Flooding is an unwelcome and increasingly frequent phenomenon in many urban areas. However, as recent UK floods have shown, serious inundation can still occur even if river discharge has not exceeded the functional flood plain capacity due to 'pluvial' or sewer flooding. This work focuses on sewer flooding and its management. The aim of this study is to develop a methodology to support effective sewer flood risk management within a decision support framework, based on a sound understanding of the urban flooding regime. An urban drainage modelling tool was used to generate a flow time series based on long-term continuous rainfall inputs (rather than individual storms). A catchment delineation method was applied on a detailed digital elevation model to predict the extent of flooding and flood stages were assigned to the different flood discharges at each property in a catchment. A flood risk model was developed to assess flood risk in terms of probability and consequence. Flood probability was identified using statistical analysis for the flood stages, and flood consequence was obtained from the relationship between the flood stage and the corresponding damage cost. Annual average flood risk for each property was identified by linking the flood probability, stage and damage. A decision support framework was developed to allow the rational comparison of different management strategies ranked in terms of highest cost benefit difference or benefit cost ratio. This was used to compare a wide range of different sewerage rehabilitation and management scenarios in terms of flood costs and benefits for a particular case study. The wider application of the framework was demonstrated.

The overall aim of this work is to determine the ability of several types of in-sewer flow regulator to limit and attenuate, in conjunction with some type of storage unit, the flow of stormwater into a sewer system. The intended result of these devices is the reduction of overland flooding and downstream surcharge during storm events. Various stormwater management (SWM) techniques, practised both in the UK and abroad, were reviewed with particular emphasis on attenuation and control. Several types of flow regulators were tested where the choice was based on their popularity (i.e orifice plates and vortex), and novelty (i.e self regulating valve and throttle hose). A full size laboratory rig was built to allow experimentation and comparison of in-situ performance of several flow regulators. This system was capable of generating flows of up to 100 I/s which was controlled by a pneumatic valve interfaced to a dedicated PC. The flow characteristics(discharge and upstream head) were measured efficiently and accurately by two ultrasonic sensors. Investigation of the blockage performance of the vortex flow regulators indicated that these devices retained gross solids at higher heads (i. e. after formation of vortex) which either caused increase in discharge (up to 25% for a given upstream head) or reduced the outlet area (i. e. blockages). In terms of head/discharge relationships, the SRV offered optimum performance with consistent and near constant outflow (discharge changing 3% over 1m change in head). Vortex devices exhibited a linear relationship which is attributed to small changes in the air-core diameter. Comparisons were also made on the performance of these devices in terms of outlet opening, storage and passage of gross solids. A theoretical model was developed for the vortex regulator by combining the hydraulic properties of the free vortex with Bernoulli's equation and applying the principle of maximum discharge. In this way discharge could be predicted for a vortex flow regulator with an inlet opening in terms of upstream head and geometry of the device. The model was calibrated using the laboratory data and verified against independent data sets on blocked devices and from external sources.

The poor quality of many receiving waters has been attributed to the frequent operation of combined sewer overflows (C. S. O. s). As the existing need for overflow structures is unlikely to change in the foreseeable future it is imperative that these deleterious effects are minimised. The present study investigates the pollution performance of three common overflow structure designs (the stilling pond, the double high side weir and the double low side weir) and aims to institute a novel pollution monitoring methodology for combined sewer overflows. Four sites in the central Sheffield area were monitored for periods from 9 to 13 months (two high-side weirs, one low-side weir and a stilling pond). Each overflow was monitored with continuous flow measurement equipment and bottle samplers to obtain samples of sewage from the storm and dry weather flows. The bottle samples were analysed for suspended solids (SS), ash, BOD, COD, pH, conductivity and ammonia. Mesh bags and frames were also installed to trap the gross solids (solids with a median size>6mm) from the inflow and the spill flow. The stilling pond and the high side weirs were found to perform well hydraulically, limiting the flow to treatment to a steady maximum. The low side weir performed unsatisfactorily, hydraulically, as the flow to treatment rose as the incoming flow increased and, for some storm conditions, a hydraulic jump formed towards the downstream end of the chamber. The first foul flush was regularly observed at the stilling pond and low side weir sites. Peak concentrations for SS were found to be 600 times greater than the dry weather flow for the same time of day. The first foul flush was rarely observed at the other sites. For the majority of storms at each site the spill concentrations were of a similar magnitude to the inflow sample concentrations. However for a large minority of SS, BOD and COD samples, the concentrations of the spill samples were significantly less than the inflow samples. t-Tests suggested that at the stilling pond and high side weir sites there is a significant reduction in the spill sample concentrations for the water quality (bottle) samples. Although the load of material spilled during an overflow event was found to be small in comparison to the inflow load, large amounts of material were spilled to the watercourse during storm events at each of the sites investigated. The storm load entering the CSO was found to be considerably influenced by peak intensity of the storm at the stilling pond site and antecedent dry weather period at one of the high side weir sites. At the other sites a number of hydrological factors were found to be influential e. g. duration. It is thought that time of year may also be important factor as this influences the type of rainfall (its duration and intensity). The types of gross solid collected at each site were similar with leaf material and sanitary towels consistently being the major items in terms of total mass. The efficiency of the stilling pond and one of the high side weirs in retaining gross solids in the flow to treatment appeared to be explained by the flow split although for 5 of 14 storms at the stilling pond and 3 of 7 at the high side weir a treatment effect was observed. The treatment factors at the low side weir were noticeably less than those for the other three sites with all being less than unity (average 0.5). This suggests that the low side weir preferentially discharges gross solid material over the weir. The treatment factors at the other high side weir were low due to inadequate sampling of the spill flow.

