, em regime permanente, incompressíveis e turbulentos; e foram baseados no método dos volumes finitos. Os cálculos de escoamento turbulento foram realizados com procedimentos baseados em equações médias (RANS), utilizando o modelo k-omega SST (Shear-Stress Transport) como modelo de turbulência. Métricas de qualidade de malha foram monitoradas e as incertezas devido à erros de discretização foram avaliadas por meio de um estudo de independência de malha e aplicação de um procedimento de estimativa de incertezas com base na extrapolação de Richardson. A validação do método numérico desenvolvido pelo presente trabalho (especificamente para simular o comportamento dinâmico do escoamento e obter numericamente o coeficiente de perda de carga do escoamento ao entrar no duto de ar e atravessar o núcleo do estator de um hidrogerador) é realizada comparando os resultados numéricos com dados experimentais publicados por Wustmann (2005). Os dados experimentais foram obtidos como referência por um teste de modelo. A comparação entre os resultados numéricos e experimentais mostra que a diferença da perda de carga para números de Reynolds mais elevados do que 5000 é no máximo de 2%, enquanto que para números de Reynolds inferiores, a diferença aumenta significativamente e atinge 10%. A hipótese mais razoável para a maior discrepância para número de Reynolds menores é a possível influência de instabilidades do escoamento no experimento, fazendo com que o regime seja não-permanente. Conclui-se que o método numérico proposto é validado para a região superior do intervalo analisado. Além da validação pelo ensaio de modelo, medições de campo foram executadas, a fim de confirmar os resultados numéricos. As medições de perda de carga no núcleo do estator de um hidrogerador real era um desafio. No entanto, apesar de todas as dificuldades e consideráveis incertezas da medição campo, o comportamento das curvas de tendência ficou alinhado com resultados numéricos. Finalmente, uma série de cálculos numéricos, variando parâmetros geométricos do design da entrada do duto de ar e dados operacionais, foram executados a fim de se obter curvas de tendência para coeficientes de perda de carga (resultados deste trabalho) a serem aplicadas diretamente à rotinas de cálculos analíticos de sistemas completos de ventilação de hidrogeradores. Paralelamente à isso, o cálculo térmico numérico foi executado na simulação do protótipo, a fim de se definir o coeficiente de transferência de calor por convecção.

The velocity field in the wake of a small scale flexible parachute canopy was measured using two-dimensional particle image velocimetry. The experiments were performed in a water tunnel with the Reynolds number ranging from 3.0-6.0 x 104. Both a fully inflated canopy and the inflation phase were investigated in a constant freestream (i.e. an infinite mass condition). The fully inflated canopy experienced a cyclic“breathing" which corresponded to the shedding of a vortex ring from the canopy. The normalized breathing frequency had a value of 0.56 +/- 0.03. The investigation of the canopy inflation showed that during the early stages of the inflation, the boundary layer on the canopy surface remains attached to the canopy while the canopy diameter increases substantially. The boundary layer begins to separate near the apex region when the diameter is ~68% of the fully inflated diameter. The separation point then progresses upstream from the canopy apex region toward the canopy skirt. During this time period, the force rapidly increases to its maximum value while the separation point of the boundary layer moves upstream towards the skirt. The force then declines rapidly and the separated boundary layer rolls-up into a large vortex ring near the canopy skirt. At the same time, the canopy is drawn into an over-expanded state after which the cyclic breathing initiates. The unsteady potential force was estimated from the rate of change of the canopy volume. It contributed no more than 10% of the peak opening force and was only significant during the early stages of inflation. The majority of the opening force was the result of the time rate of change of the fluid impulse. It accounts for approximately 60% of the peak opening force. This result shows that the formation of the viscous wake is the primary factor in the peak drag force of the canopy.

Les mesures optiques dimensionnelles sont des techniques en plein essor dont la maîtrise et l'exploitation soulèvent encore de nombreuses questions. Pour une meilleure compréhension d'un système de mesure par stéréo-corrélation d'images, des études de caractérisation et d'évaluation de l'erreur de mesure en corrélation à partir d'images synthétiques et en stéréovision à partir d'images réelles ont été réalisées. Les résultats mettent en avant l'influence de plusieurs paramètres : fenêtre de corrélation, déformation et gradient de déformation. La dernière partie de l'étude est consacrée à l'identification de comportements constitutifs à partir des mesures de champs. Deux méthodes sont mises en œuvre : la méthode des champs virtuel pour l'identification des paramètres élastiques d'un matériau et l'identification paramétrique par recalage éléments finis pour le cas des comportements élasto-plastiques. / The optical dimensional measurements are emergent techniques whose control and exploitation still address many questions. For a better understanding of the stereo-correlation measurement systems, studies are made to characterize and assess the digital image correlation measurement error by the way of synthetic images. The stereovision characterization is made using real images. The results highlight the influence of several parameters : subset size, strain and strain gradient. The last part of this study is devoted to the identification of constitutive behaviour law using full-field measurements. Two identification methods are used : the virtual fields method to determine the elastic parameters of a material and the finite element model updating method to identify the elasto-plastic behaviour law.

