Computing flooding of crossroads with obstacles using a 2D numerical model

Abstract

Typical urban flood flow features are usually computed using two-dimensional numerical models. How such modelling can be implemented in dense urban areas with obstacles is investigated. A strategy for representing the effect of urban obstacles in various flow conditions is defined. The comparison between the available laboratory measurements and the model results show that if the water depth is high enough and the flow remains subcritical, two-dimensional modelling with constant eddy viscosity provides the effect of the obstacles on the flow distribution accurately, even with a coarse mesh; moreover, an oversimplified representation of the sidewalks averaging the street cross section elevations seems sufficient. Oppositely, if the water depth is low and/or the flow becomes supercritical, the description of the flow is not relevant enough and it generates errors in the flow distribution at the crossroads.

Numerical Insights Into the Effects of Model Geometric Distortion in Laboratory Experiments of Urban Flooding

“…In contrast, for some of the more generic configurations of flow divisions and junctions, the performance of the 2‐D model was also analyzed in terms of extent of computed recirculation cells and velocity fields. While Bazin et al (2017), Shettar and Murthy (1996), and Khan et al (2000) all report a fair to good agreement between computed and observed velocity profiles, only Shettar and Murthy (1996) state that the computed length and width of recirculation cells showed an “acceptable” agreement with the observations. Contrarily, Bazin et al (2017) highlight that the 2‐D model failed to predict accurately the extent of the flow recirculations.…”

“…While Bazin et al (2017), Shettar and Murthy (1996), and Khan et al (2000) all report a fair to good agreement between computed and observed velocity profiles, only Shettar and Murthy (1996) state that the computed length and width of recirculation cells showed an “acceptable” agreement with the observations. Contrarily, Bazin et al (2017) highlight that the 2‐D model failed to predict accurately the extent of the flow recirculations. Several studies emphasize the sensitivity of the computed recirculation widths and lengths to the particular parametrization used for turbulence closure (Arrault et al, 2016; Bazin et al, 2017; Bruwier et al, 2017).…”

“…Contrarily, Bazin et al (2017) highlight that the 2-D model failed to predict accurately the extent of the flow recirculations. Several studies emphasize the sensitivity of the computed recirculation widths and lengths to the particular parametrization used for turbulence closure (Arrault et al, 2016;Bazin et al, 2017;Bruwier et al, 2017). These results suggest that, although depth-averaged 2-D computational models have been very successful in predicting flow depths, discharge partition, and, to some extent, velocity profiles, in configurations similar to that of interest here, their ability to accurately predict the extent of recirculation cells and, hence, the overall effect of form drag and local head losses, as well as their relative importance, may still be questioned.…”

“…Depth-averaged 2-D computational models have proved valid in many instances, but limitations in their validity remain when it comes to predicting flows significantly influenced by form drag and recirculations. Table S6 in the supporting information presents a selection of studies, which assessed the ability of 2-D models to deliver accurate predictions of various flow variables in configurations similar to those encountered in urban flooding: flow in a synthetic urban district (Arrault et al, 2016), as well as subcritical flow divisions (Bazin et al, 2017;Bruwier et al, 2017;Khan et al, 2000;Shettar & Murthy, 1996) and junctions in 90°s harp-edged rectangular open channels of equal width (El Kadi Abderrezzak & Paquier, 2009;Khan et al, 2000).…”

Experimental modelling of urban flooding: A review

“…One main deficiency regarding surface flow corresponds to the consideration of obstacles present in the streets (Mignot et al, 2013) and steep urban areas where mostly supercritical flow conditions take place. In such cases, hydraulic jumps occur at the street intersections (Mignot et al, 2008) and in the vicinity of obstacles (Bazin et al, 2017).…”