Characterization of the Turbulent Structure in Compound Channel Flows

Abstract

The main goal of this investigation is to characterize the turbulent structures in compound channel flows considering two geometrical conditions, a simple asymmetric compound channel and the same compound channel but with the placement of rods on the upper bank. For the simple asymmetric compound channel, three different water depths were analyzed, one corresponding to deep flows and two corresponding to shallow flows. For the compound channel with rods, three different spacing between elements were studied, including two different water depths for each spacing condition. The measurements were taken with a 2D Laser Doppler Velocimiter, at 9.0 m from the inlet for simple compound channel. For the compound channel with rods, three cross-section around 9.0 m from the inlet of the channel were measured, corresponding to locations downstream of the rod, in the middle of two rods and upstream of the rod. The measurements were performed under quasi-uniform flow condition and streamwise and vertical instantaneous velocity components were obtained. The raw data was filtered and processed in order to estimate the time-averaged velocities U and W, the turbulent intensities U' and W', Reynolds stress u'w' , streamwise integral length scale Lx, turbulence dissipation rate ε and Taylor’s micro scale  x. Taylor's frozen field hypothesis was adopted in order to transform the time record into a space record, using a convection velocity Uc. The autocorrelation function was built and the integral length scale estimated using three different stop methods of the integral: the second zero of the autocorrelation function, the first minimum, and assuming the integral length scale as the wavenumber value when the autocorrelation function reaches 1/e (this was concluded to be the most consistent method). For estimating the dissipation rate, the following methods were used: from the third order structure function, from the second order structure function and finally, and from the energy spectrum of the velocity (this was concluded to be the most consistent method). In the case of a simple compound channel, the deep flows are characterized by macro vortices with streamwise axis between the interface and main channel and between interface and floodplain, having a notable separation in the "main channel vortex" and the "floodplain vortex" meeting, due the double shear layer of the streamwise depth-averaged velocity. Shallow flows are characterized by macro vortices with vertical axis confined between the interface and main channel and originated by the depth average velocity gradient between the main channel and floodplain. A clear linear relation exists between the streamwise integral length scale, Lx, the dissipation rate, ε, and the streamwise turbulent intensity U’. Contrary to 2D fully developed open-channel flow equations that relation appears to be constant for all water depths. For the compound channel with rods, new turbulent structures are generated due the interaction between rods and flow. Downstream of rods, the horseshoes-vortex system is perfectly observed and a strong descendant flow dominate both sides of rod, turning invalid the universal laws for 2D fully developed open-channel flows. The integral length scale presents almost constant values in the vertical direction, which indicates that the wakes generated by the rods influence the entire water column. The turbulent microscale and dissipation rate acquire a streamwise variation due to the vortex propagation in the downstream direction, both presenting higher values than the ones corresponding to 2D flows.