Solute Transport in Flow Through Random Emergent Vegetation

Aquatic vegetation governs a wide range of river processes, from flood control to nutrient exchange, and crucially, pollution dispersal and treatment. This thesis presents a comprehensive experimental study, using a random array of circular cylinders, of varying diameters, based on charactersations of real species, to simulate aquatic plants, and study mass transport in vegetated flows. Optical techniques—Particle Image Velocimetry (PIV) and Laser Induced Fluorescence (LIF)—were employed to obtain velocity and concentration fields at unprecedented spatial and temporal resolutions. Concentration maps provided estimates of longitudinal and transverse dispersion coefficients over multiple reaches, and a comprehensive range of flows (50<Red<1000). Two distinct longitudinal dispersion regimes, separated at Reynolds number, Red≈ 400, were found where different physical processes drive dispersion. For Red<400 mixing is dominated by shear and thus cannot be considered strictly Fickian. For Red>400 turbulent diffusion efficiently spreads mass locally and transport is mainly advective. Transverse dispersion is mainly determined by vegetation morphology, and not Reynolds number. Generally, solutes are seen to spread independently of initial conditions, as they quickly assimilate the average physical characteristics of vegetation. Instantaneous velocity maps were obtained with PIV and glass cylinders to allow inter-stem visualisation. Mean maps show that flow heterogeneities are caused by complex stem interactions and not by a superposition of their individual effects. Dispersive fluxes grow with increasing Red, and become dominant drivers of large-scale dispersion at Red>300. After this range, turbulent fluxes become less prominent as drivers of large scale dispersion, but small-scale mixing still increases with Red, thus reducing trapping times in recirculation zones. In summary, the novel findings present a comprehensive picture of the main hydrodynamic features dominating dispersion in vegetation, and their dependence on Reynolds number. The novel inclusion of diameter and spacing distributions in the RandoSticks array reveals the existence of a range of characteristic flow scales broader than in uniform-diameter artificial systems. This variability suggests the existence of stem and wake interactions that have not been previously included in models for vegetation morphology and mixing, and further work is needed to characterise these effects, to develop more accurate, physically-based dispersion models.