The 12th International Conference on Scour and Erosion (ICSE 2025)

Accurately delineating total scour at bridge crossings is challenging due to the complex interactions between various scour components, such as scour caused by pier, abutment/embankment scour, and deck submergence. The superposition of these scour components is highly non-linear because of the intricate fluid-structure-sediment mechanism during scour evolution. The presence of scour countermeasures (e.g., riprap) and topographic complexity (e.g., compound channel) further exacerbates this complication, especially when general sediment transport affects the entire channel, preventing us from making accurate predictions. To address these challenges, we conducted systematic large-scale flume experiments and proposed new design approaches. Our findings focus on four key areas: (1) flow mechanism at the initial state of scour which indicates critical scour risks; (2) dynamic scour processes at bridge crossings under live-bed conditions, emphasizing the role of bedforms, bridge structures, and loose riprap stones in complicating scour evolution; (3) predicting temporal scour evolution; and (4) predicting equilibrium scour depth for various types of lateral and vertical flow contraction in compound channels. Particularly, the fourth area also addresses the predominant modes of time-dependent morphological alteration due to scour, offering a unified approach to assessing geotechnical and sediment loss risks. In addition to physical experiments, we have also applied advanced deep learning algorithms and developed multiple data-driven models to evaluate parametric sensitivity, predict scour depth, and extract, infer and reconstruct regional topography using sparse inputs. We expect the data-driven models to work in collaboration with empirical design equations to depict bridge scour scenarios in a more accurate and systematic way.

Bridge crossing; dynamic scour processes; multiple scour components; scour prediction models; deep learning in scour analysis