|Abstract:||Bifurcations are one of the fundamental elements of a fluvial (river) system, and diversions are a special type of asymmetrical bifurcation where one of the channels after bifurcation continues along the original channel. Diversions can be found in nature, though many of them are built to divert water and sediment from the river for various purposes. Historically, diversions were built to divert water for irrigational and navigational purposes. Recently the importance of diversions has increased, as building diversions to divert sediment (and water) have been put forth as a method to rebuild deltas that have been losing land due to rapid rise in sea-level, subsidence etc. One of the prime examples is the plan under consideration by United States Army Corps of Engineers (USACE) to re- build the Mississippi river delta through diverting water and sediment from the lower Mississippi river. Design of aforementioned diversions would be immensely benefited by a better understanding of the sediment distribution at diversions, and the hydrodynamics that drive it. One of the first and most extensive experimental studies to understand the dynamics at a diversion was conducted by Bulle in 1926, at Karlsruhe, Germany. Bulle found that a disproportionate percentage of bedload went into the lateral-channel, compared to the percentage of water entering the lateral-channel. This non-linear distribution of near-bed sediment between the two channels at a diversion is known as the Bulle-Effect; and since the seminal work of Bulle, multiple experimental studies have corroborated the phenomenon. Despite the importance of this phenomenon, till date the exact mechanism behind the Bulle-Effect is not clear. This thesis first unravels the mechanism behind the phenomenon, and then explores how Bulle-Effect might impact the morphodynamics of a diversion.
This thesis can be divided into two major parts:
1) First the mechanism behind the Bulle-Effect phenomenon is explored using high-resolution numerical simulation (Direct Numerical Simulations, Large Eddy Simulations) of flow and sediment transport for a configuration and at the scale similar to Bulle’s experiment. The simulations were conducted using the highly-scalable spectral-element based incompressible Navier-Stokes solver Nek5000, on which a Lagrangian point particle submodel was developed and implemented to model the transport of sediment. The simulations were computationally very expensive (∼ 240 million computational points), thus they required the use of the peta-scale supercomputer Blue Waters for conducting them. The simulation results showed that the phenomenon is caused by the mechanism, where most of the flow near the bottom entered the lateral-channel, even when the percentage of the total water discharge entering lateral-channel is relatively smaller. The phenomenon was found to be at play not only for sediment transported as bed- load, but also for suspended sediment that travels in the lower 25-35 percent of the water-column. These findings were found to hold across a range of Reynolds number (10 − 25000) of the flow, and for different diversion angles.
2) In the second part of the thesis, a Reynolds Averaged Navier-Stokes (RANS) based 3D hydrodynamics and sediment transport model was developed for Bulles experiments using the open-source solver Telemac-Mascaret. This model was found to capture the phenomenon satisfactorily but at a relatively lower computational cost. The substantial reduction in computational cost is important, because at this point it impossible to conduct accurate Large Eddy Simulations (LES) of flows at the scale of real rivers. Thus it becomes important to evaluate if RANS based models can capture a complex phenomenon, and if this is the case to what extent? The RANS model was then used to study the impact of Bulle-Effect on morphodynamics of the diversion. The separation of the flow from the left-bank of the lateral-channel was found to result in formation of a scour-hole under the high-flow zone and subsequent deposition sediment under the flow-recirculation. The impact of the change in morphology of the channel on Bulle-Effect was also analyzed.
The findings of this dissertation not only add to the fundamental understanding of an important phenomenon in nature, these also provide insights that will help in optimal design of engineered diversions and other facilities where vorticity and secondary-flow driven sediment/particle transport occurs. Based on disproportionately high sediment transported into the diversions of the Yellow River, China, Canal del Dique on the Magdalena River, Columbia etc., it can be conjectured that the Bulle-Effect plays a major role at the aforementioned diversions. Thus, in the future numerical simulations of real-world diversions should be conducted ( in conjunction with field measurements) in order to study the flow-structure and sediment distribution pattern at diversions, and to understand the extent to which the Bulle-Effect impacts real-world diversions.