Keywords: flow structure, gravel river bed, bedform, turbulent flow, microstructure, acoustic Doppler velocimeter.

This study presents a fundamental investigation of the flow patterns over a regular man-made gravel river bed using both laboratory experiments (i.e., ADV method) and a computational fluid dynamics (CFD) numerical simulation (i.e., RSM model) under the same flow conditions. The acoustic Doppler velocimeter (ADV) measurement provides the longitudinal, transverse and vertical velocities and turbulence intensities at different locations above the bed, which reveal the three-dimensionality and the complexity of the phenomenon. The experimental data are then used to validate the corresponding numerical simulations. A commercial CFD solver FLUENT is employed to predict the velocity and pressure fields of the gravel river bed. FLUENT is widely used and has been employed for many hydraulic engineering and geophysical applications [1]. The numerical simulations employ the Reynolds stress model (RSM) as the turbulence model and a high quality unstructured mesh with a non-equilibrium wall function. Since the The RSM modeling is advantageous compared to the commonly used (k-epsilon) model in the prediction of complex flows, such as streamline curvature, swirl, rotation, and rapid changes in strain rate due to abandoning the isotropic eddy-viscosity hypothesis. The turbulence parameters, which are difficult to be measured in experiments, can be alternatively obtained from the simulation, e.g., turbulent intensity.

From the experimental results, the turbulence intensity between particles is so strong as to decrease the mean velocity. Under the condition of relative submergence of h/d=12.75, the effect caused by the protrusion of the particle can not be sensed above 0.2h. From the velocity profile comparisons, due to the presence of the particles, the rapid acceleration at the front end of the particles causes steep gradients near the wall and high turbulence intensity. The downstream particles are in the wake of the upstream particles, at a height comparable to the height of a particle diameter, which is responsible for the peak in turbulent kinetic energy. Due to the space constraints in the near wall region, the turbulent intensity is set to zero first on wall surface, gradually increases to a peak level as departed from the wall, and then rapidly reduces to a small, constant value.

All in all, the RSM model performs well in predicting the mean flow and the turbulent parameters in this study. The discrepancy is mainly attributed to the resolution of computational meshes, which can not exactly describe the physical formation of spherical particles. Although a steady-state is considered in the RANS modeling, the Reynolds stress provided from RSM model can still be validated with the experimental data. Future work will focus on the use of a time-dependent turbulent model to examine the coherent motion and explain the nature of the coherent structures created in the turbulence. In addition, the understanding of interaction mechanism between the flow field and bed-form demands the development of both instruments and simulation technology [2].

1

A.P. Nicholas, "Computational fluid dynamics modelling of boundary roughness in gravel-bed rivers: An investigation of the effects of random variability in bed elevation", Earth Surface Processes and Landforms, 26, 345-362, 2001. doi:10.1002/esp.178

2

K.B. Strom, A.N. Papanicolaou, G. Constantinescu, "Flow heterogeneity over 3D cluster microform: Laboratory and numerical investigation", Journal of Hydraulic Engineering, 130(6):554-567, 2007. doi:10.1061/(ASCE)0733-9429(2004)130:6(554)

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £95 +P&P)