ABSTRACT
Flows in compound open-channels present important turbulent shear stresses at the transition region between the main channel and the floodplain. An accurate prediction of the eddy viscosity is thus necessary in modeling of those flows. In this paper, the small-scale variability of the eddy viscosity in compound open-channels was investigated using a Smagorinsky turbulence closure model. The calculation results from two laboratory experiments showed that the Smagorinsky turbulence model reproduced very well the velocity profiles in compound channel sections, especially the non-uniform velocity in the floodplain and in the transition region between floodplain and main channel. The computed values of eddy viscosity from this model agreed reasonably with experimental values obtained by calculating the depth-averaged Reynolds’ stress and the lateral gradient of longitudinal velocity from the measurements. The eddy viscosity varies significantly in the channel section, i.e. the small values often occurred around the central part of the floodplains and the main channel while large values appeared in the transition regions. On the other hand, four other classical models, i.e. a constant value, zero-equation, one-equation, and two-equation turbulence models were also additionally considered. It was shown that the use of complicating models such as one- and two-equations did not provide any significant improvement of flow depth and velocity compared to the Smagorinsky turbulence closure and other models.
