Recently, many countries have investigated replacing conventional visual inspection with computer-vision-based inspection to enhance the efficiency, speed, and objectivity of inspection. This paper presents a novel crack assessment framework for concrete structures that detects cracks using mask and region-based convolutional neural network (Mask R-CNN) and quantifies cracks using a few morphological operations on the detected crack masks. In this study, a Mask R-CNN is trained for crack detection using 1,102 crack regions masked on 376 concrete images. The trained Mask R-CNN model is tested on the images taken from a real concrete wall with 453 cracks whose widths range from less than 0.1 mm to 1.0 mm. The trained model successfully detects most of the cracks 0.3 mm or wider. Quantification of the cracks was then carried out using several image-processing operations on 10 randomly selected crack masks. Cracks with widths of 0.3 mm or more are quantified successfully with errors less than 0.1 mm, whereas cracks less than 0.3 mm widths show relatively larger error due to the limitation of image resolution.
a b s t r a c tThe effect of mesh type on the accuracy and computational demands of a two-dimensional Godunov-type flood inundation model is critically examined. Cartesian grids, constrained and unconstrained triangular grids, constrained quadrilateral grids, and mixed meshes are considered, with and without local time stepping (LTS), to determine the approach that maximizes computational efficiency defined as accuracy relative to computational effort. A mixed-mesh numerical scheme is introduced so all grids are processed by the same solver. Analysis focuses on a wide range of dam-break type test cases, where Godunov-type flood models have proven very successful. Results show that different mesh types excel under different circumstances. Cartesian grids are 2-3 times more efficient with relatively simple terrain features such as rectilinear channels that call for a uniform grid resolution, while unstructured grids are about twice as efficient in complex domains with irregular terrain features that call for localized refinements. The superior efficiency of locally refined, unstructured grids in complex terrain is attributable to LTS; the locally refined unstructured grid becomes less efficient using global time stepping. These results point to mesh-type tradeoffs that should be considered in flood modeling applications. A mixed mesh model formulation with LTS is recommended as a general purpose solver because the mesh type can be adapted to maximize computational efficiency.
Keywords:Porous shallow water equations Finite volume model Anisotropic porosity Dam-break flood Urban flood a b s t r a c t Porous shallow-water models (porosity models) simulate urban flood flows orders of magnitude faster than classical shallow-water models due to a relatively coarse grid and large time step, enabling flood hazard mapping over far greater spatial extents than is possible with classical shallow-water models. Here the errors of both isotropic and anisotropic porosity models are examined in the presence of anisotropic porosity, i.e., unevenly spaced obstacles in the cross-flow and along-flow directions, which is common in practical applications. We show that porosity models are affected by three types of errors: (a) structural model error associated with limitations of the shallow-water equations, (b) scale errors associated with use of a relatively coarse grid, and (c) porosity model errors associated with the formulation of the porosity equations to account for sub-grid scale obstructions. Results from a unique laboratory test case with strong anisotropy indicate that porosity model errors are smaller than structural model errors, and that porosity model errors in both depth and velocity are substantially smaller for anisotropic versus isotropic porosity models. Test case results also show that the anisotropic porosity model is equally accurate as classical shallow-water models when compared directly to gage measurements, while the isotropic model is less accurate. Further, results show the anisotropic porosity model resolves flow variability at smaller spatial scales than the isotropic model because the latter is restricted by the assumption of a Representative Elemental Volume (REV) which is considerably larger than the size of obstructions. These results point to anisotropic porosity models as being well-suited to whole-city urban flood prediction, but also reveal that point-scale flow attributes relevant to flood risk such as localized wakes and wave reflections from flow obstructions may not be resolved.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.