Chapter 5 - Soft-computing approach to scour depth prediction under wall jets

Abstract

Presented in this chapter is an evaluation of laboratory experimental data relating to local scour under the action of two-dimensional jets using soft-computing approach. Performance of the existing scour depth prediction equations is also evaluated by applying them to the existing data set. The existing prediction equations do not perform well when adjudged with experimental data. Hence, adaptive neuro-fuzzy interference system (ANFIS) and artificial neural network (ANN)-based prediction models are developed utilizing published data set, which are considered as reliable and precise methods for the prediction of equilibrium depth of scour. The main factors affecting the equilibrium depth of scour are jet Froude number, tailwater level, length of the apron, and median size of sediment. Statistical parameters are utilized to compare the outcome achieved from ANFIS and ANN models with that of the existing equations. The statistical outcome of ANFIS ( RMSE = 0.013) and ANN ( RMSE = 0.003) models is reasonably better than that of the existing equations.