L’utilisation des matériaux constitués de fibres synthétique ou naturelle est en pleine expansion et concerne de nombreux secteurs : industrie automobile, aéronautique, électrique, filtration de l’air ou applications médicales. Malgré des procédés de fabrication et des natures de fibres différents, ces matériaux ont pour point commun d’être constitués d’un réseau de fibres liées entre elles par des liaisons. Les papiers et les cartons sont, par exemple, constitués de fibres de cellulose naturelles liées chimiquement. A l’heure actuelle, les mécanismes de fissuration dans de tels milieux sont encore mal compris. Ils dépendent fortement (a) des propriétés des constituants : géométrie et propriétés mécaniques des fibres et des contacts fibre-fibre, (b) des caractéristiques des réseaux fibreux : géométrie et arrangement des fibres, et des caractéristiques du réseau poreux induit : porosité, distribution de taille des pores, répartition spatiale des pores, etc. et (c) des modes de sollicitations mécaniques. Dans ce type de matériaux, les effets d’échelles doivent être pris en compte pour compléter les approches mécaniques traditionnelles. Les récents progrès en mécanique expérimentale et en simulation numérique permettent de mener une telle étude de l’échelle de la fibre à celle du réseau fibreux.Cette thèse a donc pour but de mettre en place des outils d’analyse des microstructures et des mécanismes de fissuration dans les milieux fibreux à faible densité. Pour cela, (i) des essais de micromécaniques seront couplés à des méthodes d’imagerie (ESEM, microtomographie à rayons X, stéréocorrélation) afin de caractériser expérimentalement les milieux et leur endommagement (ii) Cette étude vient compléter les travaux expérimentaux menés dans les deux laboratoires 3SR et LGP2 (ANR ANAFIB http://anafib.hmg.inpg.fr/spip.php?rubrique1) et sera complétée par des simulations numériques des essais réalisés en collaboration avec Per Isaksson de l’Université d’Uppsala (Suède). / Materials made up of synthetic or natural fibres are increasingly developed in various domains: papermaking, composite, automotive and aeronautic industries for structural, packaging, air filtration or medical applications. Despite the variety of manufacturing processes of such materials, all of them can be considered as being formed by a network of fibres interconnected via bonds. For instance, in the case of materials made up of natural cellulosic fibres such as papers or boards, fibres are chemically linked.Crack propagation phenomena in such materials remain poorly understood even though it can be presumed that such mechanisms depend on:- (i) the geometrical and mechanical properties of the constituents of individual fibres and fibre-fibre bonds,- (ii) the architecture of the fibrous network, for example the spatial distributions of fibres, bonds and pores and the size distributions of pores and bonds,- (iii) the applied mechanical loadings.In such materials, scale effects must be investigated in order to improve the classical approaches used to understand crack propagation mechanisms. Recent progresses in both experimental mechanics and numerical simulation approaches allow such a study from the fibre scale up to the fibre network scale to be carried out.The proposed PhD aims first at developing an original experimental approach to analyse microstructure changes and crack propagation phenomena for low density papers. For that purpose x-ray microtomography or ESEM, and stereo-correlation experiments will be carried out to investigate microstructural changes and deformation mechanisms at all relevant scales (see the illustration given in ).

