Simulation of open channel bend characteristics using computational fluid dynamics and artificial neural networks

Abstract

The ability to simulate flow characteristics is one of the most important issues in the design and application of open channel bends. Three-dimensional computational fluid dynamics (CFD) and multi-layer feed-forward artificial neural networks (MLFF-ANNs) are used and compared in modeling the flow depth and velocity field in sharp bends. CFD is modeled in two phases, water and air, using the volume of fluid method. The backpropagation algorithm is applied in the training process of the ANN model. An experimental study of a 90° curved channel is undertaken to verify and compare the efficiency of the CFD and ANN models. The results show that both CFD and ANN methods can be successfully applied to the modeling of open channel bend characteristics. However, the ANN method performs significantly better than CFD.