Interparticle interactions determine the microstructure, stability, rheology, and optical properties of concentrated colloidal suspensions involved in paint, paper, cosmetic, and pharmaceutical industries, etc. Frequency domain photon migration (FDPM) involves modeling the photon transport in a multiple scattering medium as a diffusion process in order to simultaneously determine isotropic scattering and absorption coefficients from measured amplitude attenuation and phase shift of the propagating photon density wave. Using FDPM, we investigated the impact of electrostatic interaction upon the optical properties and structure of dense charged suspensions. We demonstrated that electrostatic interactions among charged polystyrene latex may significantly affect the light scattering properties and structure of dense suspensions at low ionic strength (<0.06 mM NaCl equivalent) by actual FDPM measurement. We showed that the structure factor models addressing electrostatic interaction can be used to describe the microstructure of charged suspensions and quenched scattering due to electrostatics, and demonstrated that FDPM has the potential to be a novel structure and surface charge probe for dense suspensions. We also showed that the FDPM measured isotropic scattering coefficients may respond to the change in effective particle surface charge, and displayed the potential of using FDPM for probing particle surface charge in concentrated suspensions. We presented that the interference approximation implies a linear relationship between the absorption coefficient and volume fraction of suspension. We illustrated that FDPM measured absorption coefficient varies linearly with suspension volume fraction and affirmed the interference approximation from a perspective of light absorption. The validation of the interference approximation enables us to develop the methodology for estimating absorption efficiencies and imaginary refractive indices for both particles and suspending fluid simultaneously using FDPM. We further demonstrated a novel application of FDPM measured absorption coefficients in determining pigment absorption spectra, and displayed the potential of using FDPM as a novel analytical tool in pigment and paint industry.

Parmi les centrales solaires thermiques à concentration, la technologie des centrales à tour offre l'un des rendements les plus importants de production d'énergie. Néanmoins, l'efficacité et la sécurité de ces centrales sont améliorables. En effet, les sels fondus, généralement utilisés comme fluide de transfert thermique, présentent une plage limitée d'utilisation (200-550°C), à l'origine des limites d'efficacité de la conversion thermique-électrique, ainsi que de consommations parasites d'énergie de chauffage. De plus, leurs caractères corrosif et comburant sont à l'origine de sévères contraintes de sécurité. Un nouveau concept de récepteur solaire, dont les caractéristiques permettent de s'affranchir des contraintes associées aux sels fondus, est présenté dans ce manuscrit. Il utilise des suspensions denses de particules fluidisées par un gaz comme fluide de transfert et de stockage de l'énergie thermique. Ce concept, et la technologie de récepteur associée, a été brevetée par Flamant et Hemati dans le cadre d'une collaboration entre le Laboratoire CNRS-PROMES d'Odeillo, et l'Institut National Polytechnique de Toulouse. Son développement a reçu le soutien financier du CNRS, puis de la Commission Européenne. Les propriétés thermiques du carbure de silicium ont déterminé le choix de ce solide. Le diamètre moyen des particules utilisées avoisine 60 micromètres (groupe A). Ces particules présentent d'excellentes propriétés de fluidisation pour des vitesses de gaz faibles. La construction et l'exploitation d'une maquette froide transparente ont permis de démontrer la faisabilité hydrodynamique du concept. Cette maquette est un échangeur à deux passes. Chaque passe est constituée de deux tubes verticaux en parallèle. L'une est traversée par un débit vertical ascendant de solide, l'autre descendant. Un débit de solide continu, stable et équitablement réparti a été obtenu à l'intérieur des tubes. La caractérisation hydrodynamique détaillée de l'écoulement, et du comportement globale de la maquette, en fonction des conditions opératoires, a été effectué sur la partie ascendante de l'écoulement dans l'échangeur. La construction et l'exploitation d'une maquette chaude, constituée d'un seul tube traversé par