SUMMARY
Flow modeling in a compound channel is a complex matter. Indeed, due to the smaller velocities in the floodplains than in the main channel, shear layers develop at the interfaces between two stage channels, and a momentum transfer corresponding to this shear layer affects the channel conveyance. Since a compound channel is characterized by a deep main channel flanked by relatively shallow flood plains, the interaction between the faster fluid velocities in the main channel and the slower moving flow on the floodplains causes shear stresses at their interface which significantly distort flow and boundary shear stress patterns. The distortion implies that flow field in rivers is highly non homogeneous turbulent, which lateral transport of fluid momentum and suspended sediment are influenced by the characteristics of flow in rivers. The nature of mechanism of lateral transport needs to be understood for the design of river engineering schemes that rely upon realistic flow. Furthermore, the flows in river are also almost turbulent. This means that the fluid motion is highly random, unsteady, and three -dimensional. Due to these complexities, the flow cannot be properly predicted by using approximate analytical solutions to the governing equations of motion. With the complexity of the problems, the solution of turbulent is simplified with mathematics equation. The momentum transfer due to turbulent exchanges is then studied experimentally and numerically. Experimental data is obtained by using ElectroMagnetic Velocimetry and Wave Height Gauge. The Large Eddy Simulation Sub Depth Scale (LES SDS)-2 Dimensional Horizontal (2DH) Model is used to solve the turbulent problem. Successive Over Relaxation (SOR) method is employed to solve the numerical computation based ob finite difference discretization. The model has been applied to the compound channel with smooth roughness. Some organized large eddies were found in the boundary between main channel and flood channel. At this boundary the transverse velocity profile exhibits a steep gradient, which induces significant mass and momentum exchange, acts as a source of vorticity, and generates high Reynolds stresses. The Large Eddy Simulation SDS-2DH model enables to predict quite successfully the wavelength of some observed vortices. The estimated vortex wavelengths agree again with the measurements and the theoretical predictions. The present model is proven to be a useful tool for engineering applications, as it can simulate the dynamic development of large eddies.