La modélisation à l’aide de l’approche RANS (Reynolds Averaged Navier Stokes) a été menée en trois dimensions, en considérant les régimes permanent et non permanent, dans le but de simuler les écoulements au niveau des jonctions, bifurcations et déversoirs d’orage. A travers ces trois exemples d’étude, plusieurs stratégies de modélisation (différentes combinaison lois de paroi/modèles de turbulence, différents algorithmes de couplage pression/vitesse, différents schémas de discrétisation spatiale, différentes conditions aux limites, différents types et tailles de maille, etc.) ont été testées et évaluées à l’aide de plusieurs indicateurs de performance (de type RMS –Root Mean Square) en s’appuyant sur les données de vitesses (vitesses moyennes in situ et champs de vitesses obtenus par PIV en laboratoire), hauteurs d’eau, débits (répartition de débits en bifurcation en laboratoire ou débits déversés in situ). Les résultats obtenus sont transposables aux autres cas de singularités et ouvrages spéciaux rencontrés en réseau d’assainissement et montrent que : i) les résultats des simulations 3D sont sensibles à la rugosité, aux conditions limites de hauteur et de vitesse ; ii) dans les trois cas d’étude, les schémas de discrétisation du second ordre et l’algorithme de couplage pression/vitesse PISO sont appropriés ; ii) la loi de paroi scalable couplée aux modèles de turbulence de type k-ε pour le cas des jonctions (avec un débit latéral inférieur ou égal au débit principal) ou des déversoirs semblent convenir, tandis que le modèle de turbulence RSM associé à la loi de paroi enhanced ou scalable permet de mieux représenter les écoulements à travers les bifurcations ou au niveau des jonctions lorsque le débit latéral est dominant. Sur la base de ces résultats, une méthodologie de modélisation plus générale définie en six étapes et fondée sur le guide proposée par Jakeman et al. (2006) a été mise au point. La méthodologie ainsi définie a été utilisée pour i) améliorer l’instrumentation du déversoir OTHU (Observatoire de Terrain en Hydrologie Urbaine) situé à Ecully, à partir de la simulation de sa courbe de fonctionnement et en fournissant les incertitudes sur les débits déversés obtenus ; ii) simuler l’implémentation des capteurs débitmétriques à l’aval d’une jonction. Elle a permis de concevoir et de dimensionner un nouveau dispositif de maîtrise des flux d’eau et de polluants déversés (technologie DSM – dispositif de surveillance et de maîtrise des flux déversés au milieu naturel). Ce dispositif a fait l’objet d’un dépôt de brevet international. Enfin, la mise en œuvre de cette méthodologie a été à l’origine de la découverte d’une nouvelle structure d’écoulement dans la branche latérale d’une bifurcation à 90°. L’analyse des résultats des simulations des écoulements mettant en évidence cette nouvelle structure a montré qu’il était possible de prédire l’apparition de cette dernière à partir du rapport d’aspect et du nombre de Froude. / The understanding of sewer flows behaviour is a key component for better urban drainage monitoring and management. However, these flows are conveyed across singularities (such as bends, drops, deviations, etc.) and special structures (combined sewer overflows –CSOs–, channels junction, dividing flow structures, etc.). These singularities and specific structures exhibit complex geometries, leading to open channel turbulent, three-dimensional and multiphase (pollutants and storm and sewer waters) flows. Using three-dimensional CFD (Computational Fluid Dynamics) platform allows a better understanding of mechanisms of contaminants transport through these structures and singularities, leading to a better sewer monitoring. In this thesis, 3D-RANS (Reynolds Averaged Navier Stokes) modelling approach under steady-state conditions is used in order to study flows within CSOs, junctions and bifurcations. Through