This thesis examines ways in which the design of storm water drainage networks can be optimised and proposes, develops and tests some such methods. The introduction is followed by a r~su~ of current design practice and an examination of previous work on the drainage optimisation problem. Methods of estimating the construction cost of a drainage ne:twork are detailed and fwctions proposed for modelling these costs. The optimisation problem may logically be split into two areas, namely optimising fixed plan networks and optimising variable plan networks. The former involves the simultaneous selection of gradients and diameters for a network of pipes fixed in plan. A new Dynamic Programming model is proposed for this, having several advantages over previously published methods. The main area of innovation is, however, in optimising variable plan networks. The general plan optimisation problem is seen to be far too complex for solution. However~ taking the special case of road drainag~ networks» two possible modes of optimisation are defined. These are, firstly, the positioning of an unknown number of manholes along a drain running between two fixed manholes, and secondly, the positioning of an unknown number of cross-drains along a road carriageway. Both modes include the simultaneous choice of pipe gradients and diameters. Models for these modes are proposed, with practical computer programs being developed and tested. Both models use a novel form of Dynamic Programming conceived and developed during this research. The thesis ends with a brief outline of a Dynamic Programming solution to a rather different variable plan problem, followed by suggestions of areas for further study and conclusions of both a specific and a general nature.

The use of storage tanks in sewerage systems has increased in recent years. The primary functions of such tanks are to attenuate flow and to retain pollutants within the sewer system. The main problem is to provide storage and effective separation of gross and suspended solids without incurring poor self - cleansing and associated high maintenance costs. The size of the required storage volume is dependant on the purpose for which the tank is to be used, but the end product of any design analysis is a fixed volume of storage. This project has involved the development of fullscale and laboratory computer controlled monitoring systems for the purpose of flow visualisation and digital imaging and for the comparative assessment of the sediment removal performance of different geometric configurations of overflow and storage tank. These systems used sophisticated control procedures and the latter had the facility to generate a flow hydrograph of any shape and duration and to superimpose on this hydrograph a pollutograph of synthetic cohesive sediment of any distribution. A wide variety of tanks have been constructed, mainly rectangular in plan shape (some circular), but both online and off-line and with and without an overflow structure. The use of benching to the chamber floor and the inclusion of a dry-weather flow channel were common but not universal features.

A model is developed, using a unit hydrograph approach for sewer flow synthesis and a simple mixing model to calculate the pollution load from combined sewer overflows, to simulate the long-term behaviour of storm overflows in combined sewerage systems. It is shown that this procedure synthesizes overflow operation characteristics to acceptable engineering accuracy, measured relative to the adopted standard for UK practice, namely predictions from the WALLRUS suite of engineer's software. This is achieved at only a small fraction of the computer runtime and so makes practicable a wide range of overflow performance and river impact studies using local rainfall records of unlimited extent. The model is applied successfully to three drainage networks. Results show that the model, COSSOM, achieves predictions with respect to runoff volume and overflow characteristics well within : 00 % of full. WALLRUS applications. Sensitivity studies demonstrate that performance of the model improves when catchment unit hydrographs are obtained, by preliminary application of WALLRUS using a rainfall intensity close to the maximum in the observed data and for rain duration close to the time of concentration of each sub-catchment under study. It is also shown that with little loss in accuracy relative to application of the WALLRUS approach for unit hydrograph development, COSSOM can be operated with other rainfall/runoff/pipeflow models.

