Probabilistic models and reliability analysis of scour depth around bridge piers

Abstract

Scour at a bridge pier is the formation of a hole around the pier due to the erosion of soil by flowing water; this hole in the soil reduces the carrying capacity of the foundation and the pier. Excessive scour can cause a bridge pier to fail without warning. Current predictions of the depth of the scour hole around a bridge pier are based on deterministic models. This paper considers two alternative deterministic models to predict scour depth. For each deterministic model, a corresponding probabilistic model is constructed using a Bayesian statistical approach and available field and experimental data. The developed probabilistic models account for the estimate bias in the deterministic models and for the model uncertainty. Parameters from both prediction models are compared to determine their accuracy. The developed probabilistic models are used to estimate the probability of exceedance of scour depth around bridge piers. The method is demonstrated on an example bridge pier. The values of the model parameters suggest that the maximum sour depth predicted by the deterministic HEC-18 Sand and HEC-18 Clay models tend to be conservative. Evidence is also found that the applicability of the HEC-18 Clay method is not limited to clay but can also be used for other soil types. The main advantage of the HEC-18 Clay method with respect to the HEC-18 Sand method is that it predicts the depth of scour as a function of time and can be used to estimate the final scour at the end of the design life of a structure. The paper addresses model uncertainties for given hydrologic variables. Hydrologic uncertainties have been presented in a separate paper.