The primary objective of this research is to develop a mathematical framework that could be used to model or predict the rate of fluid absorption and release in fibrous sheets made up of solid or porous fibers. In the first step, a two-scale two-phase modeling methodology is developed for studying fluid release from saturated/unsaturated thin fibrous media made up of solid fibers when brought in contact with a moving solid surface. Our macroscale model is based on the Richards’ equation for two-phase fluid transport in porous media. The required constitutive relationships, capillary pressure and relative permeability as functions of the medium’s saturation, are obtained through microscale modeling. Here, a mass convection boundary condition is considered to model the fluid transport at the boundary in contact with the target surface. The mass convection coefficient plays a significant role in determining the release rate of fluid. Moreover the release rate depends on the properties of the fluid, fibrous sheet, the target surface as well as the speed of the relative motion, and remains to be determined experimentally. Obtaining functional relationships for relative permeability and capillary pressure is only possible through experimentation or expensive microscale simulations, and needs to be repeated for different media having different fiber diameters, thicknesses, or porosities. In this concern, we conducted series of 3-D microscale simulations in order to investigate the effect of the aforementioned parameters on the relative permeability and capillary pressure of fibrous porous sheets. The results of our parameter study are utilized to develop general expressions for kr(S) and Pc(S). Furthermore, these general expressions can be easily included in macroscale fluid transport equations to predict the rate of fluid release from partially saturated fibrous sheets in a time and cost-effective manner. Moreover, the ability of the model has been extended to simulate the radial spreading of liquids in thin fibrous sheets. By simulating different fibrous sheets with identical parameters but different in-plane fiber orientations has revealed that the rate of fluid spread increases with increasing the in-plane alignment of the fibers. Additionally, we have developed a semi-analytical modeling approach that can be used to predict the fluid absorption and release characteristics of multi-layered composite fabric made up of porous (swelling) and soild (non-swelling) fibrous sheets. The sheets capillary pressure and relative permeability are obtained via a combination of numerical simulations and experiment. In particular, the capillary pressure for swelling media is obtained via height rise experiments. The relative permeability expressions are obtained from the analytical expressions previously developed with the 3-D microscale simulations, which are also in agreement with experimental correlations from the literature. To extend the ability of the model, we have developed a diffusion-controlled boundary treatment to simulate fluid release from partially-saturated fabrics onto surfaces with different hydrophilicy. Using a custom made test rig, experimental data is obtained for the release of liquid from partially saturated PET and Rayon nonwoven sheets at different speeds, and on two different surfaces. It is demonstrated that the new semi-empirical model redeveloped in this work can predict the rate of fluid release from wet nonwoven sheets as a function of time.

Capillarity is often exploited in self-cleaning, drag reducing and fluid absorption/storage (sanitary products) purposes just to name a few. Formulating the underlying physics of capillarity helps future design and development of optimized structures. This work reports on developing computational models to quantify the capillary pressure and capillary forces on the fibrous surfaces. To this end, the current study utilizes a novel mass-spring-damper approach to incorporate the mechanical properties of the fibers in generating virtual fibrous structures that can best represent fibrous membranes. Such virtual fibrous structures are then subjected to a pressure estimation model, developed for the first time in this work, to estimate the liquid entry pressure (LEP) for a hydrophobic fibrous membrane. As for accurate prediction (and not just estimation) of the capillary pressure, this work also presents an energy minimization method, implemented in the Surface Evolver code, for tracking the air–water interface intrusion in a hydrophobic fibrous membrane comprised of orthogonally oriented fibers. This novel interface tracking algorithm is used to investigate the effects of the membrane’s microstructure and wetting properties on its resistance to water intrusion (i.e., LEP). The simulation method developed in this work is computationally affordable and it is accurate in its predictions of the air–water interface shape and position inside the membrane as a function of pressure. Application of the simulation method in studying effects of fiber diameter or contact angle heterogeneity on water intrusion pressure is reported for demonstration purposes. Capillary forces between fibrous surfaces are also studied experimentally and numerically via the liquid bridge between two parallel plates coated with electrospun fibers. In the experiment, a droplet was placed on one of the polystyrene- or polyurethane-coated plates and then compressed, stretched, or sheared using the other plate and the force was measured using a sensitive scale. In the simulation, the liquid bridge was mathematically defined for the Surface Evolver finite element code to predict its 3-D shape and resistance to normal and shearing forces, respectively, in presence of the contact angle hysteresis effect. Despite the inherent non-uniformity of the fibrous surfaces used in the experiments and the simplifying assumptions considered for the simulations, reasonable agreement was observed between the experiments and simulations. Results reveal that both normal and shear force on the plates increase by increasing the liquid volume, or decreasing the spacing between the plates.

No description available.

