Impact and Blast Resistance of Concrete Gravity Dams: Mechanisms, Response and Prediction

Le résumé fourni par la source

This chapter investigates the blast resistance of concrete gravity dams under air blast, contact blast, and underwater explosion (UNDEX) loading. An integrated framework is established by combining scaled model experiments, numerical simulations, theoretical fracture analysis, and physics-informed prediction. The similarity laws governing dynamic pressure, acceleration, and time are first verified through air blast tests and corroborated by Arbitrary Lagrangian-Eulerian simulations, supporting the extrapolation of model-scale responses to prototype dams. The chapter then compares damage evolution under different loading scenarios. Air blasts mainly produce localized surface damage, contact blasts generate scalable crest fracture patterns, and UNDEX loading induces strong spatially nonuniform responses controlled primarily by standoff distance and detonation depth. To explain these failure patterns, a theoretical damage surface model is developed from fluid-elastoplastic theory and fracture mechanics. Comparison with experimental fracture paths and curve-similarity metrics shows that mixed Mode I-II fracture governs the explosion-dominated region, whereas Mode I fracture contributes to the inertia-dominated segment. Finally, physics-informed neural networks (PINNs) are introduced for response prediction. By embedding a modified Euler-Bernoulli beam equation into neural network training, the PINN model improves the prediction of acceleration and displacement and generalizes better than a purely data-driven neural network under untrained blast conditions. The proposed framework links scaled experimentation, mechanism-based interpretation, and predictive modeling, providing a basis for blast-resistant design, damage assessment, vulnerability analysis, and emergency decision-making for critical dam infrastructure. It also clarifies how experimental scaling, numerical corroboration, and learning-based prediction can be combined to support resilient hydraulic infrastructure under extreme threats and hazards.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.