Neural Network Formula for Local Scour at Piers Using Field Data

It is known that the scour at bridge piers is among the leading causes of bridge failure. However, the mechanism of flow around a pier structure is so complicated that it is difficult to establish a general empirical model to provide accurate estimation for scour. Each of the proposed empirical formulas yields good results for a particular data set, but cannot accurately or reliably predict scour from various data sets. In this study, an alternative approach, an artificial neural network (ANN) model, is proposed to estimate the local scour depth with numerous field data bases. The local scour depth was modeled as a function of six non-dimensional variables: V-2/gb, y/b, b/D-50, l/b, pier skew angle, and pier shape. Four hundred and ten field data points were used for the training and testing of the ANN model. The predicted results showed that the neural network could provide better performance than the empirical equations.