Este trabalho visa estudar o fluxo de um fluido em um meio poroso (fibroso) durante o processo de Moldagem por Transferência de Resina (RTM) utilizando e comparando dois softwares comerciais: o ANSYS CFX, software de simulação numérica CFD não-dedicado a RTM, e o PAM-RTM, software de simulação numérica dedicado a esta aplicação. Foram utilizados dados experimentais, empregando uma pré-forma de tecido (0/90) de fibra de vidro e que foi impregnada por um óleo vegetal em injeção radial de RTM. Diversos experimentos foram realizados variando-se a pressão de injeção e o teor volumétrico de fibras a fim de observar o efeito no comportamento do fluxo. Com estes dados, alimentou-se o ANSYS CFX e o PAM-RTM utilizando equações dos meios porosos conhecidas, e foram feitas análises numéricas de todos os casos experimentais. Resultados de permeabilidade, viscosidade do fluido, tempo de preenchimento do molde, campo de pressão, variação da fração volumétrica das fases presentes e vetor velocidade de fluxo são apresentados e analisados. Concluiu-se que há uma sólida relação entre o experimental e o simulado (erros inferiores a 10%), o software PAM-RTM possui menor erro do que o ANSYS CFX quando comparados aos resultados experimentais, mas este último apresenta maior versatilidade de análises. / This work aims to study the flow of fluid through a porous (fibrous) media that occurs during Resin Transfer Molding (RTM) of composite materials and to compare it with the numerical results obtained with two commercial simulation software, ANSYS CFX, a general-use CFD package, and PAM-RTM, which is dedicated to RTM modeling. Experimental data were produced using a glass-fiber cloth (0/90) perform, which was impregnated with a vegetable oil in an RTM radial infiltration. Several experiments were performed varying injection pressure and fiber volume content in order to observe their effect on the flow behavior. These data were input into ANSYS CFX and PAM-RTM using known porous media equations and they were used to model all experiments. Results of permeability, fluid viscosity, mold filling time, pressure field, variation of volume fraction of the phases and the flow velocity vector are presented and analyzed. An excellent correlation between experimental and simulated results was found (errors less than 10%), and PAM-RTM yielded smaller error than ANSYS CFX, but the latter allows greater flexibility.

Este trabalho visa estudar o fluxo de um fluido em um meio poroso (fibroso) durante o processo de Moldagem por Transferência de Resina (RTM) utilizando e comparando dois softwares comerciais: o ANSYS CFX, software de simulação numérica CFD não-dedicado a RTM, e o PAM-RTM, software de simulação numérica dedicado a esta aplicação. Foram utilizados dados experimentais, empregando uma pré-forma de tecido (0/90) de fibra de vidro e que foi impregnada por um óleo vegetal em injeção radial de RTM. Diversos experimentos foram realizados variando-se a pressão de injeção e o teor volumétrico de fibras a fim de observar o efeito no comportamento do fluxo. Com estes dados, alimentou-se o ANSYS CFX e o PAM-RTM utilizando equações dos meios porosos conhecidas, e foram feitas análises numéricas de todos os casos experimentais. Resultados de permeabilidade, viscosidade do fluido, tempo de preenchimento do molde, campo de pressão, variação da fração volumétrica das fases presentes e vetor velocidade de fluxo são apresentados e analisados. Concluiu-se que há uma sólida relação entre o experimental e o simulado (erros inferiores a 10%), o software PAM-RTM possui menor erro do que o ANSYS CFX quando comparados aos resultados experimentais, mas este último apresenta maior versatilidade de análises. / This work aims to study the flow of fluid through a porous (fibrous) media that occurs during Resin Transfer Molding (RTM) of composite materials and to compare it with the numerical results obtained with two commercial simulation software, ANSYS CFX, a general-use CFD package, and PAM-RTM, which is dedicated to RTM modeling. Experimental data were produced using a glass-fiber cloth (0/90) perform, which was impregnated with a vegetable oil in an RTM radial infiltration. Several experiments were performed varying injection pressure and fiber volume content in order to observe their effect on the flow behavior. These data were input into ANSYS CFX and PAM-RTM using known porous media equations and they were used to model all experiments. Results of permeability, fluid viscosity, mold filling time, pressure field, variation of volume fraction of the phases and the flow velocity vector are presented and analyzed. An excellent correlation between experimental and simulated results was found (errors less than 10%), and PAM-RTM yielded smaller error than ANSYS CFX, but the latter allows greater flexibility.

