Emulator-aided optimization of detention tanks for flood reduction

Conference item

Stormwater detention tank was widely used as an effective measure in improving the performance of an urban stormwater drainage system (USDS) and mitigating urban flood risk during intensive storm events. Different layouts of detention tanks can largely affect their efficiency; therefore, it is crucial to determine their appropriate locations. This study investigates the feasibility of integrating urban hydrological simulation into an optimization model, meanwhile taking into consideration the constraints of realistic flooding control criteria, for the purpose of developing a general methodology for the optimal location design of detention tanks. For demonstration of the proposed method, the Storm Water Management Model (SWMM) was run on a hypothetical urban catchment, concerning different specifications of detention tanks in the USDS. The performance criteria of the nominated detention tanks were set as the computed value of the total flood volume in the USDS, as these criteria were closely connected to the loss ratio of property assets and infrastructures in an urban flood event. An artificial neural network (ANN) model was then established as an emulator to save the computational effort in hydrological simulation. Genetic algorithm (GA) was employed to automatically search the optimal solution in the optimization process. The optimal design of detention tanks could lead to a reasonable reduction of flood volume with a relatively lower cost when compared to an empirical design. The proposed study could help water management authorities or governmental agencies in reaching a more cost-effective decision-making.

Authors: Lu, W; Qin, X; Yu, J

• Publication status: Published
• Peer review status: Peer reviewed
• Publisher: International Association for Hydro-Environment Engineering and Research
• Host title: 37th IAHR World Congress
• Journal: 37th International Association for Hydro-Environment Engineering and Research World Congress
• Publication date: 2017-08-01
• Acceptance date: 2017-04-19
• EISSN: 2521-7127
• ISSN: 2521-7119
• Keywords: flood mitigation; artificial neural network; optimization; Storm water detention tank;; genetic algorithm