Explicit Neural Network-derived formula for overtopping flow on mound breakwaters in depth-limited breaking wave conditions

Abstract

Sea level rise due to climate change, as well as social pressure to decrease the visual impact of coastal structures, have led to reduced crest freeboards, and this increases the overtopping hazard. In previous studies, pedestrian safety during overtopping events was assessed considering the overtopping layer thickness (OLT) and the overtopping flow velocity (OFV). This study analyzed the statistics of OLT and OFV on mound breakwaters without crown walls during severe wave storms. Small-scale 2D physical tests were conducted on mound breakwaters with dimensionless crest freeboards between 0.29 and 1.77, testing three armor layers (single-layer Cubipod®, and double-layer cubes and rocks) in depth-limited breaking wave conditions and with two bottom slopes. Neural Networks were used to develop new estimators for the OLT and OFV exceeded by 2% of the incoming waves with a high coefficient of determination (0.866 ≤ R2 ≤ 0.876). The best number of significant figures in the empirical coefficients of the new estimators was determined according to their variability. The 1-parameter Exponential and Rayleigh distribution functions were proposed to estimate the extreme values of OLT and OFV with 0.803 ≤ R2 ≤ 0.812, respectively.