Numerical Exploration of the Shielding Effect of Vegetation on Nearshore Structures under Wave Run-Up Loading

• Published in: Journal of engineering mechanics 2024-08, Vol.150 (8)
• Main Authors: Moris, Joaquin P., Kyprioti, Aikaterini P., Irwin, Christopher H., Taflanidis, Alexandros A., Kennedy, Andrew B.
• Format: Article
• Language: English
• Subjects: Coastal structures; Computational fluid dynamics; Configurations; Errors; Free surfaces; Gaussian process; Human influences; Hurricanes; Prediction models; Risk assessment; Shear forces; Shielding; Shorelines; Statistical analysis; Statistical methods; Surface runoff; Technical Papers; Vegetation; Water pressure
• ISSN: 0733-9399; 1943-7889
• DOI: 10.1061/JENMDT.EMENG-7332

AbstractOverland flow from inundation events (IEs) such as hurricanes or tsunamis produce destructive loading conditions on nearshore structures. The magnitude of this loading and its associated impact can be substantially reduced by the existence of natural and/or anthropogenic obstacles, such as planted vegetation patches or neighboring structures. This paper examines the shielding effects offered by onshore coastal vegetation to nearshore structures, by establishing a numerical framework in combination with statistical computing methods to explore how vegetation properties, such as density and, more important, geometry relative to a row of shoreline infrastructure elements, impacts the resultant IE loads and subsequently the structural vulnerability. Computational fluid dynamics (CFD) simulations are first conducted to predict the intensity measures (IMs) of interest (momentum flux, free surface elevation, water pressure, and depth-averaged velocities, as well as base shear forces) on the shoreline structures for different excitation intensities (wave heights) and vegetation configurations. This information is used to develop a predictive model for the IMs using surrogate modeling techniques in order to expedite the succeeding risk assessment process. Gaussian process (GP) regression is specifically adopted as the surrogate modeling technique here, since it can accommodate noise in the training observations (CFD numerical errors) and quantify the associated predictive uncertainty (originating from the surrogate model) with minimal additional computational complexity. The spatial correlation between some of the examined IMs is explicitly addressed at the GP calibration stage. The developed predictive model is then used to estimate the structural vulnerability, with the surrogate model prediction errors explicitly considered when calculating the probability of exceedance for the different damage states, as warranted in such a vulnerability setting. Comparisons among different vegetation configurations, including the bare earth case, support an in-depth exploration of the impact of vegetation geometry on the shielding effects it can offer. Two case studies are examined, demonstrating applicability of the proposed framework for both system- and component-level vulnerability assessments.