these three structures, several modelling strategies (wall law/turbulence model combination, velocity/pressure coupling algorithm, spatial discretisation schemes, boundary conditions, computational mesh –shape and size of cells–, etc.) are tested and evaluated thanks to performance indicators (such as RMS –Root Mean Square– indicators) based on velocities (in situ mean velocities and PIV velocity fields obtained in laboratory), water depths and discharge (discharge repartition for bifurcation in laboratory or in situ overflow discharge, for CSOs). Results deriving from these tests are transposable to other singularities or special structures encountered in sewer network and suggest that: i) simulated CFD results are sensitive to the roughness coefficient; ii) for the three studied structures, second-order discretisation schemes and SIMPLE or PISO velocity/pressure coupling algorithm are appropriate; iii) scalable wall function associated to the group of k-ε turbulence model for junction flows (with a lateral inflow lower or equal to the main inflow) or for CSOs is appropriate, whereas RSM turbulence model associated to enhanced wall function allows a better representation of bifurcation flows or junctions flows when the lateral inflow is greater than the main inflow. Based on these results and on Jakeman et al. (2006) guideline, a six steps-methodology focused on the using of RANS approach modelling has been proposed. This six steps-methodology is used in order to i) enhance the monitoring of an OTHU (Observatoire de Terrain en Hydrologie Urbaine) CSO located at Ecully accounting for uncertainties on overflow discharge values ;ii) simulate the performance of flowmeters downstream of a junction, defining the best location for these sensors. This methodology is used to design the new overflow discharge measurement device. This device is an international patent. Finally, the application of the methodology led to point out a new flow structure, occurring in the downstream lateral branch of a 90° bifurcation.

L’évaluation du débit en réseaux hydrauliques à surface libre est une problématique actuelle sur le plan scientifique, à forts enjeux technologiques, économiques et écologiques. Dans cette thèse, de nouvelles méthodologies d’instrumentation, basées sur une synergie entre mesures non intrusives de hauteur d’eau et modélisation numérique ont été développées. Celles-ci s’appliquent d’une part à des collecteurs dont le fonctionnement hydraulique est complexe et, d’autre part, à des ouvrages non-standard (Venturi, déversoirs d’orage). Ce travail de thèse multidisciplinaire vise une meilleure compréhension de l’écoulement pour en déduire des relations Q=f(hi) plus robustes, spécifiques à chaque site et associées à leurs incertitudes; mais également l’identification de possibles modifications du site de mesure afin d’améliorer l’estimation du débit. Au final, l’applicabilité des méthodologies développées a été éprouvée au travers de plusieurs études sur sites réels. / The evaluation of the flow rate in free surface water systems is a current scientific problem, related to high technological, economical and ecological issues. In this study, new methods of instrumentation based on a synergy between non-intrusive water level measurements and numerical modeling have been developed. These methods are applied first to sewer pipes with complex hydraulic conditions then to non-standard hydraulic structures (Venturi flumes, Combined Sewer Overflows). This multidisciplinary work aims at a better understanding of the flow to identify more robust site-specific Q=f(hi) relationships related to their uncertainties. It also aims at the identification of possible modification of the measurement site in order to improve the flow rate evaluation. Finally, the applicability of the developed methodologies has been tested through several real site studies.

