Percorrer por autor "Baddavidana, Dinush Shayaman Priyankera"
A mostrar 1 - 1 de 1
Resultados por página
Opções de ordenação
- Assessment of wave overtopping at Praia da Vitória, Terceira, Azores, with SWASH modelPublication . Baddavidana, Dinush Shayaman Priyankera; Pinheiro, Liliana; Garzon, Juan L.; Zózimo, Ana CatarinaThe primary objective of this dissertation is to enhance the accuracy of overtopping predictions using the SWASH (Simulating WAves till SHore) model, for inclusion in the Early Warning System (EWS), HIDRALERTA, currently operational in Praia da Vitória. While existing tools within HIDRALERTA offer valuable insights, they have limitations in accurately representing waves and overtopping details. To address these limitations, this research tested the capabilities of the SWASH model by simulating past storm events and typical wave conditions across the two-dimensional model of the entire port and bay areas with complex coastal structures and bathymetry. Outcomes of the simulations were compared to predictions from the NN_OVERTOPPING2 neural network and observed images from extreme events. In the SWASH model, coastal structures were modelled as impermeable layers, with bottom roughness incorporated to enhance energy dissipation from roughness and seepage. An unstructured triangular mesh was used for the computational domain. Manning’s coefficient for the outer slope of the south breakwater, with tetrapod armour units, was calibrated using a physical model test conducted in February 2025 at the National Laboratory of Civil Engineering (LNEC). The model was replicated in SWASH at prototype scale, using two test cases that resulted in overtopping for calibration. The results demonstrate that the SWASH model effectively simulates wave propagation and overtopping in harbour and bay areas, especially over complex coastal structures. Its accuracy in estimating wave propagation and overtopping is sensitive to wave energy interactions with the bottom and slopes, influenced by the Manning coefficient. While unstructured meshes enable large-domain simulations, the quality of the mesh affects both prediction accuracy and computational time. The model is stable when using implicit time integration but unstable with explicit methods. Despite some simulation instability, most simulations reached the maximum allowed simulation time on the National Distributed Computing Infrastructure (INCD) cluster. The model’s accuracy is heavily dependent on bathymetric data, with results generally aligning with NN_OVERTOPPING2 predictions and observed images from extreme events, although underpredictions can occur. Parallel processing is currently limited to structured grids in SWASH and, there is potential for efficiency improvements with unstructured meshes. Future studies should compare SWASH results with field data to fully assess its performance.