2D and 3D Computational Modeling of Surface Flooding in Urbanized Floodplains: Modeling Performance for Various Building Layouts

Understanding the strengths and limitations of the modeling capacity of surface flooding in urbanized floodplains is of utmost importance as such events are becoming increasingly frequent and extreme. In this study, we assess two computational models against laboratory observations of surface urban flooding in a reduced-scale physical model of idealized urban districts. Four urban layouts were considered, involving each three inlets and three outlets as well as a combination of three- and four-branch crossroads together with open spaces. The first model (2D) solves the shallow-water equations while the second one (3D) solves the Reynolds-averaged Navier-Stokes equations. Both models accurately predict the flow depths in the inlet branches. For the discharge partition between the outlets, deviations between the computations and laboratory observations remain close to the experimental uncertainties (maximum 2.5 percent-points). The velocity fields computed in 3D generally match the measured surface velocity fields. In urban layouts involving mostly a network of streets, the depth-averaged velocity fields computed by the 2D model agree remarkably well with those of the 3D model, with differences not exceeding 10%, despite the presence of helicoidal flow (revealed by the 3D computations). In configurations with large open areas, the 3D model captures generally well the trajectory and velocity distribution of main surface flow jet and recirculations; but the 2D model does not perform as well as it does in relatively channelized flow regions. Visual inspection of the jet trajectories computed by the 2D model in large open areas reveals that they substantially deviate from the observations. Advancing our modeling capacity of urban flooding is of utmost importance for improving the design of risk reduction measures. During extreme urban flooding, complex flow patterns develop in urban environments, involving three-dimensional flow structures. Though, urban floods are commonly simulated with two-dimensional computational models. So far, no detailed comparison between flow fields predicted by two- and three-dimensional computational models were conducted and assessed against reference data such as experimental observations for representative configurations of urban flooding. In this study, we assess two computational models against laboratory observations of urban flooding in a reduced-scale physical model of an idealized district. Predictions of 2D and 3D computational models were compared against laboratory experiments representing urban flooding in a steady-state Both models perform equally well to predict upstream flow depth, outlet discharge partition, and velocity field in street networks In urban layouts with large open spaces, only the 3D model accurately predicts the velocity field