Author(s): Ilhan Ozgen; Dongfang Liang; Reinhard Hinkelmann
Keywords: Grid convergence; Isolated obstacle; Mesh refinement techniques; Porosity shallow water model
Abstract: The integral porosity shallow water model (IP) is a modified form of the depth-averaged shallow water flow model, which utilizesporosity terms to account for unresolved sub-grid-scale topography such as buildings to enable fast urban flooding simulations. Existing research has repeatedly pointed out that the IP model is inherently oversensitive to the mesh design. This paper presents a detailed grid convergence study of the IP model for simulating a laboratory experiment on the interaction between a dam-break wave and an obstacle in a channel, which is featured by the highly complex non-hydrostatic flow with a backwards-propagating hydraulic jump. We compare three different mesh refinement techniques with up to six levels of refinement: (1) uniform, (2) manual, (3) locally coarsened. For this investigated case, the modeling error due to the shallow water assumptions is more significant than that due to the porosity treatment. Neither a conventional shallow water model, nor the integral porosity model is able to predict the measured data well owning to non-hydrostatic flow conditions and a backwards propagating hydraulic jump. We show that the integral porosity model results converge to the conventional shallow water model results at locations that are not affected by these non-hydrostatic flow conditions. We conclude that, when the obstacle density is low, high-frequency oscillations may appear in the domain owing to K'arm'an vortex shedding. These cannot be captured accurately by the integral porosity shallow water model, unless high resolutions similar to those in the conventional shallow water models are used. However, the benefit of the porosity model is lost by using high resolutions.