Alexander C. Demetracopoulos

Effect of bed dunes on spatial development of open-channel flow

The spatial development of turbulent, sub-critical open-channel flow over five identical dunes is studied by the numerical solution of the RANS equations utilizing theVOF free-surface formulation and the k–e or Spalart–Allmaras turbulence models. Results are presented for smooth and rough walls and several dune dimensions. One of the cases was also studied experimentally. The flow separation at each dune crest generates a recirculation in the dune lee-side and reattachment at a distance, which increases with increasing dune height and decreasing dune length. Turbulence does not fully develop over the dunes, while the majority of turbulent kinetic energy production takes place in the recirculation region. The spatially mean, free-surface level decreases in the flow direction over the dunes, while the free-surface amplitude increases with increasing dune height. The dune drag coefficient increases with increasing dune height, while the contribution of form resistance on the drag increases with increasing dune height and decreasing dune length.

Numerical Computation of Turbulence Development in Flow Over Sand Dunes

The spatial development of sub-critical, turbulent, open-channel flow over a bottom with five dunes is studied. The steady-state flow is described by the RANS equations utilizing the k−ɛ turbulence model. The free-surface treatment is based on the rigid-lid approximation method, while the numerical solution is based on a finite-volume discretization. The Froude number is Fr=0.3, while both smooth and rough bottom cases are considered. Mean velocity and turbulence numerical predictions are in good agreement to available experimental data and analytical expressions. Then, results are presented for the streamwise development of eddy viscosity, Reynolds stresses and turbulence production of the flow over bed with train of five fixed dunes with dune length to water depth ratio L/d = 5 and dune height to water depth ratio h/d = 0.25. Numerical results are compared to available experimental data and overall agreement is satisfactory. Turbulence is generated mostly by flow separation at the dune crest and does not reach an equilibrium state over the dunes.

Groyne spacing role on the effective control of wall shear stress in open-channel flow

ABSTRACT The objective of an effective configuration for a series of groynes in an open channel is the attainment of large bed shear stress in the main channel for channel deepening and small sidewall shear stress in the groyne fields for bank protection with the lowest cost, i.e. with the largest spacing between groynes. A Reynolds-averaged Navier–Strokes, finite-volume, numerical model was used and it was validated against experimental data for the case of a single groyne. For series of groynes with uniform spacing between them, the most effective configuration was the one with spacing equal to six groyne lengths. For the range of parameters considered, a substantial improvement of the effectiveness was achieved by implementing a novel non-uniform configuration where the spacing between groynes was reduced to one and a half groyne lengths in the first four groyne fields, and remained equal to six groyne lengths in the subsequent ones.

Contributions towards the prediction of velocity distribution in open channel flows

The steady state, depth-averaged hydrodynamic equations are used for the computation of transverse profiles of the longitudinal velocity in prismatic and natural channels. Turbulence closure of the comlete set of equations is accomplished (a) via a simle model in which vt is assumed proportional to the shear velocity and the local depth, and (b) the depth-averaged k-e model of Rastogi & Rodi. Subsequently, the governing equations are simplified to their purely parabolic form and the transverse distance is replaced by the cumulative discharge as the second independent variable. Turbulence closure in the latter set of equations is accomplished with model (a). All models are solved numerically via Patankars scheme and the results are compared with literature data corresponding to laboratory and field measurements. It is shown that the purely parabolic, transformed-coordinates model is more efficient than, and equally accurate with, the complete model.