3D numerical modelling of flow exchanges between flooded streets and residential blocks during urban flooding in an idealized district

Flow exchanges between flooded streets and the interior of buildings through broken doors and windows can have a significant impact on the flow characteristics and the flood hazard in urban floods. Traditionally, urban floods are modelled with 2D shallow-water equations, however, this modelling approach is not always capable to reproduce such complicated flow exchanges. This study aims to investigate the added value of 3D non-hydrostatic numerical modelling over the modelling with 2D shallow-water equations for the accurate prediction of flow exchanges between flooded streets and the interior of buildings. Systematic comparisons were carried out between the results of 2D depth-averaged and 3D numerical modelling against laboratory observations for eight configurations of an urban block surrounded by flooded streets. Each urban block configuration had a different number and locations of openings in the block perimeter. The 3D model predicted the flow discharge partition at the street outlets more accurately than the 2D model, with an average RMSE of about 1 percentage point compared to the measurements. The flow in the interior of the urban block was typically characterized by large recirculation zones, which were successfully reproduced by the 3D model in terms of number, locations and directionality of recirculating flow, with only limited discrepancies in the sizes and shapes for some cases. The superiority of the 3D model over the 2D was particularly evident in the case where the streets meet a large open space. In addition, 3D models perform better than 2D models in cases with high vertical velocities. The vertical velocities are prominent in areas where flow jets and flow contractions are observed near the openings of the urban blocks, and can reach values up to 20% of the total depth-averaged velocities. The vertical variation of the flow pattern is rather limited from the bottom to the surface, but the difference between the magnitude of the surface velocity and the corresponding depth-averaged velocity reaches 50% in areas of complex flow patterns near the openings of the urban block. From an engineering perspective, this study informs practitioners when to opt for the more demanding 3D modelling instead of the traditional modelling with 2D shallow-water equations in the context of flow exchanges between flooded streets and the interior of buildings during urban floods.