Estimation of maximum scour depth around bridge piers under ice-covered conditions using data-driven methods

The scouring around bridge foundations is a significant concern in civil engineering. Several research has been conducted experimentally and numerically to study the maximum scour depth around the foundations of a bridge in open channel conditions. In cold regions, where ice forms on lakes, reservoirs, and rivers, the interaction between ice and hydraulic structures is further complicated. The flow distribution varies significantly leading to deeper and larger scouring around bridge foundations. The present study took advantage of the data-driven methods to investigate the impacts of ice conditions to maximum scour depth. In total, 96 datasets were extracted from previous studies. The M5 decision tree, Genetic Expression Programming (GEP), and Particle Swarm Optimization (PSO) were used in data analysis. The models were developed using the train-and-test approach to avoid over-training. The mentioned developed models were further compared with a previous empirical equation associated with the prediction of maximum scour depth around piers under ice conditions. Finally, a sensitivity analysis method was applied to evaluate the performance of the data-driven models. It was found that the approaching velocity and pier diameter play the most crucial role in determination of the maximum scour depth. The result is generally in agreement with those of previous experimental studies.