Action Filename Description Size Access License Resource Version
Show more files...


The study of turbulent flows has always been a challenge for scientists. Turbulent flows are common in nature and have an important role in several geophysical processes related to a variety of phenomena such as river morphology, landscape modeling, atmospheric dynamics and ocean currents. At present, new measurement and observation techniques suitable for fieldwork can be combined with laboratory and theoretical work in order to advance in the understanding of river processes. In this Ph.D. dissertation, an Acoustic Doppler Velocity Profiler (ADVP) suspended from a deployable structure allowed the investigation of turbulent gravel-bed river flows. The ADVP, which was developed by the Laboratoire d'Hydraulique Environnementale (LHE), permits to obtain over the entire water depth, three-dimensional quasi-instantaneous information on the fluctuating velocity flow field in the production and inertial subranges of the spectral space. Improvements on the ADVP data quality were made with the implementation of a correction methodology for errors due to data aliasing. This reduced the range-velocity ambiguity in Doppler-based instruments. The results presented in this dissertation contribute to the understanding of transport and mixing processes in river flows. They are based on three sets of field measurements made in gravel-bed rivers with blockage ratios of h/D84≈3.0 and aspect ratios B/h between 23 and 32. The measurements were made during low-water periods. The fieldwork provided results on the mean and instantaneous velocity field. The flow was divided into three inviscid vertical layers with different mean field and Reynolds stress characteristics: the roughness layer, the blending or intermediate layer and the surface layer. In the lower layers of the flow three types of mean velocity profile were found: mono-logarithmic, s-shaped due to bed perturbations and double-logarithmic downstream bed perturbations. The determination of the shear stress distribution for each of these profile types is studied. In double-log profiles, the friction velocity and roughness length determined for the outer logarithmic layer are required for the velocity profile parameterization. The s-shaped profiles are described by a tangent-hyperbolic function in the lower layers compatible with an external log layer. Limitations of 2D open-channel theories to parameterize the velocity distribution and to characterize the bottom drag are discussed. Bottom drag occurs in the predominant momentum direction. The direction changes as a function of the local bed forms. To estimate bottom drag one has to consider the actual momentum transport direction which varies with the flow depth. The wall effect of the riverbank is visible until y/h≈5. Bottom topography produces important secondary mean motion in the flow. A permanent structure of the flow was described in the upper layers, near the surface (z/h<0.80): the Surface Layer Organized Movement – SLOM. It is composed by local jets (CH regions) and by lower velocity regions (CL regions) associated with a compensatory secondary motion with streamwise vorticity. Lateral momentum transfer exists between adjacent CL and CH regions. In one river, the bed form presented signatures of possible streamwise ridges. All the turbulent characteristics of the flow respond to the periodic riverbed morphology. The strips were produced during high water events affecting river processes on the long-term. The D-shaped profiles are investigated. They relate to the CL regions and are formed where the velocity is lower near the surface. The maximum velocity is situated at around zUmax/π~0.80. The occurrence of D-shaped profiles shows a dependence on the local Froude number; the SLOM might be an inviscid response to the bed forms. Two distinct approaches in the study of the turbulent velocity field were made: an analysis of the mean turbulence characteristics and the analysis of particular instantaneous features in the turbulent flow. The normal Reynolds stress distributions are anisotropic because streamwise turbulent intensity (TI) dominates the TKE (50 to 80% of the total energy with maximum at z/h≈0.70). Spanwise and vertical contributions vary along the flow depth. The surface and the bottom layers exert a strong influence on the vertical TI profile inducing a parabolic distribution. Self-similarity of the flow is only found inside the blending layer. In the roughness layer all Reynolds shear stresses become equally important. Empirical formula established by previous authors to describe the Reynolds stress tensor components are analyzed. The diffusive terms in the TKE budget equation are negligible and consequently the pseudo-dissipation may be considered equal to the actual dissipation ε~ ≈ ε. Production mostly happens inside the boundary layer and is mainly due to the gradients of the streamwise component. Dissipation is more widely spread in the flow with a maximum value near the bottom. Production and dissipation follow the exponential laws e5.1z/h and e3.9z/h, respectively. The repartition of the flow energy through the eddy scales is studied and a characteristic curve is found having as parameters the scale corresponding to the peak energy (Λmax), and a dimensionless parameter representing the energy dispersion through the scales (σ+). Empirical power laws are presented for the turbulence scales, Taylor (λ), Kolmogorov (η), integral (L), energetic (Λ) and mixing length (lm). The relationship between the different scales is also studied. With conditional sampling techniques, particular features of the turbulent velocity and their role in flow dynamics were described: detection and analysis of bursting packets inside the boundary layer and the study of the passage of large-sale uniform momentum regions (UMR), which here were called Streamwise Velocity Pulsation (SVP). Innovative tools are applied in the study of the instantaneous velocity measurements: wavelet decomposition and multiresolution analysis, empirical mode decomposition (EMD) and phase averaging based on the Hilbert transform. SVP corresponds to a non-periodic passage of large-scale UMR with streamwise velocities alternately higher or lower than the mean. This phenomenon was identified and characterized. The SVP Strouhal number is within the range 0.13