une suspension dense en écoulement ascendant, chauffé par 3 fours d'une puissance totale de 5,6 kW, a permis d'estimer la capacité de transfert thermique de ce nouveau type d'échangeur. Le contrôle et la stabilité des conditions opératoires a permis d'évaluer l'effet de ces dernières sur le transfert thermique entre l'échangeur et la suspension dense de fines particules le traversant. La modélisation par 3 approches du transport ascendant de la suspension dense a également été réalisée. Une approche corrélative 1D basée sur le formalisme du modèle Bulle-Emulsion, adapté afin de tenir compte de l'entraînement des particules dans le sillage des bulles. Ce modèle permet de représenter la structure diphasique de l'écoulement. Une autre approche 1D a été utilisée. Elle repose sur la résolution des équations locales de conservation de masse et de quantité de mouvement sur chaque phase gaz et solide. Cette méthode permet de s'affranchir des hypothèses du modèle Bulle-Emulsion. Enfin, la simulation numérique 3D a été réalisée sur un maillage complet du système, de telle sorte que les conditions aux bornes imposées son identiques à celle imposée par l'opérateur (débit de fluidisation, débit d'aération, débit de solide, pression de la nourrice). Cette dernière apporte des informations sur la structure locale de l'écoulement, dont les caractéristiques permettent d'expliquer l'efficacité du transfert thermique entre la suspension et la paroi observé expérimentalement. / Among concentrating solar power plants, solar tower technology is one of the more power efficient. Nevertheless, their efficiency and safety can be improved. Indeed, molten salts, commonly used as heat transfer fluid, have a limited range of operating temperature (470-820K), thus lowering the thermal-electrical conversion efficiency, and increasing parasitic power consumption. Moreover, they are corrosive and combustion agent, leading to severe safety constraints. A new concept of solar receiver is presented in the present study, the characteristics of which avoid most of the molten salts drawbacks. It uses dense gas-particle suspension as heat transfer and storage fluid. This concept and the associated technology has been patented by Flamant et Hemati in the frame of a collaboration between the PROMES-CNRS Laboratory of Odeillo and the Polytechnic National Institute of Toulouse. Its development has been first supported by the CNRS, and later by the European Commission. Thermal properties of silicon carbide have determined the choice of this solid. The mean diameter of particles is around 64 microns (A group). These particles have excellent fluidisation properties at low gas velocities. The construction and the operation of a transparent cold mockup allowed demonstrating the hydrodynamic feasibility of this concept. This mockup is composed of two passes. Each pass is composed of two tubes in parallel. One pass is upward flow of solid, the other is downward flow. A steady, stable and evenly distributed solid flow has been set inside the tubes. The global behaviour of the system and the hydrodynamics of the suspension has been evaluated as a function of operating parameters on the upward pass. The construction and the operation of a hot mockup allowed estimating the heat transfer efficiency of this new kind of exchanger. On this mockup, the dense suspension flows upward inside a single tube, heated by three ovens of 5.6 kW total power. Thanks to the control and stability of the operating parameters, their effects on the heat transfer between the tube and the dense gas-solid suspension has been accurately determined. Modelling of the suspension upward flow has been performed using 3 approaches. The first one is based on the 1D Bubble-emulsion formalism, adapted to take into account the solid entrainment by the bubble wakes. It allows modelling the diphasic structure of the flow. The resolution of the local mass and momentum balance equation on each phase has also been performed. It allows to sidestep the Bubble-Emsulion assumptions, and to study the effects of drag models. 3D simulation has been performed on a complete mesh of the system, so that the boundary conditions are the same as those imposed by the operator (fluidisation flow rate, aeration flow rate, solid flow rate, dispenser pressure). These simulations give information on the local structure of the suspension flow, influencing on the heat transfer efficiency between the exchanger wall and the suspension.