The objective of the present work is to determinate the pressure drop and the heat transfer coefficients, normally applied to analytical calculations of hydrogenerators thermal design, obtained by applying numerical calculation (Computational Fluid Dynamics - CFD) and validated by experimental results and field measurements. The object of study is limited to the most important region of the ventilation system (the cooling air ducts of stator core) to get numerical results of heat transfer and pressure drop coefficients, which are impacted mostly by the entrance of air ducts. The numerical calculations considered three-dimensional, steady-state, incompressible and turbulent flow; and were based on the Finite Volume methodology. The turbulent flow computations were carried out with procedures based on RANS equations by selecting k-omega SST (Shear-Stress Transport) as turbulence model. Grid quality metrics were monitored and the uncertainties due to discretization errors were evaluated by means of a grid independence study and application of an uncertainty estimation procedure based on Richardson extrapolation. The validation of numerical method developed by the present work (specifically to simulate the flow dynamics behavior and to obtain numerically the pressure drop coefficient of the airflow to enter and pass through the Stator Core Air Duct in a hydrogenerator) is performed by comparing the numerical results to experimental data published by Wustmann (2005). The reference experimental data were obtained by a model test. The comparison between numerical and experimental results shows that the difference of pressure drop for Reynolds numbers higher than 5000 is 2% at maximum, while for lower Reynolds numbers, the difference increases significantly and reaches 10%. It is presented that the most reasonable hypothesis for higher discrepancy at lower Reynolds numbers can be assigned to the experiment\'s non-steady-state condition. It is to conclude that the proposed numerical method is validated for the upper region of the analyzed range. Additionally to the model test validation, field measurements were executed in order to confirm numerical results. Measurements of pressure drop in the stator core of a real hydrogenerator were a challenge. Nevertheless, despite all the difficulties and considerable high field measuring uncertainties, trend curves behavior are similar to numerical results. Finally, series of numerical calculation, varying geometrical parameters of the air-duct inlet design and operational data, were done in order to obtain pressure drop coefficients trend curves to be directly applied to analytical calculation routines of whole hydrogenerator ventilation systems. Parallel to it, thermal numerical calculation was executed in the prototype simulation in order to define the convective heat transfer coefficient. / O objetivo do presente trabalho é determinar os coeficientes de perda de carga e transferência de calor, normalmente aplicados nos cálculos analíticos de design térmico de hidrogeradores, obtido pela aplicação de cálculo numérico (Computacional Fluid Dynamics - CFD) e validado por resultados experimentais e medições de campo. O objeto de estudo é limitado à região mais importante do sistema de ventilação (os dutos de ar de arrefecimento do núcleo do estator) para obter resultados numéricos dos coeficientes de transferência de calor e de perda de carga, que são impactados principalmente pela entrada de dutos de ar. Os cálculos numéricos consideraram escoamentos tridimensionais, em regime permanente, incompressíveis e turbulentos; e foram baseados no método dos volumes finitos. Os cálculos de escoamento turbulento foram realizados com procedimentos baseados em equações médias (RANS), utilizando o modelo k-omega SST (Shear-Stress Transport) como modelo de turbulência. Métricas de qualidade de malha foram monitoradas e as incertezas devido à erros de discretização foram avaliadas por meio de um estudo de independência de malha e aplicação de um procedimento de estimativa de incertezas com base na extrapolação de Richardson. A validação do método numérico desenvolvido pelo presente trabalho (especificamente para simular o comportamento dinâmico do escoamento e obter numericamente o coeficiente de perda de carga do escoamento ao entrar no duto de ar e atravessar o núcleo do estator de um hidrogerador) é realizada comparando os resultados numéricos com dados experimentais publicados por Wustmann (2005). Os dados experimentais foram obtidos como referência por um teste de modelo. A comparação entre os resultados numéricos e experimentais mostra que a diferença da perda de carga para números de Reynolds mais elevados do que 5000 é no máximo de 2%, enquanto que para números de Reynolds inferiores, a diferença aumenta significativamente e atinge 10%. A hipótese mais razoável para a maior discrepância para número de Reynolds menores é a possível influência de instabilidades do escoamento no experimento, fazendo com que o regime seja não-permanente. Conclui-se que o método numérico proposto é validado para a região superior do intervalo analisado. Além da validação pelo ensaio de modelo, medições de campo foram executadas, a fim de confirmar os resultados numéricos. As medições de perda de carga no núcleo do estator de um hidrogerador real era um desafio. No entanto, apesar de todas as dificuldades e consideráveis incertezas da medição campo, o comportamento das curvas de tendência ficou alinhado com resultados numéricos. Finalmente, uma série de cálculos numéricos, variando parâmetros geométricos do design da entrada do duto de ar e dados operacionais, foram executados a fim de se obter curvas de tendência para coeficientes de perda de carga (resultados deste trabalho) a serem aplicadas diretamente à rotinas de cálculos analíticos de sistemas completos de ventilação de hidrogeradores. Paralelamente à isso, o cálculo térmico numérico foi executado na simulação do protótipo, a fim de se definir o coeficiente de transferência de calor por convecção.

