Enhanced biological phosphorus removal (EBPR) was researched from the performance of a modified University of Cape Town (UCT), anaerobic-anoxic/nitrifying-aerobic process. The work focussed on high P influent where milk was compared to carbohydrates as exogenous added carbon and typical settled sewage. The results confirmed that at equal COD load in the influent (minimum COD:P (250:5) ratio for EBPR), milk always provided sufficient soluble substrate than the carbohydrate mix, but also improved the EBPR performance. The laboratory scale treated 10L/day where 2 parallel treatment trains for milk and an equivalent carbohydrate mix as supplement to compare and study the P sequestration from hypothesised P ligands in milk and easily assimilable carbon (AOM) after fermentation for biological P uptake. The aerobic bioreactors used submerged flat sheet membranes (AeMBR) to improve the effluent quality and reduce the suspended solid residues. The results suggested extra benefits from adding calcium chloride (CaCl2) (200 ml at 250 mM/day or 200 mg/L treated) to form P complexes both in the anaerobic and aerobic zones (100 ml CaCl2 250mM/zone/day). To complete P removal a calcium phosphate (CaPO4) further treatment stage (post membrane final effluent (F.E.)) was added for nucleation. The combination of, A2O-N, exogenous carbon and calcium addition improved the performance of the EBPR, and enabled the laboratory units to achieve less than the 1 mg/L P required by the EU Directive. The process was tested at higher than normal P loads (maximum 100 mg/L) (domestic wastewater influent 15 mg/L). Experiments with influent P load ≤ 50mg/L, with 1% milk as AOM were compared to the carbohydrate mix and could remove soluble P to less than 1mg/L above 97% and less than 2 mg/L more than 99% of the in the time respectively. With an influent P load of 60mg/L (maximum 100 mg/L), the soluble P in the F.E. with milk was below 5 mg/L and below 8 mg/L with carbohydrates mix. The results showed that most of the phosphorus was retained by the sludge during the anoxic-aerobic phases. The remaining phosphate in the F.E. was able to pass through AeMBR pore size (0.4 μm) and needed to be chelated by the nucleation process. The results indicated this A2O-N modifications achieved stable nutrient removal and also offered the potential for more sustainable phosphorus recovery. The EBPR without AOM was 25% less efficient compared to milk and never achieved the E.U standard of 1mg/L in final effluent. The flat sheet membrane always achieved a NTU final effluent below 1 and the TOC always greater than 90% removal or less than the EU 125 standard regardless of the feeding COD/P ratio.

Ce travail vise à étudier un procédé de traitement d'eaux usées industrielles par cristallisation en milieu fondu sur paroi froide. Deux effluents modèles sont choisis : un mélange eau/acétone et un mélange eau/acide propanoïque. Dans un premier temps, l'équilibre solide-liquide du mélange est étudié pour connaitre les limites de l'étude en température et en concentration. Le procédé de traitement de l'eau est alors conduit, selon un mode opératoire précis. Deux modes de fonctionnement sont mis en œuvre, un mode statique et un mode dynamique avec une circulation en boucle de la solution. La concentration en impuretés dans la glace est analysée après chaque cycle de congélation. L'étude paramétrique, conduite suivant un plan d'expériences, a mis en avant les effets importants de la concentration initiale de la solution et de la rampe de refroidissement. La microstructure de la glace est également analysée par microscopie en chambre froide pour interpréter les mécanismes d'incorporation d'impuretés au sein de la glace. Les inclusions de liquide sont sous la forme de poches de solution à faible vitesse de croissance et sont localisées dans les joints de grain à plus forte vitesse. Enfin, la modélisation du procédé fonctionnant en mode statique, par les éléments finis et en 2D axisymétrique avec frontière mobile, montre la présence de mouvements de convection. En mode dynamique, l'hydrodynamique de l'écoulement dans l'espace annulaire est décrite par une modélisation 3D prenant en compte la position de l'entrée et de la sortie. Les résultats démontrent la faisabilité du procédé et permettent des avancées significatives dans la compréhension des phénomènes mis en jeu / This work aims to study a process of industrial wastewater treatment by melt crystallization on a cold wall. Two effluent model solutions are chosen: water/acetone and water/propionic acid binary mixtures. First, the solid liquid phase diagrams are determined experimentally in order to delimit the operating range of temperature and concentration. Then, a parametric study of the wastewater treatment process by freezing is performed, by means of an experimental design, for two working modes, static mode and dynamic mode by adding a recirculation loop, respectively. The impurity concentration in the ice is analyzed after each freezing cycle. The process requires applying very precise conditions and the ice concentration mainly depends on the initial solution concentration and on the applied cooling rate. The ice microstructure is also characterized by optical microscopy in a cold chamber and gives insights into the mechanism of impurity incorporation: the liquid inclusions are localized under the form of solution pockets at low growth rate or between the polycrystals at higher growth rate. Lastly, 2D axisymmetric modelling of the process in static mode, based on finite elements and taking into account the moving boundary, shows the presence of buoyancy loops in relation with the density dependence of the solution with temperature. In dynamic mode, the hydrodynamics in the annular space is described by a 3D model to account for the positions of the inlet and outlet pipes. All the results demonstrate the process feasibility and allow better understanding of the occurring phenomena