L'objectif principal de ce travail a porté sur l'optimisation du dimensionnement et de la gestion des installations de traitement des eaux usées par boues activées fonctionnant à basse température (10-12°C) et soumise à une contrainte de traitement poussé sur l'azote. Dans ce but, nous avons étudié la faisabilité de la nitrification à des charges massiques stabilisées supérieures à 0.1 kgDBO5.(kgMVS.j)-1. Le fonctionnement d'une installation pilote alimentée par un effluent synthétique, puis par un effluent urbain réel, a été étudié pour différentes charges massiques stabilisées à une température de 11 +/- 1°C. Les résultats très détaillés acquis en continu sur deux périodes de six mois ont suggéré le rôle et la hiérarchie des facteurs influençant la capacité de nitrification. Ils ont permis d'adapter l'outil de simulation dynamique Activated Sludge Model n°1 (ASM1). La double démarche expérimentale et numérique mise en oeuvre a permis de dégager les rôles des conditions de fonctionnement (âge des boues, charge volumique en azote, durée de présence d'oxygène) et des propriétés de l'effluent à traiter (ratios caractéristiques, part des différentes fractions) sur les performances de traitement de l'azote et sur la composition de la boue. Les valeurs limites des paramètres de gestion (charge massique maximale, durée d'aération journalière), permettant à différentes configurations d'installation (bassin unique, réacteur avec zone d'anoxie en tête) de respecter une concentration en azote total inférieure à 10 mgN/L dans le rejet, ont pu être dégagées. Des stratégies de gestion dans le cas de variations de flux importantes en entrée d'installation (cas des zones touristiques de montagnes) ont pu être évaluées (fonctionnement à charge massique variable par adaptation de la durée d'aération, ou bien utilisation d'un bassin d'aération à volume variable).

Nitrogen is the major component of biosphere. Paradoxically, nitrogen pollution is the concern globally. Ammonia pollution is due to its unceasing rejection into nature such as groundwater, current water and the atmosphere. This phenomenon constitutes a threat for the humanity, land and aquatic flora, and consequently disturbs the balance of natural ecosystem. Recently, that situation has lead to develop various techniques and/or technologies for ammonia removal from municipal and industrial wastewaters. Particularly in the environmental biotechnology area, two main objectives were recently aimed in many research activities: the development of new configurations of competitive bioreactors and the monitoring of partial nitrification process, which are the fundamental basis of this thesis project. In this study, the partial ammonium oxidation process, also called nitrite route, was studied in a 60 litre jet-loop submerged membrane bioreactor pilot plant. The research was organized around six chapters. An exhaustive literature review of the state-of- art of the biological nitrification process and the membrane technologies was performed. The materials and measurement methods were presented. The colorimetric method, the chromatography analysis, the biomass estimation by the suspended solids (SS), the aggregates size measurement, the gas holdup, the gas-liquid mass transfer, the bubbles gas diameter determination, the medium rheology aspects, etc., and the complete equipment of the bioreactor were studied in detail. The plant automation functioning was also studied. Membrane module (Mitsubishi Sterapore-L) characterization was carried out and three characteristic parameters were estimated: the membrane intrinsic resistance Rm, the membrane permeability Lp and the membrane porosity εm. Estimations revealed good agreement between experimental results and theoretical methods based on the Darcys law and the Carman-Kozeny law applicable in microfiltration system. Hydrodynamics and aeration aspects were studied. The mixing in the jet-loop system was characterized by the mixing time (tmix) and the circulation time (tc), respectively. The results showed that the characteristic times (tmix and tc) decrease with an increase in input gas flowrate and the circulated liquid flowrate. A model correlation involving the air and the combined liquid effects was proposed to describe the circulation time evolution. The classical non-steady state clean water test was used to determine the gas-liquid mass transfer coefficient (kLa). It was found to be influenced by the combined action of air and recirculated-liquid flowrates and a correlation has been proposed to describe their influence. The interpretation of kLa results and the system mixing data showed that the developed reactor corresponds to a near perfect mixing tank. This criterion was satisfactorily verified by literature data. The gas holdup (εg) was directly measured by the volume expansion method. In the absence of liquid circulation, εg ranged between 1 and 4% for the investigated range of gas liquid superficial velocities. It was found to increase linearly with the air superficial velocity, which corresponds to the bubbly flow regime. However, in the presence of liquid flowrate, εg slightly increased (from 1 to 6%) with increase in the superficial liquid velocity. A model has been proposed to correlate εg and the air and the recirculated-liquid velocities. The average diameter of the bubbles gas (dB) in the system was also estimated by the Leibson theoretical model based on the Reynolds number at the orifice of the gas distributor. Finally, biological aspects were studied. Respirometry measurements were conducted to characterize the process medium. The mass transfer, the gas holdup and the medium viscosity were determined. The obtained data allowed estimating the α factor and the β factor, respectively. The interaction of the growth of microorganisms into the process and the membrane performance was also investigated and a correlation model was proposed to describe membrane fouling with time. The optimal conditions for ammonium partial oxidation were determined using process monitoring and simulation. Dissolved oxygen (DO), temperature (T) and hydraulic retention time (HRT) were selected to achieve a high nitrite accumulation in the system. The results obtained showed that the selected parameters should be fixed at DO ≈ 2 mgO2.l-1, HRT ≈ 6 7 h and T = 30°C, respectively. The partial nitrification was simulated by the use of the TwoPopNitrification model included into the BioWin 2.2 software. For these simulations, a sequencing ammonia oxidation assumption was adopted: the nitrozation followed by the nitration step, respectively. The corresponding kinetics and stoichiometric constants were estimated by combining literature data and experimental nitrification results. For these estimates, the ammonium oxidation was monitored on several process samples taken at different times. The estimates were also delivered by monitoring the ammonium oxidation on the process operated in the batch mode. The plotting of simulations and experimental results revealed good agreement. In order to investigate the process performance in terms of biological stability, a long time period (≈ 600 days) was simulated. The results showed that a high stable nitrite accumulation (> 95%) could be achieved when the above optimal conditions are imposed to the system. However, after a long time, the accumulated nitrite is converted into nitrate and then the system is disrupted. For the simulated experimental conditions, the process disruption period was located between 180 and 350 days. At this period, a corresponding theoretical purge flowrate was found to range between 0.15 10-3 m3.d-1 and 3.0 10-3 m3.d-1. Simulations also showed that increasing the purge flowrate decreases the sludge retention time and then favours nitrite accumulation into the process. That is an interesting strategy to increase the performance of the biological partial nitrification process.