Numerical modelling of impact pressure and load characteristics of tidal bores on inclined piles and their rapid prediction using convolutional neural network method

Tidal bores are a type of highly nonlinear discontinuous flow that carry tremendous energy. As such, pile structures such as bridge piers located in bore-affected estuarine zones are subjected to significant hydrodynamic impact forces, which may lead to severe structural damage. To safeguard human life and property, it is crucial to assess the impact loads of tidal bores in advance and implement effective protective measures. In this study, a high-resolution numerical tidal bore flume was established based on the open-source solvers OpenFOAM and wave2Foam, enabling the accurate simulation of tidal bore propagation and its interaction with pile structures. The hydrodynamic response and structural behavior of both vertical and obliquely inclined piles under tidal bore impact were systematically investigated, focusing on tidal bore evolution, local flow fields, hydrodynamic loads, and dynamic pressure. The influence of bore type (undular, weak breaking, and breaking), tidal bore height, initial water depth, inclination angle ( α ), and incident direction ( β ) on the hydrodynamic characteristics near pile structures was analyzed in detail. Furthermore, based on numerical simulation results, this study developed empirical formulas for predicting the maximum hydrodynamic load and dynamic pressure exerted on pile structures under different tidal bore conditions, incorporating the effects of inclination and incident angles. In addition, a one-dimensional convolutional neural network (1D-CNN) model was constructed to predict the hydrodynamic evolution of tidal bore impacts. Using a dataset of water level time series collected at multiple monitoring points as input, the model effectively predicted the water run-up on pile surfaces and the corresponding peak hydrodynamic loads. Results show that as the pile inclination angle α increases from −12.5 ° to 12.5 ° , both the maximum load and peak pressure decrease, with larger changes under stronger bores. As the incident direction β shifts from 180 ° to 0 ° , water run-up increases, and piles with β > 90 ° experience higher peak loads. Overall, the load increases as β moves from 0 ° to 180 ° , especially under stronger tidal bores. The proposed 1D-CNN model shows strong predictive capability and robustness, capturing the impact patterns across various angles and directions. • A high-precision numerical tidal bore flume was developed in OpenFOAM by redefining boundary conditions based on the one-dimensional shallow-water propagation theory, enabling accurate reproduction of bore generation, evolution, and breaking near structures. • Systematic simulations quantified the effects of bore type, bore height, pre-bore water depth, pile inclination, and incident direction on hydrodynamic loads and pressure distributions, from which new semi-theoretical formulas for maximum load and impact pressure were derived. • A one-dimensional convolutional neural network (1D-CNN) model was established using multi-point water level sequences to rapidly predict pile run-up and peak hydrodynamic loads, achieving high accuracy and computational efficiency. • The integration of high-fidelity CFD modeling and deep learning prediction provides a unified framework for tidal bore impact analysis, overcoming the limitations of empirical methods and advancing intelligent assessment of coastal structural safety.