Calculation of Friction velocity in Straight Rectangular Channel in Unsteady flow condition Using Artificial Neural Network

Abstract

Generally, the flow in natural rivers is unsteady. Accurate assessment of various flow properties like friction velocity and bed shear stress in an open channel flow under unsteady condition is of crucial importance to hydraulic engineers since it helps in estimation of erosion, sediment transport etc. Bed shear stress can be properly predicted by accurate calculation of the friction velocity which is generally influenced by the geometry, roughness, and hydraulic parameters of the channel. In the past, very few studies have been carried out to calculate the bed shear stress in unsteady open channel flows. This study proposes an artificial neural network (ANN) model for the prediction of bed shear stress in straight rectangular channels in unsteady flow condition for both rising and falling limb of hydrograph. The most influential parameters such as depth of flow, discharge, rising and falling time of hydrograph, bed slope of the channel, and roughness condition are considered as input parameters. Vast amount of experimental data from previous researches comprising of input parameters have been used for both training and validation of the model. The models utilised here are back-propagation neural network (BPNN) models, which can perform well for broad ranges of independent parameters. A statistical error analysis employing large data sets is used to confirm the efficacy of the models. The result shows that the ANN network is giving R2 value of 0.9186 for rising limb and 0.9334 for falling limb of hydrograph.