Experimental Study and Numerical Simulation of Vegetated Alluvial Channels

Abstract

Vegetation in rivers increases flow resistance and bank stability, reduces bed resistance and flow conveyance, improves water quality, promotes habitat diversity, and alters both mean and turbulent flow. By reducing bed resistance and altering turbulent characteristics, vegetation can change the distribution of deposition and erosion processes. To understand all above mentioned vegetation effects, more research is needed. The goal of this dissertation was to determine the impacts of vegetation on bed resistance and sediment transport and identify a best approach for quantifying vegetation induced friction resistance. To achieve this, both experimental study and numerical simulation were performed. A series of laboratory experiments were conducted in an open channel flume to investigate the impacts of vegetation density on bed resistance and bed load transport for emergent vegetation condition. The bed resistance in a mobile bed channel is equal to the summation of grain and bed form resistances. An attempt has been made to make a separation between grain and bed form resistances, which is challenging and has never been reported in literature. An alternative approach is used to calculate the grain resistance. A new iterative method was derived to calculate the bed form resistance. Empirical relations were formulated to calculate the bed form resistance and bed load transport rate using a newly defined flow parameter that incorporates the vegetation concentration. The bed elevations and bed form height were measured by the Microsoft Kinect 3D Camera. It was found that the height of bed form depends on the vegetation concentration, which determines whether ripple/dune or scour holes are dominant on the bed surface. For sparsely vegetated flows, the bed form height and resistance are decrease rapidly as the vegetation concentration was increased, and they decreased gradually when the vegetation concentration was high. To quantify the vegetation induced friction resistance, a 3D numerical simulation was conducted using the Delft3D-FLOW open source program. The study area is Davis Pond freshwater marsh area near New Orleans, Louisiana. The dominant vegetation type for the study area is Panicum hemitomon. The study area was divided into several sub-areas depending on the existence of channels, overbanks, and vegetation height. Several approaches were used to approximate the vegetation roughness; a constant Manning's n coefficient, a time-varying n or Chezy's C coefficient, and the modified momentum and k-ɛ equations for each subarea. To quantify the time varying roughness coefficients, four equations for calculating n values were incorporated in the Delft3D-FLOW program in addition to two options offered by this program to calculate C values. It is concluded that the use of the time varying roughness coefficient gives better results than other approaches. Among the selected equations to calculate the time varying vegetation roughness, the equations that account for the effect of the degree of submergence and the vegetation frontal area per unit volume, symbolized as a, gave the closet matches with the observations. The sensitivity of modeling results to the selection of vertical grid (σ–and Z-grids), a value, and grid size were analyzed. It is found that using the σ-grid yielded more accurate results with less CPU times and the best range of a value for the Panicum hemitomon vegetation type is from 8.160 to 11.220 m⁻¹. Also it was observed that the adoption of a coarse mesh gives reasonable simulation results with less CPU time compared with a fine mesh. A non-linear relation between the vegetation resistance, in terms of n value, and degree of submergence was observed.