부산대학교 도서관

Severe flood events throughout Europe in recent years have increased political, public and scientific awareness of the risks posed by large flood events. In response, engineers and researchers have transferred their attention from rural studies to consideration of urban areas where risk is concentrated. Computational fluid dynamics methods have been extensively employed in the evaluation of in-channel and out-of-bank flow processes in natural rivers in the last 20 years. In the case of urban flood events, computationally efficient methods are required to estimate flood risk at fine scale details over wide areas. The overall aim of this thesis was therefore to understand the controlling features of urban areas for floodwavc propagation and subsequently, develop computationally efficient methods to evaluate flood risk. The first component of this research was focused on determining the features of urban areas that modulate floodwave dynamics. Subsequently, the effect of grid resolution on the representation of urban features and flood propagation was investigated to determine the compromise between computational cost and model performance. It was found that floodwave propagation through urban areas in the UK is controlled by the distribution of building sizes and separation distances. The second part of this thesis details the development and evaluation of sub-grid scale porosity techniques aimed at harnessing high resolution topographic data sets within coarse resolution numerical models. For the first time, this research presents a consistent and rigorous evaluation of a variety of porosity techniques for flood modelling. The results suggest that representing the broad scale effect of buildings and obstacles on flood flows provides the best compromise between data demands, pre-processing requirements and computational cost. Indeed, the use of a porosity techniques yielded model performance at least as good as standard model configurations at double the resolution for an order of magnitude less computation time. The techniques developed here provide a structured approach towards flood risk assessment for engineers in data rich areas as well as proposing methodologies for data sparse regions.