Modèle monodimensionnel de type Piston-Dispersion. Le terme de dispersion axiale englobe alors toutes les imperfections de l'écoulement : injection des fluides non homogène dans l’espace et étalée dans le temps, effet de parois, et enfin volumes morts derrière les obstacles noyés dans le tamis (poutres, conduites...) dans le cas des Lits Mobiles Simulés. Cette représentation, quoique très simpliste, s'avère généralement suffisante tant que l’étalement des fronts de concentration est d'abord induit par les limitations au transfert externe, interne (macro/microporeux) et par la thermodynamique du système. Par contre, lorsque l’adsorbant employé présente d’excellentes performances de transfert, une approche aussi simple s'avère extrêmement risquée. En effet, les phénomènes dispersifs associés à l’adsorbant (transfert et thermodynamique) et à l’hydrodynamique ont des contributions de même ordre de grandeur sur la dispersion des fronts de concentration. Dans ce cas, une description plus réaliste de l'écoulement est requise afin de mieux appréhender son effet sur les performances de séparation.Dans ce contexte, l’objectif de ce projet de thèse est de mettre en place une méthodologie pour prendre en compte ces phénomènes hydrodynamiques lors de l’extrapolation d’un procédé de séparation par adsorption. Pour cela, nous proposons une étude du couplage entre les phénomènes hydrodynamiques et le phénomène d’adsorption. / Hydrodynamics inside industrial Simulated Moving Bed (SMB) adsorption columns can be complex due to the presence of internal distribution devices, free flow chambers and heterogeneous injections. These have to be taken into account in SMB numerical models to scale-up purposes. In the present thesis, a CFD approach is adopted as an intermediate step to develop a 1D model simple enough to be used for cyclic SMB simulations while being able to represent realistic hydrodynamics. This model results from the interpretation of the moments of the fluid age distribution, transported by CFD according to the method developed by Liu and Tilton (2010) that allows to estimate the degree of mixing (Liu, 2011) of the adsorption columns. The resulting 1D model consists in the two examples provided by Zwietering (1959) of a completely segregated system and a maximum mixedness system. This model is able to reproduce the residence time distribution of the CFD model of an adsorption column, while being representative of the internal flow patterns. This results in a good representation of the coupling of adsorption and hydrodynamics by the 1D model. When integrated in a SMB simulator and compared to the traditionally used dispersed plug flow model, the new 1D model demonstrates that for most of the adsorption column geometries considered a detailed hydrodynamic description is mandatory. Such detailed hydrodynamic description is even more important when employing adsorbents with better mass transfer performances than those currently used for the p-xylene purification, which is expected in the upcoming years.

Composite propellants are mainly composed of: crystalline oxidizer, metallic fuel, and polymeric binder. Additives, such as plasticizers, catalysts, bonding agents and curing agents may also be incorporated to propellant compositions in small amounts. These ingredients should be mixed rigorously in order to obtain a uniform microstructure throughout the cast propellant profile. The quality of the propellant mixture has to be determined quantitatively to improve the product quality and to reduce costs. In this study, it was aimed to develop an easy, cost effective and rapid test method for determining the optimum mixing conditions for the manufacturing process of solid rocket propellants. An analytical method used in the literature for assessing mixing quality in highly filled polymeric systems is wide-angle x-ray diffractometry (WA-XRD). After finding out the concentration distribution of the components indirectly by WA-XRD, degree of mixing was identified using statistical methods. To accomplish this, series of samples were taken from various locations of the mixing chamber and analyzed by WA-XRD. Degree of mixing calculations based on ratio of intensity arising from aluminum phase over total crystal intensity, and the ratio of intensity arising from ammonium perchlorate phase over total crystal intensity gave satisfactory results. Radial mixing efficiency of planetary mixer was determined, and it was found that mixing at the center was more effective than mixing at the outer regions. Also, by scanning electron microscopy technique (SEM), interactions between binder and solid loading during mixing process were observed. It was seen that polymeric matrix gradually encloses solid particles during mixing.