Nous étudions la rhéologie d'un ensemble de grains sphériques et frottants, par la simulation numérique, à l'échelle des grains, d'écoulements de cisaillement sous une contrainte normale P contrôlée, en présence d'un liquide interstitiel. En faible teneur, ce liquide se présente sous forme de ménisques intergranulaires qui transmettent des forces capillaires attractives ; s'il sature l'espace intergranulaire, on s'intéresse alors à l'écoulement de Stokes de la suspension dense ainsi constituée, où dominent les forces visqueuses. Les assemblages de grains secs constituent un système de référence aux propriétés mécaniques bien connues, en particulier l'approche de l'état critique de la mécanique des sols dans la limite quasi-statique. L'effet des ménisques capillaires qui joignent les grains en présence d'un liquide en faible saturation (régime pendulaire) est étudié pour les taux de cisaillement allant du régime quasi statique au régime inertiel. La rhéologie est caractérisée par le frottement interne apparent, la compacité de l'assemblage, les différences de contraintes normales et diverses variables internes, fonctions de deux paramètres de contrôle adimensionnés : le nombre inertiel I et la pression réduite P*, qui compare les forces de confinement à l'adhésion dans les contacts. Notre étude concerne les états homogènes, ce qui exclut les états de cisaillement localisés observés à faible P*, de l'ordre de 0,1. Le coefficient de frottement interne augmente de 0.35 (cas sec) à 0.9 environ pour P*=0.4, tandis que la compacité décroît de 0.59 à 0.52. L'important effet des forces capillaires sur la rhéologie, sensible pour des P* de plusieurs unités, est relié à la texture anisotrope des contacts et des ponts liquides. Lorsque P* décroît, nombre de contacts cohésifs sont maintenus pour des intervalles de déformation de plusieurs unités, survivant aux effets de rotation et de cisaillement de l'écoulement, et forment des amas percolants dans le système entier. Les résultats sont modérément sensibles à la saturation dans le régime pendulaire, mais fortement affectés par l'hystérèse de la conformation des ménisques. En présence de forces visqueuses et non plus capillaires, une version simplifiée de la dynamique stokésienne est adoptée dans laquelle les forces de lubrification entre proches voisins, supposées dominantes, sont les seules interactions hydrodynamiques. La rhéologie est fortement influencée par les contacts intergranulaires directes, qu'autorise la coupure à courte distance de la singulérité de lubrification du fait de la rugosité de surface des particules. Le même état critique que celui des grains secs est approché dans la limite quasi-statique. Nous discutons de lois rhéologiques exprimées en fonction du nombre visqueux qui remplace alors le nombre inertiel, et de la divergence de la viscosité effective à l'approche de la compacité critique en écoulement permanent, ou de la compacité maximale des assemblages aléatoires pour les configurations isotropes désordonnées / We numerically simulate the shear flow of dense assemblies of 3D frictional spherical grains under a fixed normal stress P in steady-state, either in the presence of a small amount of an interstitial liquid, which gives rise to capillary menisci and attractive forces, or in the fully saturated state, when the mechanical properties of suspensions in Stokes flow are controlled by hydrodynamic and contact forces. Dry grain assemblies are used as a reference system for which the rheological properties - in particular the approach to the critical state – are rather well known and can be measured with good accuracy. A non-saturating wetting fluid creates capillary attractive intergranular forces, the effects of which on the rheology are investigated in the pendular state, with shear rates ranging from quasistatic to inertial regimes. The system behavior is characterized by the dependence of internal friction coefficient, solid fraction, normal stress differences and internal state parameters on two dimensionless control parameters: the inertial number, I and the reduced pressure, P*, comparing confining forces to contact tensile strength. We focus on steady homogeneous flows, excluding localized flow patterns which we observe to occur for low P* (of order 0.1). The apparent internal friction coefficient increases to 0.9 in the quasistatic limit for P*=0.4, from its dry value 0.35, while solid fraction decreases from 0.59 to 0.52. We relate the significant effect of capillary forces on the macroscopic behavior of the system, up to P* values of several unities, to fabric anisotropy parameters of contact and distant interactions. As P* decreases, many cohesive contacts are observed to survive the tumbling motion associated to the shear flow, and their average age exceeds the reciprocal shear rate. Corresponding clusters of grains with enduring capillary bonds gather a large proportion of grains and percolate through the sample. The results are shown to be moderately sensitive to saturation within the pendular range, yet rather strongly affected by the hysteretic nature of liquid bridges. In the presence of viscous forces, assuming lubrication effects to dominate the hydrodynamic interactions, we adopt a simplified version of the (overdamped) Stokesian dynamics approach, in which hydrodynamic interactions only couple close neighbours. Rheological properties are strongly influenced by direct intergranular contacts and friction, which are permitted due to a very small distance lubrication cutoff modeling surface asperities. The same critical state as in the dry case is approached in the quasistatic limit. We discuss expressions of rheological laws involving the viscous number instead of the inertial number, and the divergence of effective viscosities in steady flow and in isotropic random suspensions as either the critical state or the random close packing solid fraction are approached

