Prediction of Scour Depth around Bridge Piers Using M5 Model Tree

Scour is natural phenomenon which happens dut to erosive action of flowing water in a river stream. It lowers the river bed by removal of sediments around the pier and ultimately exposing foundation of the structure around the bridge piers. Accurate prediction of maximum local scour is must for safe and economical design of the bridges. Many formulae are available in the literature and mostly are developed based on either laboratory data or field data or a combination of both. A number of equations have also been developed over the years to predict local scour depth using experimental and a few pier equations have formed using field data. Most of these equations are empirical in nature. This article presents an alternative in the terms of linear regression, simple regression and M5 model tree techniques to the conventional empirical equations in predicting the maximum depth of scour of bridge pier. The results obtained are tabulated in terms of statistical error i.e., correlation coefficient and rmse for 73 dataset of bridge pier scour published in literature. The outcome of this study suggest that the M5 model tree based modeling can be adopted as an alternate to linear regression equations, simple linear regression and the conventional empirical equations. Efforts were also made to estimate the maximum depth of scour in non dimensional form as well and results indicated the improvement to great extent and had shown better results. The sensitivity analysis on the data sets were also carried out to determine the relative significance of each of the independent input parameters on the estimation of maximum pier scour depth which indicated that diameter of the pier is the most important parameter affecting the scour.