This thesis investigates a special type of energy system, namely energy use in supermarkets through modelling, simulations and field studies. A user-friendly computer program, CyberMart, which calculates the total energy performance of a supermarket, is presented. The modelling method described in this thesis has four phases: the first phase is the de-velopment of a conceptual model that includes its objectives, the envi-ronment and the components of the system, and their interconnections. The second phase is a quantitative model in which the ideas from the conceptual model are transformed into mathematical and physical rela-tionships. The third phase is an evaluation of the model with a sensitivity analysis of its predictions and comparisons between the computer model and results from field measurements. The fourth phase is the model ap-plication in which the computer model answers questions identified in the beginning of the modelling process as well as other questions arising throughout the work. Field measurements in seven different supermarkets in Sweden were car-ried out to: (i) investigate the most important parameters that influence energy performance in supermarkets, (ii) analyse the operation of new system designs with indirect system implementation in Sweden during recent years, and (iii) validate the computer model. A thorough sensitivity analysis shows a total sensitivity of 5.6 %, which is a satisfactory result given a 10% change in the majority of input parame-ters and assumptions, with the exception of outdoor temperatures and solar radiation that were calculated as extreme values in METEO-NORM. Comparisons between measurements and simulations in five supermarkets also show a good agreement. Measurements and simula-tion results for a whole year were not possible due to lack of data. CyberMart opens up perspectives for designers and engineers in the field by providing innovative opportunities for assessment and testing of new energy efficient measures but also for evaluation of different already-installed system designs and components. The implementation of new energy-saving technologies in supermarkets requires an extensive inte-grated analysis of the energy performances of the refrigeration system, HVAC system, lighting, equipment, and the total energy usage. This analysis should be done over a long period, to evaluate and compare the real energy performance with the theoretical values calculated by Cyber-Mart. / QC 20100330

Ce travail porte sur l'endommagement des matériaux composites stratifiés utilisés notamment pour la réalisation de pièces structurales minces. La dégradation de ces matériaux induite par la création et le développement de surfaces de décohésion internes est abordée sous deux angles. Une campagne expérimentale a tout d'abord été menée sur des stratifiés en carbone-époxy réalisés par infusion de résine liquide et sollicités en traction uniaxiale. Cette étude propose une analyse originale à l'aide de trois techniques optiques permettant une caractérisation de l'endommagement par mesures de champs : cinématiques (par stéréo-corrélation d'images), thermiques (par thermographie infrarouge) et densimétriques (par tomographie à rayons X). Le second volet du travail concerne la modélisation de la microfissuration dans le contexte d'une anisotropie initiale. A cette fin, une homogénéisation bidimensionnelle de milieux orthotropes fissurés permet la prise en compte de défauts d'orientation arbitraire et des effets unilatéraux (ouverture-fermeture des microfissures) au sein d'une formulation énergétique en déformation. Sur cette base, un modèle de comportement est proposé dans le cadre de la thermodynamique des processus irréversibles avec variables internes. Des simulations numériques permettent de démontrer les capacités prédictives de la formulation, en particulier la représentation du comportement non linéaire de ces matériaux, l'interaction entre les anisotropies initiale et induite et la restitution des propriétés élastiques lors de la fermeture de défauts. / This work deals with the damage of laminated composite materials used in particular for the production of thin structural parts. The degradation of these materials induced by the creation and growth of internal microcracks is considered from two angles. An experimental campaign was first conducted on carbon-epoxy laminates made by liquid resin infusion and loaded in uniaxial tension. This study proposes an original analysis using three advanced optical techniques that allow the damage characterization through full-field measurements : kinematic (with stereo-image correlation), thermal (with infrared thermography) and density (with X-ray tomography). The second part of the work concerns the modeling of microcracking in the context of initial anisotropy. To this end, a two-dimensional homogenization of orthotropic cracked media allows consideration of arbitrary orientation of defects and unilateral effects (opening and closing of microcracks) within a strain energetic formulation. On this basis, a constitutive model is proposed in the framework of thermodynamics of irreversible processes with internal variables. Numerical simulations demonstrate the predictive capabilities of the formulation, in particular the representation of the nonlinear behavior of these materials, the interaction between initial and induced anisotropies and the recovery of elastic properties at the closure of microdefects.