En santé publique, de nouveaux programmes de surveillance et de gestion sont à proposer notamment pour les masses d'eaux fortement affectées par l'urbanisation et les rejets urbains par temps de pluie. Les niveaux, les réservoirs et les sources de contamination microbiologique des milieux aquatiques doivent être évalués et identifiés, et leur incidence sur la dissémination des bactéries pathogènes et les risques d'exposition des populations humaines comprises. Parmi les agents pathogènes retrouvés dans les milieux hydriques, Pseudomonas aeruginosa représentent une préoccupation sanitaire majeure. A partir de deux sites expérimentaux, les objectifs de ce travail de thèse furent de définir la contribution des rejets d'eaux usées sur la prévalence de P. aeruginosa dans les cours d'eau récepteurs et d'étudier l'écologie des formes introduites dont ces habitats incluant leur dynamique spatiotemporelle. Les sédiments, les biofilms de surface (périphyton) et les végétaux aquatiques submergés, permettraient leur survie voire la multiplication de certains génotypes de Pseudomonas aeruginosa. La répartition des indicateurs de contaminations fécales et de la bactérie pathogène Aeromonas caviae ont également été étudiée. Il a été mis en évidence que les forces hydrauliques, le morpho-dynamisme de la rivière et les variations saisonnières étaient des facteurs structurants de la répartition des contaminants microbiens analysés. La dangerosité des souches isolées a été évaluée et montré que toutes les souches avaient un potentiel de virulence élevée. Ceci s'est confirmé par la détection de clones épidémiques majeurs au sein de la collection dont des souches apparentées aux clones PA14 ou C. Les capacités métaboliques de ces souches ont été étudiées dont leur antibio-résistance / In Public Health, new monitoring and management programs are needed, especially for the aquatic environments strongly affected by urbanization and urban wet weather discharges. The levels, reservoirs and sources of microbiological contamination of the aquatic environment should be assessed and identified, and their impact on the spread of pathogenic bacteria and the risk of exposure of human populations understood. Among the pathogens found in water environments, Pseudomonas aeruginosa is of major health concern. From two experimental sites, the objectives of this work were to define the contribution of wastewater discharges on the prevalence of P. aeruginosa in receiving watercourses and study the ecology of the introduced forms, including their spatiotemporal dynamic and preferential habitats. Sediments, surface biofilms (periphyton) and the submerged aquatic vegetations appeared to favour the survival or growth of some genotypes of Pseudomonas aeruginosa. The distribution of fecal indicator bacteria and of Aeromonas caviae were also studied. It was highlighted that hydraulic forces, the morpho-dynamics of the river and the seasonal vaiations were determinant factors in the distribution of the analyzed microbial contaminants. The health hazard of the clones found in these systems was estimated through indirect molecular approaches. It was shown that all Pseudomonas aeruginosa strains had a high virulence potential and that some were related to the PA14 and C clones which are spread worldwide and pathogenic

The discovery of antibiotics is considered as one of the most significant scientific achievements of the 20th century – lives of millions of people and animals have been saved. Thenceforth, substantial amounts of administered antibiotics and their metabo-lites have been excreted into waste stream via urine and faeces. In this dissertation, primary focus is the qualitative balance of 14 antibiotics and one metabolite in urban water management and in urban waters, respectively. In particular, antibiotics pre-scribed to human beings are drained in the urban sewer system and finally enter the environment: (i) Continuously via the effluent of the wastewater treatment plant after a partially effective removal or degradation or (ii) Intermittent via combined sewer overflow structures due to capacity limitations of the urban drainage system. The fate and the potential effects and risks of these substances on ecosystems and hu-man health are of major concern – their direct