Modelling solid-fluid multiphase flows are crucial to many applications such as fluidized beds, pyrolysis and gasification, catalytic cracking etc. Accurate modelling of the fluid-particle forces is essential for lab-scale and industry-scale simulations. Fluid-particle system solutions can be obtained using various techniques including the macro-scale TFM (Two fluid model), the meso-scale CFD-DEM (CFD - Discrete Element Method) and the micro-scale PRS (Particle Resolved Simulation method). As the simulation scale decreases, accuracy increases but with an exponential increase in computational time. Since fluid forces have a large impact on the dynamics of the system, this study trains deep learning models using micro-scale PRS data to predict drag forces on ellipsoidal particle suspensions to be applied to meso-scale and macro-scale models. Two different deep learning methodologies are employed, multi-layer perceptrons (MLP) and 3D convolutional neural networks (CNNs). The former trains on the mean characteristics of the suspension including the Reynolds number of the mean flow, the solid fraction of the suspension, particle shape or aspect ratio and inclination to the mean flow direction, while the latter trains on the 3D spatial characterization of the immediate neighborhood of each particle in addition to the data provided to the MLP. The trained models are analyzed and compared on their ability to predict three different drag force values, the suspension mean drag which is the mean drag for all the particles in a given suspension, the mean orientation drag which is the mean drag of all particles at specific orientations to the mean flow, and finally the individual particle drag. Additionally, the trained models are also compared on their ability to test on data sets that are excluded/hidden during the training phase. For instance, the deep learning models are trained on drag force data at only a few values of Reynolds numbers and tested on an unseen value of Reynolds numbers. The ability of the trained models to perform extrapolations over Reynolds number, solid fraction, and particle shape to predict drag forces is presented. The results show that the CNN performs significantly better compared to the MLP in terms of predicting both suspensions mean drag force and also mean orientation drag force, except a particular case of extrapolation where the MLP does better. With regards to predicting drag force on individual particles in the suspension the CNN performs very well when extrapolated to unseen cases and experiments and performs reasonably well when extrapolating to unseen Reynolds numbers and solid fractions. / M.S. / Multiphase solid-fluid flows are ubiquitous in various industries like pharmaceuticals (tablet coating), agriculture (grain drying, grain conveying), mining (oar roasting, mineral conveying), energy (gasification). Accurate and time-efficient computational simulations are crucial in developing and designing systems dealing with multiphase flows. Particle drag force calculations are very important in modeling solid-fluid multiphase flows. Current simulation methods used in the industry such as two-fluid models (TFM) and CFD-Discrete Element Methods (CFD-DEM) suffer from uncertain drag force modeling because these simulations do not resolve the flow field around a particle. Particle Resolved Simulations (PRS) on the other hand completely resolve the fluid flow around a particle and predict very accurate drag force values. This requires a very fine mesh simulation, thus making PRS simulations many orders more computationally expensive compared to the CFD-DEM simulations. This work aims at using deep learning or artificial intelligence-based methods to improve the drag calculation accuracy of the CFD-DEM simulations by learning from the data generated by PRS simulations. Two different deep learning models have been used, the Multi-Layer Perceptrons(MLP) and Convolutional Neural Networks(CNN). The deep learning models are trained to predict the drag forces given a particle's aspect ratio, the solid fraction of the suspension it is present in, and the Reynolds number of the mean flow field in the suspension. Along with the former information the CNN, owing their ability to learn spatial data better is additionally provided with a 3D image of particles' immediate neighborhood. The trained models are analyzed on their ability to predict drag forces at three different fidelities, the suspension mean drag force, the orientation mean drag, and the individual particle drag. Additionally, the trained models are compared on their abilities to predict unseen datasets. For instance, the models would be trained on particles of an aspect ratio of 10 and 5 and tested on their ability to predict drags of particles of aspect ratio 2.5. The results show that the CNN performs significantly better compared to the MLP in terms of predicting both suspension mean drag force and also mean orientation drag force, except a particular case of extrapolation where the MLP does better. With regards to predicting drag force on individual particles in the suspension, the CNN performs very well when extrapolated to unseen cases and experiments and performs reasonably well when extrapolating to unseen Reynolds numbers and solid fractions.

