Prediction of the scour depth of bridge pier using artificial neural network model and comparison with empirical equations

Document Type : Original Article

Authors

1 Department of Water and Environmental Engineering, Faculty of Civil Engineering, Shahrood University of Technology, Shahrood, Iran.

2 M. Sc student of Water structures, Department of Water and Soil, Faculty of Agriculture, Shahrood University of Technology, Shahrood, Iran.

Abstract

The impact of flow on the bridge piers creates turbulent flows that cause scour around it, and as result collapse of the bridges. Despite to most research which done in this area, but due to the complexity of the phenomena and the many parameters involved in the phenomenon, there is still no exact relationship or fundamental solution to predict scour depth. In the present study, using the measured data, the scour depth was estimated with data mining methods such as artificial neural network, linear and nonlinear regression models and also empirical relationships. The data were used in two ways, with dimensions and non-dimension, which were obtained using dimensional analysis. The results showed that the artificial neural network model able to predict scour depth with determination coefficient (R2) equal to 0.97 and 0.81, as well as RMSE error equal to 0.06 m and 0.32, respectively, when data was used with dimension and non-dimension forms. Also, the empirical equations of the Colorado State University between the empirical relationships predicted scour depth with R2 and RMSE error equal to 0.84 and 0.52. Comparison of the results of different models shows that the best results are related to the artificial neural network model and it decreased error of prediction 70, 85.5 and 87.7% compared to linear regression model, nonlinear regression model and the empirical equation of the Colorado State University, respectively.

Keywords

Main Subjects

Advanced Technologies in Water Efficiency
Volume 1, Issue 1
December 2022
Pages 70-90

Files

History

  • Receive Date: 29 August 2021
  • Revise Date: 02 October 2021
  • Accept Date: 19 October 2021

Share

How to cite

Statistics