Author(s): Siva Krishna Reddy; Venu Chandra

Linked Author(s):

Keywords: Pier scour; Pier shape effect; Non-uniform piers; Scour depth prediction; Artificial neural networks

Abstract: Local scour around bridge piers is a significant factor that affects the stability of a bridge structure. Accurate pier scour depth estimation is important for safe and economical bridge design, which requires understanding the complex three-dimensional flow field around a pier and associated sediment transport. Scour depth around a pier is related to the flow, sediment, and pier characteristics. Several empirical equations are available in the literature to calculate the design scour depth, but accurate estimation is not possible due to the complexity of the scouring process. Artificial Neural Networks (ANN) can be an alternative tool for modeling the scour process and estimating scour depth. This study investigates the application of ANNs to estimate the maximum scour depth around different-shaped bridge piers (uniform and non-uniform piers). The feed-forward back-propagation neural network is used to develop the scour models. Dimensionless independent variables are supplied to the ANN models to obtain the maximum scour depth at different-shaped piers. For uniform piers, the RMSE values are 0.172 and 0.205 during model calibration and validation, respectively, and the corresponding R2 values are 0.888 and 0.834. Further, an ANN model is developed to predict the scour depth at non-uniform piers. The RMSE values are 0.333 and 0.625 during calibration and validation, respectively, and the corresponding R2 values are 0.973 and 0.897. About 95% of the datasets are confined within the margin of ± 25% error. It is concluded that the ANN can predict the local scour depth accurately for different-shaped bridge piers.

DOI: https://doi.org/10.1007/978-981-97-6009-1_35

Year: 2022