Abstract

Bridge failures due to scour occur more frequently than expected, although bridge design considers conservative scenarios, evidencing a stochastic scour behaviour and the need for improvements in current design procedures. In this paper, a new method for determination of design local scour depth at bridge piers is proposed. The results show that peaks-over-threshold series are better suited than annual-maximum series of maximum scour depths for frequency analysis. Extreme scour depths follow the heavy-tailed Cauchy distribution. Scour depths with a 100-year return period resulted in 22 to 202% of the equilibrium scour depth currently used in design, evidencing the need for a different criterion to select the design scour depth. The proposed method computes the design local scour depth based on its probability of occurrence, introducing a novel probabilistic approach in bridge design for enhanced infrastructure risk management compatible with multi-hazard analysis.