Este trabalho visa estudar o fluxo de um fluido em um meio poroso (fibroso) durante o processo de Moldagem por Transferência de Resina (RTM) utilizando e comparando dois softwares comerciais: o ANSYS CFX, software de simulação numérica CFD não-dedicado a RTM, e o PAM-RTM, software de simulação numérica dedicado a esta aplicação. Foram utilizados dados experimentais, empregando uma pré-forma de tecido (0/90) de fibra de vidro e que foi impregnada por um óleo vegetal em injeção radial de RTM. Diversos experimentos foram realizados variando-se a pressão de injeção e o teor volumétrico de fibras a fim de observar o efeito no comportamento do fluxo. Com estes dados, alimentou-se o ANSYS CFX e o PAM-RTM utilizando equações dos meios porosos conhecidas, e foram feitas análises numéricas de todos os casos experimentais. Resultados de permeabilidade, viscosidade do fluido, tempo de preenchimento do molde, campo de pressão, variação da fração volumétrica das fases presentes e vetor velocidade de fluxo são apresentados e analisados. Concluiu-se que há uma sólida relação entre o experimental e o simulado (erros inferiores a 10%), o software PAM-RTM possui menor erro do que o ANSYS CFX quando comparados aos resultados experimentais, mas este último apresenta maior versatilidade de análises. / This work aims to study the flow of fluid through a porous (fibrous) media that occurs during Resin Transfer Molding (RTM) of composite materials and to compare it with the numerical results obtained with two commercial simulation software, ANSYS CFX, a general-use CFD package, and PAM-RTM, which is dedicated to RTM modeling. Experimental data were produced using a glass-fiber cloth (0/90) perform, which was impregnated with a vegetable oil in an RTM radial infiltration. Several experiments were performed varying injection pressure and fiber volume content in order to observe their effect on the flow behavior. These data were input into ANSYS CFX and PAM-RTM using known porous media equations and they were used to model all experiments. Results of permeability, fluid viscosity, mold filling time, pressure field, variation of volume fraction of the phases and the flow velocity vector are presented and analyzed. An excellent correlation between experimental and simulated results was found (errors less than 10%), and PAM-RTM yielded smaller error than ANSYS CFX, but the latter allows greater flexibility.

Dans ce travail, les propriétés thermiques effectives du milieu fibreux sont déterminées en tenant compte du couplage conduction et rayonnement. Un échantillon numérique fibreux statistiquement homogène composé de deux phases a été généré en empilant des cylindres finis absorbant dans le vide. Ces cylindres sont dispersés selon des fonctions de distribution de la position de leur centre et de leur orientation. L'interpénétration des cylindres est permis. L'extinction, l'absorption et la diffusion sont caractérisées par des fonctions statistiques radiatives qui permettent de savoir si le milieu est Beerien (ou non). Elles sont déterminées précisément à l'aide d'une méthode de Monte Carlo. On montre que la phase gazeuse a un comportement Beerien et que le phase fibreuse a un comportement fortement non Beerien. Le champ de puissance radiative déposée dans le milieu fibreux est calculé en résolvant un modèle qui couple une Équation du Transfert Radiatif Généralisée (ETRG) et une Équation du Transfert radiatif Classique (ETR). Le modèle de conduction thermique est basé sur une méthode de marche aléatoire ne nécessitant aucun maillage. La simulation du mouvement Brownien de marcheurs dans les fibres permet de résoudre l'équation de l'énergie. L'idée de la méthode est de caractériser la température d'un volume élémentaire par une densité de marcheurs, qui peuvent parcourir le milieu. Le problème est gouverné par les conditions aux limites ; Une concentration constante de marcheurs (ou un flux constant) est associée à une température imposée (ou un flux). / In the present work, the effective heat transfer properties of fibrous medium are determined by taking into account a coupling of heat conduction and radiation. A virtual, statistically homogeneous, two-phase fibrous sample has been built by stacking finite absorbing cylinders in vaccum. These cylinders are dispersed according to prescribed distribution functions defining the cylinder positions and orientations. Cylinder overlappings are allowed. Extinction, absorption and scattering are characterised by radiative statistical functions which allow the Beerian behaviour of a medium to be assessed (or not). They are accurately determined with a Monte Carlo method. Whereas the gaseous phase exhibits a Beerian behaviour, the fibre phase is strongly non Beerian. The radiative power field deposited within the fibrous material is calculated by resolving a model which couples a Generalized Radiative Transfer Equation (GRTE) and a classic Radiative Transfer Equation (RTE). The model of conduction transfer is based on a random walk method without meshing. The simulation of Brownian motion of walkers in fibres allows the energy equation to be solved. The idea of the method is to characterize the temperature in an elementary volume by the density of walkers, which roam the medium. The problem is governed by boundary conditions ; A constant concentration of walkers (or a constant flux) is associated with a fixed temperature (or flux).