Optimisation de la gestion en temps réel des réseaux urbains de drainage basée sur la qualité des eaux

Abstract

Real-time control of sewerage systems is increasingly preferred to reduce the magnitude of floods and overflows impacts on the natural environment. This dynamic control approach allows monitoring, analyzing and controlling the operation of these networks with a small lag time, by corrective or preventive actions. These actions consist in remotely control regulation structures in order to minimize wastewater discharges to receiving waters during rain events. However, three prerequisites are essential to the success of real-time control. The first one is to acquire accurate knowledge of the network hydraulics’ behavior by using modeling. Then proceed with the quantitative and qualitative forecast of flows and loads to finally establish optimal set points. The work outlined in Article 1 allows the simulation of drainage networks hydraulics’ behavior, using two modeling approaches. The first, which is a global approach, is based on the Generalized Rational Method for which the entire network upstream of a regulation location or of the pipe section to be evaluated is aggregated as a single node. The global hydraulic model based on this method uses the generalized Bernoulli equation applied to each pipe to be evaluated. The second approach, more detailed, is based on the coupling of the Nonlinear Reservoir Model to simulate runoff at the outlet of each sub-network with the Saint-Venant equations, to laminate the resulting hydrographs within the network. The case study was conducted to compare the results obtained by these two approaches by applying them to small synthetic and real networks. Substantially equivalent results were obtained. The Generalized Rational Method however, when compared with the Non-Linear Reservoir method, has the advantage of achieving the pursued objectives by requiring much less field data and modeling efforts. Conclusions of this first article are to the effect that the proposed model can adequately simulate networks behaviors with an acceptable degree of accuracy for its use for controlling in real-time sewer systems and for hydraulic behavior needs assessment purpose. The primary focus of article 2 is to develop a simple but still efficient real-time flow and load prediction tool using only a minimal set of watershed data at the outfall of an urban sewer shed. This tool is aiming at emulating sewer sheds inclination to attenuate hydrographs, without requiring any detailed modeling of a given sewer network to get the final hydrograph. For such reason and in order to allow computation of direct runoff of desired magnitude and duration at an outlet, a structured procedure is proposed through the use of a linked Improved Rational Hydrograph (IRH) method with a Modified Linear Muskingum semi-empirical model transfer function. The IRH equation allows generating realistic flows and loads from a minimal rainfall and sewersheds data set. The modified Muskingum model, once transformed in an autoregressive model enables the simulation of routing hydrographs within the networks. The system of equations is solved by coupling the resulting discrete convolution equation with a Kalman Filter, to enable in real time a dynamic calibration of the Muskingum model coefficients and parameters, by readjusting them using the computed prediction errors at each time steps. Using this Filter has allowed to overcome the lack of reproducibility that can happen during an event, and from one event to another. As an indirect result, the use of this filter has permitted to get rid of the calibration procedure traditionally required under classic static modeling. Using both models in combination has generated a noteworthy gain in performance in comparison with their separate utilization. In addition, the proposed algorithm has allowed managing non stationary models which parameters vary over time, and in doing so has improved the accuracy of the desired forecast. Article 3 is specifically focused on establishing optimal set points for RTC based on water quality parameters at outfalls of urban sewer networks, to enable the control of the capture of urban wet weather wastewater. The main task of this work was to assess the performance of this water quality-based approach. The methodology was tested successfully using a simplified generic interceptor network and measurements from the City of Hamilton’s (Ontario) 2010 CSO Water Quality Characterization and Monitoring Program. This work with a generic simplified interceptor network has demonstrated that this approach can capture more pollutants than a static scenario or a flow maximization scenario. The case study has shown that despite the fact the generic interceptor is a simplify representation, results are showing that a water quality-based management approach could be a suitable approach to pollution control. Both Mathematical Loads Optimization and Rules based Loads Maximization scenarios considered in this study performed better than the Flow Maximization scenario and at the same time, confirmed the validity of the proposed quality based approach. While the gain in performance for a Flow Maximization scenario in comparison with a Static Base scenario is 53%, it reached 83% for the Rules based Loads Maximization scenario and 100% for the Mathematical Loads Optimization scenario respectively. This approach, as proposed in this study, could be an efficient mean to prioritize overflows at less vulnerable locations based on water quality parameters, in order to make maximum use of available capacity within the collection system, to carry the most polluted component of the wet weather flow to be treated at the treatment plant. Consequently, this tool, while using very little data, could be considered as being an adequate complementary tool to be used as an extension of a flow maximization approach, within a broaderefficient pollution control system, for the management of urban sewer networks.

Date	22 Oct 2012
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Saad Bennis (Supervisor)