Experimental analysis of flow through and scour around vegetation islands

Sand deposits are formed by the accumulation of sediments transported through natural estuaries, lagoons and stream systems. The small sand deposits that evolve in time cause the formation of larger, vegetated, and prominent sand dunes. These structures formed as a result of a temporal process are called "vegetated islands". Given the rapid growth and spread of the vegetation on these islands, a considerable drag force or form resistance is applied during the passage of floods. Furthermore, additional turbulence is produced, the nearby flow is obstructed, and the flood wave is delayed. In addition, these structures produce some secondary-flow patterns due to the flow and by increasing the bed shear stress, they lead to the formation of narrow cross-sections on both sides of the vegetated island that can be efficient in deeper and low flow conditions. These vegetated islands move towards the flow as they develop and new vegetation sprouts around the downstream. This complex structure, consisting of vegetation that becomes thinner and shorter from downstream to upstream, gets more and more streamlined against the flow. In this study, during the evolution of such a vegetated island, the alteration of flow structure, and the resulting consequences of these alterations on the bed (resulting scour pattern) have been investigated. A three-stage modeling strategy is planned not only for the empirical observation of the development stages of the vegetated island to take place, but also to quantitatively assess the flow pattern and bed evolution in close vicinity to these islands. These are early-stage, middle-stage and mature-stage vegetated island, respectively. These stages of the vegetated islands are planned to be simulated by an idealized form of a group cylinder; a circular area with a diameter of 9 cm, which is composed of 0.5 mm-diameter cylinders at three different placement densities, each of which corresponds to one of the three stages mentioned above. With this setup, two types of experiments have been carried out: rigid-bed experiments and loose-bed experiments. Both types of experiments were conducted on a 26 m long, 0.98 m wide and 0.85 m deep channel (but the flow depth were kept at around 30 cm). The first type is rigid-bed experiments where point velocity measurements were performed around the structure and vortex street formations behind the structure were visualized. For this purpose, an acoustic Doppler veocimeter (ADV) was used, and pointwise velocity measurements were performed at the downstream of the vegetated island models on two planes: The first plane was the vertical plane (x-z plane) along the flume centerline, and the second plane was the horizontal plane at a distance of z/h=0.6 from the bed along the flume centerline. Here z is the vertical coordinate, x is the streamwise coordinate and h is the flow depth. Across each of these two planes, hundreds of pointwise measurements were conducted. The spacing of the velocity measurement grid were increased as one goes to downstream. Each measurement was conducted for a 100-second duration and sampled with 100 Hz. Once the time series of there-dimensional velocity components (u, v and w along x, y and z directions, respectively) were obtained, these data sets were then subjected to analyses, in which the time-averaged and turbulent (fluctuating) parameters of the flow were determined as spatial variables. These parameters are time-averaged velocity components, fluctuating velocity components (standard deviations instantaneous velocity components) also known as the Reynolds normal stresses, turbulent kinetic energy per unit mass and finally the turbulent shear stresses (or Reynolds shear stresses) also known as the cross-correlation of turbulence. As such, 10 (ten) characteristic parameter were obtained at each and every measurement point. Yet, one of the Reynolds shear stress component (v-w correlation) was not presented in this study since it would be physically irrelevant given the present configuration of the obstacles (i.e. vegetated islands) and the measurement locations. Once these parameters were obtained at each and every location of the measurement grid, a spatial interpolation for each of the parameter at each of the configuration were performed, yielding the contour plots of the flow structure at the wake of the three vegetated island configurations (i.e. a total of 10 x 3 = 30 contour plots). As an additional analysis, vector representations of time-averaged velocity components were plotted on top of the contour plots, which rendered the flow picture more meaningful form the sense that one can follow the evolution and directionality of the flow together with spatial distribution of the flow parameters. For flow visualization experiments, food dye with two different colors (red and green) was used, for the two sides of the vegetated island models. In the meantime, video recording from the plan view of the flume was performed. This exercise gave the opportunity to carry out a Lagrangian analysis, which proved to be very useful while interpreting the results of the velocity measurements as well as scour tests. The second type of experiments conducted were the loose-bed experiments, where scour patterns around the structure on the active sand corridor were investigated for identical flow characteristics with the rigid-bed experiments. For the loose-bed experiments, a 3.50 meter-long, channel-wide and 0.20 meter-thick artificial sand layer (i.e. a sand pit) was placed in the flume. However, the flow depth over the sand layer was kept identical with the previous experiments, simply by increasing the overall water depth in the flume. The sand in the sand pit was uniform medium-coarse sand, which resulted with clear-water scour regime (in which the far-fields Shields parameter is below the critical value for initiation of motion on the bed). The experiments were lasted for almost 8 hours for a clear and well-developed picture of the scoured bed. Once a scour test was completed. The water in the flume was gradually drained and a laser-scanner was utilized to scan the scoured bed. This laser scanned had a very high precision and the resulting scan results were processed to maintain a 3D geometric model of the scoured bed. As such, the scour morphometry and characteristic scour dimensions (scour depth, scour width, scoured area and scoured volume) could be assessed for each and every modelled stage of the vegetated islands. The results of the experimental study showed that the early-stage vegetation islands intervene the flow much less than the mid-stage and mature-stage vegetated islands. The flow diversion, the strength of recirculation region and shedding were all weaker in the early-stage case. The main reason was found to be the strong bleed flow in the early-stage vegetated islands, which gradually diminished with the increasing solid volume fraction with the development (growing-up) of the island. The flow visualization results showed that a very long steady wake region was developed for the early-stage vegetated island with very poor vortex shedding, whereas strong vortex shedding with shorter steady wake regions were seen in the mid-stage and mature-stage vegetated islands. Similar results were evident in the loose-bed (scour) experiments. The scour depth as well as scour extent was much less in the early-stage vegetated island configuration compared to the other two configurations tested. Furthermore, the traces of wake disturbance in terms of ripples on the loose-bed, was much less in the early-stage case. A very-well defined sediment deposition in the form of a long ridge was also evident at the end of each scour test, but this feature became much more pronounced as the vegetation density (or the age of vegetated island) increased. Finally, the scour geometries for each case were reported in detail in the thesis. The obtained results of this study not only shed light to how different stages of vegetated islands interact with the flow and the riverbed, but also how the vegetated islands contribute to flow resistance, turbulence mixing and morphological evolution in rivers. The reported results also constitute an important library of data to validate/tune their model, for those who apply numerical modeling methods to study the vegetated island phenomenon.