The spatiotemporal evolution of fine tailings in sand-bed rivers
Advisor: Professor Marcelo H. García
The spreading of mine tailings in riverine systems has been studied through classical morphodynamics evolution analysis limited to natural sediment conditions. However, the density of tailings can substantially differ from the density of natural sediment in rivers due to the tailing's mineralogical composition. To overcome the density difference between the materials, a 1D mathematical formulation based on the active layer concept is developed to predict the spatiotemporal evolution of the mixture of two materials with different densities that interacts in a watercourse within the active layer and in the substrate. This formulation is based on the assumption of a strong correlation between the density of the tailings and mineral content (or metals). Herein, this formulation is used to estimate the spatiotemporal evolution of the spreading of a single iron tailings pulse deposited in the Paraopeba River (Brazil) on January 25, 2019. As the sediment density and content of iron tailings has a strong correlation, the sediment density is used as a surrogate to track the temporal distribution of the iron tailings along the river. The model formulation aims to determine the sediment density evolution for different fractions within the active layer and in the substrate. The analysis also includes a re-calibration for the Sorting Engelund-Hansen Equation, SEH (An et al.,2020), as the original parameters were determined based on experiments in large flumes that are limited for sand-bed rivers.
The SEH equation is selected due to the lack of bed and suspended load formulations developed for rivers with substantial amounts of iron tailings in a riverbed. The field measurements campaigns conducted along rivers affected by iron tailings deposition (i.e., Doce River and Paraopeba River) demonstrate the predominance of sand-tailings mixtures to move in suspension. Thus, this suspension component can be determined using the vertical velocity profile and the suspended sediment distribution for equilibrium conditions. This approach requires the determination of the near-bed concentrations that are usually estimated via sand-based formulations. As the presence of iron tailings is not typically found in riverbeds, this research aims to determine whether sand-based formulations reasonably predict near-bed concentrations of two poorly-sorted iron tailings samples subjected to bed shear stress in an annular flume. One of the main findings of this research is that the near-bed concentrations can be predicted by adding two dimensionless parameters to the sand-based entrainment formulations: the relative roughness (H/D) of the sample and the particle Reynolds number (Repi) for fractions smaller than the thickness of the viscous sublayer (δv).
Finally, this thesis explores the mechanisms that govern the spreading of a single pulse of heavy and fine iron tailings moving along a sand-bed river and the impact the downstream aggradation has on its water stages. To reach these goals, this dissertation uses the field measurements
database from the Paraopeba River, as this watercourse has been continuously monitored after the tailing dam breach that led to an uncontrolled deposition of iron tailings into the river. The analysis demonstrates that a combination of translation and dispersion are the mechanisms that govern the spreading of this single pulse. The translation component is explained by the fact that these tailings are dynamically more movable than the ambient sand, even though they are heavier than the ambient sand of the river. On the other hand, the dominance of dispersion is explained mainly by the morphology of the river, the low-velocity zones, the mixing of the tailings with the ambient material, and tailings burial in the river substrate. This study also shows that the river's water surface elevations (WSE) are not sensitive to the aggradation nor the smoothing of the riverbed generated by the spreading of the iron tailings, as bedrock outcrops and pool-riffle sequences highly control the river hydrodynamics.