Prediction of flow and its resistance in compound open channel

Abstract

Flooding situation in rivers is a complex phenomenon and affects the livelihood and economic condition of the region. This complex condition has long been identified and intensive research work has been carried out to find out its remedial solution. During flood, river overtops its banks and spreads to flood plains, called compound channel. It has been observed that the flow velocity in flood plain subsection is slower than the velocity of its course. Due to this difference in velocities between main
channel and flood plain, a large shear layer produces.This large shear layer retains complex turbulent structures of different scales within it. These turbulent structures produce extra resistance to flow, which induces uncertainty in predictions of flow and its resistance. It has been also observed that, numerical turbulence models can be used to find the point to point information. Hence the analysis of turbulent structures is prevalent in this situation. Earlier, researchers have adopted various numerical, analytical and empirical models to analyze turbulent flow in compound channels, generally for low development length. Therefore, in this study an effort is made to analyze the turbulent structure by Large Eddy
Simulation method (LES) to predict the flow and its resistance involved in it. The LES is carried out taking sufficient development length so that uniform turbulent flow is developed. The development length is incorporated in the computational domain. However, it is a fact that numerical simulation of compound channels with different hydraulic conditions are computationally very expensive and arduous. Therefore,this analysis is further done by using adaptive approaches such as Artificial Neural Network (ANN) and Artificial Neuro Fuzzy Inference System (ANFIS).These models are used to predict flow and its resistanceof a compound channel for different hydraulic conditions.All the proposed models are compared well with the standard modes previously developed.The proposed models are found to give better results compared to other models when applied to globa data sets.