Prediction of the depth of local scouring at a bridge pier using a gene expression programming method

Abstract Local scouring around the piers of a bridge is one major reasons for failure, potentially resulting in heavy losses terms both economy and human life. Prediction accurate depth local difficult task due to many factors that contribute this process, however. The main aim study thus offer new formula prediction pier using modern fine computing modelling technique known as gene expression programming (GEP), with data obtained from numerical simulations used compare GEP performance standard non-linear regression (NLR) model. best then determined based on three statistical parameters: determination coefficient (R2), mean absolute error (MAE), root squared (RMSE). A total set 243 measurements, by simulation Flow-3D, intensity flow, ratio width, flow depth, Froude number, shape factor divided into training validation (testing) datasets achieve this. results suggest model provides better predicting compared conventional NLR model, R 2 = 0.901, MAE 0.111, RMSE 0.142. sensitivity analysis further has greatest impact scour other input parameters. gives predictor scouring, and, addition, offers special feature providing explicit compressed arithmetical allow calculation such scouring.