toxic effect to all trophic levels as well as the global spread of antibiotic resistance genes are challenging. Hence, an assessment of microbial community activity due to antibiotic exposure is presented. In particular, systematic work has been carried out to study the presence and character-istics of 14 antibiotics in urban waters. In detail, investigations were conducted to gain scientific knowledge with respect to adsorption, desorption, abiotic, biotic and photolyt-ic degradation as well as activity-inhibition of microorganism communities in sewage and of natural freshwater biofilm communities, respectively, due to inevitable urban drainage overflows. In order to provide information to assist potential management strategies, which miti-gate surface water pollution and minimize the adverse impacts of antibiotics on activity of microorganism communities, the following specific topics were addressed: ⑴ The occurrence of 14 antibiotics and one metabolite were determined in sewag-es at three sampling sites in the city of Dresden, Germany. ⑵ The adsorption affinities of 14 antibiotics and one metabolite to size dependent sewer sediments were determined in experimental investigations, three sam-pling campaigns and subsequently an antibiotic-specific adsorption coefficient, normalized to organic content, was quantified. ⑶ The desorption affinity and -dynamics of 14 antibiotics and one metabolite were quantified in size dependent sewer sediments in experimental investigation and with statistical analysis. ⑷ The abiotic, biotic and photolytic degradation affinity of 14 antibiotics and one metabolite were quantified based on batch experiments with three different sewages at 7°C and 22°C, with artificial irradiation and different dilution ratios of the sewage at 30°C and subsequently a model framework decrypted ranges of abiotic, biotic and photolytic degradation coefficients. ⑸ The occurrence of three antibiotics, namely ciprofloxacin, clarithromycin and doxycycline was determined in sewage sampled during dry weather conditions in a small catchment of Dresden, which spills intermittently combined sewage (a mixture of sewage and storm water) to an adjacent brook in the case of capacity limitations of the urban drainage system during periods of intense rainfall and subsequently the three antibiotics were determined in the adjacent brook water. ⑹ Then, the activity-inhibition of microorganism community in sewage of this small catchment was quantified due to an exposition with three different antibiotics and three different antibiotic concentrations. ⑺ Last but not least, the activity-inhibition of natural freshwater biofilm communities in the adjacent brook was quantified via exposure to three antibiotics, which were individually dosed in three different concentrations, and also in mixture. ⑻ Finally, a two-dimensional hierarchical cluster analysis with dendrogram and heat map based on before mentioned activity inhibition of natural freshwater biofilm communities were conducted to identify hot spots of antibiotic tolerant and resistant bacterial subpopulations due to inevitable urban drainage system overflows.:List of Figures IV List of Tables VIII Symbols and Abbreviations XII List of Publications on the Ph.D. topic XIX 1 General Introduction 2 1.1 Background 2 1.2 Aims and Objectives 3 1.3 Innovation and Contribution to the Knowledge 4 1.4 Outline of this Thesis 4 1.5 References 6 2 Adsorption and Desorption Affinity of 14 Antibiotics and One Metabolite for particulate components in urban drainage systems 10 2.1 Introduction 11 2.2 Materials and Methods 12 2.2.1 Study area 12 2.2.2 Sewer sediment and sewage sample collection 12 2.2.3 Sediment fractionation 13 2.2.4 Antibiotic determination in sewage and sediment 13 2.3 Results and Discussion 18 2.3.1 Antibiotics in composite sewage samples 18 2.3.2 Antibiotics adsorbed to sewer sediments 19 2.3.3 Organic-bound antibiotic load as a linear function of liquid concentration 20 2.3.4 Adsorption dynamics and adsorption coefficient determined by bath experiments 20 2.3.5 Mineral composition of sewer sediment SED#1B 23 2.3.6 Initial characteristics of sediment SED#1B 23 2.3.7 Desorption dynamics and desorption coefficient of SED#1B 24 2.4 Conclusions 25 2.5 References 26 3 Abiotic, Biotic and