Literature survey made in this research indicated that the roughness has a significant effect on the design of a self-cleansing sewer. A conceptual model was therefore developed taking into account the effect of roughness in the sediment transport in sewers. This model was later tested against experimental data obtained from a test rig. The rig had a pipe which was specially fabricated in two halves so that it could be divided to coat the interior with uniform sand grains to provide roughness. The experimental data shows a good correlation with the model developed. Two relationships i.e. for smooth and rough pipes, were derived from the results. The effects of volumetric sediment concentration, pipe diameter, sediment size and effective roughness on these relationships were in turn examined. Head loss formulae acquired by past researchers were slightly modified to suit the range covered by the experimental data. The relationships developed in this research were subsequently applied to sewer design. These were later compared to the criterion of 0.76 m/s proposed by the British Standards. In the case of smooth pipes, it shows conclusively that the criterion of 0.76 m/s produces excessive slopes for pipe diameter up to 1.0 metre. However, for rough pipes the criterion gives insufficient slopes to maintain a self-cleansing sewer. This is valid up to a pipe diameter of 0.3 metre. Beyond this diameter the criterion gives high slopes. Comparisons were also made with studies in the wider field of sediment transport on fixed bed. These have set in context the proposed formulae for smooth and rough pipes in the design of storm sewers.

Ce travail présente le Contrôle en Temps Réel (CTR) des Réseaux d’Assainissement (RA) dans le cadre des villes intelligentes. Le CTR nécessite de comprendre le fonctionnement du RA et d'effectuer des simulations sur des évènements mesurés, prévus et synthétiques. Par conséquent, un système de Surveillance en Temps Réel (STR) a été installé sur le site expérimental, et combinée à un modèle de simulation. Une méthode d'auto-calage des modèles hydrauliques et un système de prévision des conditions aux limites, ont été développés. Visant à protéger les citoyens et d'atténuer les conséquences des inondations, le CTR est composé d'un système de prévision des inondations suivi d'une gestion dynamique. Le concept et les méthodes proposés ont été appliqués sur le campus de l'Université de Lille 1, au sein du projet SunRise. STR a été trouvé très utile pour comprendre le fonctionnement du RA et pour le calage du modèle de simulation. L'Algorithme Génétique suivi par Pattern Search ont formé une procédure d'auto-calage efficace. NARX Neural Network a été développé et validé pour la prévision des conditions aux limites. Une fois l’opération du RA est analysée, le CTR a été développé. NARX Neural Network a été trouvé capable de prévoir les inondations. Une gestion dynamique pour augmenter la capacité de rétention du réservoir, a été étudiée sur la base du calcul de la variation temporaire de l’ouverture d’une vanne, et les résultats ont été satisfaisants en utilisant l'Algorithme Génétique et l’Algorithme des Abeilles, comme méthodes d'optimisation. Une gestion qualitative a également été examinée et testée pour vérifier son potentiel dans la réduction des volumes d'inondation. / This work presents the Real Time Control (RTC) of Urban Drainage Systems (UDS) within smart cities. RTC requires to understand the UDS operation and to perform simulations on measured, forecasted and synthetic events. Therefore, a Real Time Monitoring system (RTM) was implemented on the experimental site, and combined to a simulation model. A model auto-calibration process and hydraulic boundary conditions forecast system were developed, in order to simulate the hydrologic-hydraulic response. Aiming to protect the citizens and mitigate flooding consequences, the RTC was composed of a flooding forecast system followed by a dynamic management strategy. The proposed concept and methodologies were applied and evaluated on the Lille 1 University Campus, within the SunRise project. RTM was found very helpful in understanding the system operation and calibrating the simulation model. Genetic Algorithm followed by Pattern Search formed an effective auto-calibration procedure for the simulation model. NARX Neural Network was developed and validated for forecasting hydraulic boundary conditions. Once understanding the UDS operations, the RTC was developed. NARX Neural Network was found capable to forecast flooding events. A dynamic management for increasing a tank retention capacity, was studied based on calculating a Valve State Schedule, and results were satisfying by using Genetic Algorithm and a modified form of Artificial Bee Colony, as optimization methods. A qualitative management was also proposed and tested for verifying its potential in reducing flooding volumes.