High-performance process-based simulation methods for stormwater management in urban areas

Rapid urbanisation is considered to be one of the reasons of increased risk for urban flooding, which has become a widespread natural hazard worldwide causing significant economic damage and loss of lives, as well as deteriorating stormwater quality and environment. Sustainable Drainage Systems (SuDS) are proposed as a promising approach for stormwater management to attenuate runoff, mitigate flood risk, and meanwhile create other benefits, including reducing non-point source (NPS) pollutants through infiltration, sorption and biological processes. To better design and implement SuDS to optimise their performances, a computationally efficient and robust physical-based numerical modelling tool is essential. A variety of numerical models have been reported for simulating the performance of SuDS in managing both the runoff and stormwater quality. Simulation of stormwater quality involves modelling the pollutant wash off and transport in urban areas, usually based on empirical methods or models solving the advection and diffusion equation. However, these methods have clear limitations in not being able to simulate the full physical processes of pollutant wash-off and transport and trace out the trajectories and the pathways of pollutants. The simulation of runoff with SuDS treatment has been widely carried out using hydrological models and simplified hydrodynamic models. Most of the hydrological models cannot provide sufficient details of the spatial variability and the simplified hydrodynamic models are not able to accurately predict the dynamics of transient flows. The limitations restrict the application of the current prevailing models in providing sufficient details in designing and implementing SuDS to effectively manage runoff and flooding from intense rainfall. There is a need to address the technical issues and develop new modelling approaches to improve the practice of SuDS design, implementation and assessment. A new coupled hydrodynamic and particle-based stormwater quality model is developed in this work to simulate the full-process dynamics of NPS particulate pollutants induced by rainfall in urban areas, from pollutant detachment, transport to deposition. A novel particle-based Hairsine-Rose (H-R) model is proposed to describe the detachment and deposition process of pollutant particles. This model is coupled with a random-walk particle-tracking modelling mothed to simulate the transport of pollutant particles. The whole model can directly trace out the trajectories of individual particles and hence identify the pathways of pollutant. Another full coupled hydrodynamic-SuDS model is further proposed by integrating the hydrology simulation of SuDS and surface hydrodynamic simulation together. The proposed model can simulate the hydrodynamics of runoff, overland flow and flooding interacting with SuDS devices. The two coupled models are implemented on Graphic Processing Units (GPUs) through the NVIDIA CUDA modelling platform to achieve high-performance computing. Several idealised and experimental test cases are successfully simulated to validate the two new coupled models. The full coupled hydrodynamic-SuDS model is finally applied to a real-world case study in Can Tho City, Vietnam to evaluate the performances of different types of SuDS for flood mitigations. Successful simulation of the test cases and application to the case study demonstrate that the two new coupled models can support accurate simulation of both flow hydrodynamics and stormwater quality related to SuDS, potentially providing new tools to better support for design, assessment, and operation of different types of SuDS.