L’optimisation de l’incorporation et de la libération de principes actifs dans les produits formulés constitue un des enjeux majeurs des industries pharmaceutiques et cosmétiques. L'objectif principal de notre étude est de proposer un modèle prédictif de la diffusion de petites sondes au sein des émulsions concentrées. Pour cela, il a fallu considérer à la fois la formulation d’émulsions concentrées stables et leur caractérisation rhéologique et structurelle ainsi que la prédiction des paramètres de transfert des sondes au sein des émulsions concentrées. On entend par paramètres de transfert, tous les paramètres permettant de caractériser les différents mécanismes de transfert de sondes dans les émulsions concentrées pris en compte dans notre système, à savoir le coefficient de diffusion dans les phases continue et dispersée, le coefficient de transfert à l’interface eau/huile, le coefficient de partage à l’équilibre de la sonde entre les deux phases de l’émulsion. Une nouvelle approche de caractérisation de la structure des émulsions concentrée a permis l’obtention des paramètres importants de structure (taille des gouttes et épaisseur du film de la phase continue). L’étude détaillée des mécanismes et processus diffusionnels est réalisée avec la prise en compte des résultats liés à la caractérisation structurelle du système d’étude. Ainsi, un modèle de diffusion fondé sur une approche phénoménologique est proposé pour prédire l'évolution des profils de concentration de la sonde dans les émulsions concentrées. Les cinétiques expérimentales de libération des sondes sont comparables à celles simulées par le modèle sans paramètres ajustables. Cette comparaison montre une bonne adéquation entre le modèle de diffusion et l’expérience / In the field of controlled release technology for new drugs, models that can predict its delivery during application are important for device design. The main objective of this work is to develop a predictive model able to describe the drug delivery from highly concentrated water-in-oil emulsions. These systems consist of deformed droplets dispersed in a continuous film. Their structure’s characteristics make them favourable for their use as releasing devices. A combination of different transfer mechanisms has been implemented in a mathematical model in order to simulate release experiments under different operating conditions (volume fraction, oil/surfactant ratio). A sensitivity analysis has been performed to point out the most relevant parameters affecting the drug’s release: drug partition and diffusion coefficients. Partition coefficient of the drug for different surfactant concentrations has been obtained through a predictive thermodynamic model UNIFAC, and the diffusion coefficient using Chang and Wilke equations in addition to the Stefan- Maxwell development. An original and simple technique has been used to determine indirectly the mean droplet size of the concentrated emulsions, through measurements of continuous phase’s thickness by analysis of incoherent polarized steady light transport through emulsion samples. In a general view, the diffusion model proposed for small drug diffusion in concentrated emulsions, which was first proposed for diluted emulsions, predicts successfully the evolution of mandelic acid concentrations during release experiments undertaken in perfect sink conditions

The target of this master's thesis were the known methods for testing the quality of transfer parameters of data networks. I have studied the RFC 2544 document to analyze these tests. From that information and by studying other existing web-based applications I was able to design a concept of my own application that will allow to measure basic quality parameters of transfers made through the Internet (the parameters: downstream, upstream, latency, variance of these and a traced route). My application doesn't require any modifications on the user's system. It's built on the server-based programming language PHP and uses the relational database engine MySQL to store measurement and user data. On the client side, it's assisted by the JavaScript scripting language. Both registered users and visitors are allowed to perform the listed measurements. Registered users are able to browse the history of their own benchmark results and also to send messages to others. There is an administrative account to oversee the operations. Another part of my thesis work is an application for Windows that performs the same measurements but does not use JavaScript.