Des solutions d'isolat de protéine de lactosérum (WPI) et d'isolat de protéine de soja (SPI) ont été chauffées à différentes concentrations en protéines conduisant à la formation d'agrégats fractals polydisperses de taille moyenne variable. Lastructure des solutions a été analysée par diffusion de la lumière en fonction de la concentration en protéine. La compressibilité osmotique et la longueur de corrélation dynamique diminuent quand la concentration augmente deviennent indépendantes de la taille initiale des agrégats pour les suspensions denses. Pour une taille d'agrégat donnée, la viscosité augmente initialement exponentiellement avec la concentration croissante puis diverge. Plus lesagrégats sont grands, plus l’augmentation de la viscosité apparaît à des concentrations faibles. La dépendance avec la concentration de la viscosité des solutions d'agrégats fractals est beaucoup plus forte que celle de microgels. Le comportement de mélanges de différents types d’agrégats (fractals/fractals ; fractals/microgels et WPI/SPI) a étéétudié principalement par rhéologie.Le recouvrement de fluorescence après photoblanchiment (FRAP) a été utilisé pour étudier la diffusion de chaînes de dextran marquées par des fluorophores dans des solutions d’agrégats et des gels de WPI. Une diffusion brownienne estobservée dans des suspensions d’agrégats et des gels faibles formés juste au-delà de Cg avec un coefficient de diffusion (D) qui diminue avec l'augmentation de la concentration mais, avec une dépendance plus faible que celle de la viscosité (). A des concentrations plus élevées, des gels densément réticulés sont formés, ce qui induit une forte diminution de la mobilité des chaînes de dextran. Pour ces systèmes, la recouvrance de la fluorescence est logarithmique avec le temps,suggérant une distribution exponentielle des coefficients de diffusion. La diffusion des chaînes de dextran a également été étudiée en fonction de la concentration en protéines pour les suspensions de trois types d'agrégats de WPI (petits et grands fractals et microgels). / Polydisperse fractal aggregates of varying average sizes were formed when solutions of whey protein isolate and soy protein isolate were heated at different protein concentrations and at neutral pH. The structure of these fractals aggregates solutions was analyzed by light scattering as a function of protein concentration. In dense suspension, the osmotic compressibility and the correlation length decreases with increasing concentration and become independent of the initial aggregate size. In this concentration regime, the aggregates are strongly interpenetrated and can be visualized as a set of "blobs". For a fixed aggregate size, the viscosity initially increases exponentially with increasing concentration and then diverges at the gel point. Larger fractal aggregates show a more important increase of the viscosity with increasing concentration than smaller aggregates, because they are less dense. The increase of the viscosity was much stronger for large fractal aggregates than for homogeneous microgels (microgels were formed by heating the WPI solution in present of CaCl2) of the same size.Dynamic light scattering, rheology and FRAP measurements were performed to investigate mixtures of different type of aggregates of WPI (fractals/fractals, fractals/microgels) and fractals of mixtures of WPI and SPI. Flow measurements were used to characterise the rheological properties of the aggregate suspension whereas Fluorescence recovery after Photobleaching (FRAP) was used to determine the self diffusion of fluorophore-labelled dextrans chains in mixtures over a wide range of concentrations. The results were compared to the concentration dependence of zero shear viscosity, gel stiffness, osmotic compressibility and correlation length. Brownian diffusion of the dextran chains was observed in aggregate suspensions and weak gels formed just above the gel point with a diffusion coefficient that decreased with increasing concentration, but the dependence was weaker than that of the viscosity. At higher concentrations, densely crosslinked gels were formed, which induced a sharp decrease in the mobility of the dextran chains. For these systems, the recovery of fluorescence was logarithmic over time, suggesting an exponential distribution of diffusion coefficients.