Photolytic Degradation Coefficients of 14 Antibiotics and One Metabolite 32 3.1 Introduction 34 3.2 Materials and Methods 35 3.2.1 Study area and sample collection 35 3.2.2 Experimental set up 35 3.2.3 Modelling framework 38 3.2.4 Procedure of model calibration 40 3.3 Results and Discussion 43 3.3.1 Primary metabolic parameter 43 3.3.2 Secondary metabolic parameter 44 3.4 Conclusions 50 3.5 References 50 4 Activity-Inhibition of Microorganisms due to an Exposition with different Antibiotics and Concentrations 56 4.1 Assessing Antibiotic Resistance of Microorganisms in Sanitary Sewage 56 4.1.1 Introduction 57 4.1.2 Material and Methods 58 4.1.2.1 Sampling Site and Antibiotic Agents 58 4.1.2.2 Analyzing Antibiotics 60 4.1.2.3 Respiration Rate 60 4.1.3 Results and Discussion 60 4.1.3.1 Concentration Range of Antibiotics and Typical Sewage Parameters 60 4.1.3.2 Oxygen Uptake Rate 62 4.1.4 Summary and Conclusions 63 4.1.5 References 64 4.2 Hot Spots of Antibiotic Tolerant and Resistant Bacterial Subpopulations in Natural Freshwater Biofilm Communities due to Inevitable Urban Drainage System Overflows 66 4.2.1 Introduction 68 4.2.2 Material and Methods 69 4.2.3 Results and Discussion 72 4.2.4 Conclusions 76 4.2.5 References 76 5 Summery and General Coclusions 82 5.1 Adsorption and Desorption Affinity 82 5.2 Abiotic, Biotic and Photolytic Degradation 83 5.3 Activity-Inhibition of Microorganism Communities due to Antibiotic Exposure 84 5.4 Enhancement of the Stockholm County Council (2014) assessment of antibiotics 84 5.5 References 87 6 Proposed Directions of Future Research 90 7 Appendixes 94 7.1 Chapters 94 7.2 Figures 95 7.3 Tables 115 7.4 References 139

Artificial Neural Networks (ANNs) have been comprehensively researched, both from a computer scientific perspective and with regard to their use for predictive modelling in a wide variety of applications including hydrology and the environment. Yet their adoption for live, real-time systems remains on the whole sporadic and experimental. A plausible hypothesis is that this may be at least in part due to their treatment heretofore as “black boxes” that implicitly contain something that is unknown, or even unknowable. It is understandable that many of those responsible for delivering Early Warning Systems (EWS) might not wish to take the risk of implementing solutions perceived as containing unknown elements, despite the computational advantages that ANNs offer. This thesis therefore builds on existing efforts to open the box and develop tools and techniques that visualise, analyse and use ANN weights and biases especially from the viewpoint of neural pathways from inputs to outputs of feedforward networks. In so doing, it aims to demonstrate novel approaches to self-improving predictive model construction for both regression and classification problems. This includes Neural Pathway Strength Feature Selection (NPSFS), which uses ensembles of ANNs trained on differing subsets of data and analysis of the learnt weights to infer degrees of relevance of the input features and so build simplified models with reduced input feature sets. Case studies are carried out for prediction of flooding at multiple nodes in urban drainage networks located in three urban catchments in the UK, which demonstrate rapid, accurate prediction of flooding both for regression and classification. Predictive skill is shown to reduce beyond the time of concentration of each sewer node, when actual rainfall is used as input to the models. Further case studies model and predict statutory bacteria count exceedances for bathing water quality compliance at 5 beaches in Southwest England. An illustrative case study using a forest fires dataset from the UCI machine learning repository is also included. Results from these model ensembles generally exhibit improved performance, when compared with single ANN models. Also ensembles with reduced input feature sets, using NPSFS, demonstrate as good or improved performance when compared with the full feature set models. Conclusions are drawn about a new set of tools and techniques, including NPSFS and visualisation techniques for inspection of ANN weights, the adoption of which it is hoped may lead to improved confidence in the use of ANN for live real-time EWS applications.

In the last decades and due to the Water Framework Directive implementation, the urban wastewater cycle management has become more complex. The concept of integrated urban wastewater system management is introduced, and it becomes necessary to consider some new data as the sewer system characteristics or the receiving body. In this sense, environmental decision support systems (EDSS) are very useful and powerful tools to support during the decision making process. A new EDSS for the integrated urban wastewater system management has been developed integrating data from several sources (bibliographic, theoretical or historical) and a real case-based virtual system able to perform simulations. Results demonstrate the benefits of using this kind of systems in comparison with a standard approach, which do not use the expert knowledge and encourages continuing with this research in order to improve the quality and efficiency of this type of EDSSs / En les darreres dècades, i degut a la implementació de la Directiva Marc de l’Aigua, la gestió del cicle urbà de l’aigua ha esdevingut més complexa. S’introdueix el concepte de gestió integrada de conca, i és necessari considerar certa informació, com les característiques de la xarxa de clavegueram o del medi receptor. D’aquesta manera, els sistemes de suport a la decisió ambiental (EDSS) són eines potentíssimes que faciliten la presa de decisions en aquest camp. S’ha desenvolupat un nou EDSS per a la gestió integrada de conca que utilitza una base sòlida de coneixement expert, integrant informació de diverses fonts (bibliogràfiques, teòriques o històriques) i dos sistemes virtuals basats en dades reals sobre els quals és possible realitzar simulacions. Els resultats demostren que aquest sistema presenta beneficis importants respecte a una gestió estàndard sense sistema expert, i esperona a prosseguir amb la recerca i el desenvolupament per a millorar-lo

La gestion en temps réel (GTR) est considérée comme une solution économiquement efficace pour réduire les déversements par temps de pluie car elle optimise la capacité disponible des réseaux d'assainissement. La GTR permet d'éviter la construction de volumes de rétention supplémentaires, d'augmenter l'adaptabilité du réseau aux changements de politiques de gestion de l'eau et surtout d'atténuer l'impact environnemental des déversoirs d'orage. À la suite de l'intérêt croissant pour la GTR fondée sur la qualité de l'eau (QBR), cette thèse démontre une stratégie simple et efficace pour les charges polluantes déversées par temps de pluie. La performance de la stratégie QBR, basée sur la prédiction des courbes masse-volume (MV), est évaluée par comparaison avec une stratégie typique de GTR à base hydraulique (HBR). Une étude de validation de principe est d'abord réalisée sur un petit bassin versant de 205 ha pour tester le nouveau concept de QBR en utilisant 31 événements pluvieux sur une période de deux ans. Par rapport à HBR, QBR offre une réduction des charges déversées pour plus d'un tiers des événements, avec des réductions de 3 à 43 %. La stratégie QBR est ensuite mise en oeuvre sur le bassin versant de Louis Fargue (7700 ha) à Bordeaux, France et comparée à nouveau à la stratégie HBR. En implémentant QBR sur 19 événements pluvieux sur 15 mois, ses performances sont constantes et apportent des avantages précieux par rapport à HBR, 17 des 19 événements ayant une réduction de charge variant entre 6 et 28.8 %. La thèse évalue en outre l'impact de l'incertitude de prédiction de la courbe MV (due à l'incertitude de prédiction du modèle) sur la performance de la stratégie QBR, en utilisant un événement pluvieux représentatif. La marge d'incertitude qui en résulte est faible. En outre, l'étude de sensibilité montre que le choix de la stratégie QBR ou HBR doit tenir compte des dimensions réelles des bassins et de leur emplacement sur le bassin versant. / Real time control (RTC) is considered as a cost-efficient solution for combined sewer overflow (CSO) reduction as it optimises the available capacity of sewer networks. RTC helps to prevent the need for construction of additional retention volumes, increases the network adaptability to changes in water management policies, and above all alleviates the environmental impact of CSOs. Following increasing interest in water quality-based RTC (QBR), this thesis demonstrates a simple and nothing-to-lose QBR strategy to reduce the amount of CSO loads during storm events. The performance of the QBR strategy, based on Mass-Volume (MV) curves prediction, is evaluated by comparison to a typical hydraulics-based RTC (HBR) strategy. A proof-of-concept study is first performed on a small catchment of 205 ha to test the new QBR concept using 31 storm events during a two-year period. Compared to HBR, QBR delivers CSO load reduction for more than one third of the events, with reduction values from 3 to 43 %. The QBR strategy is then implemented on the Louis Fargue catchment (7700 ha) in Bordeaux, France and similarly compared with the HBR strategy. By implementing QBR on 19 storm events over 15 months, its performance is consistent, bringing valuable benefits over HBR, with 17 out of 19 events having load reduction varying between 6 and 28.8 %. The thesis further evaluates the impact of MV curve prediction uncertainty (due to model prediction uncertainty) on the performance of the QBR strategy, using a representative storm event. The resulting range of uncertainty is limited. Besides, results of the sensitivity study show that the choice of the QBR or HBR strategy should take into account the current tank volumes and their locations within the catchment.