Author(s): Robin Meurice; Sandra Soares-Frazao

Linked Author(s): Robin Meurice

Keywords: Finite volumes; Flushing; Mobile bed; Morphodynamics; Transient flow

Abstract: Sediment discontinuity is a major drawback of dams. While their reservoirs often see their capacity progressively diminish because of sedimentation, the downstream reaches often face several morphological issues. Since most and part of the coarse and fine sediment are respectively blocked upstream of the dam, the downstream river sees a reduction of the sediment yield and an alteration in the grain size distribution. This will lead to inevitable changes in the bed composition, which might have consequences on the development of lotic species. Also, hungry waters deprived of sediment will further erode the bed, leading to bank and infrastructure instabilities in the downstream reach. For better river management, one should hence better consider the sediment dynamics upstream and downstream of dams. While routing strategies consisting directly in passing the sediment downstream of the dam before it deposes in the reservoir should certainly always be considered first, removing the already deposited sediment is sometimes necessary. Flushing hence constitutes an interesting alternative to the energy and money expensive dredging. Nevertheless, the flows of these flushing events also come with serious issues, such as fish asphyxiation, remobilisation of long-time deposited pollutants, destruction of macroinvertebrate populations, or significant morphological changes, with possible infrastructure damage. These operations must thus be carefully prepared to limit the impacts downstream as much as possible. And to perform such a preparation, numerical modelling can constitute a powerful tool. The 2D finite-volume model that we developed in that perspective is derived from the shallow-water equations and is made of eight different equations. It combines the two-phase approach that considers momentum exchanges between the liquid and solid phases, and the two-layer approach, which tries to reproduce the dual asymptotic behaviour of the regular concentration profile, i.e. the closer to the surface, the lower the concentration, and the closer to the bed, the closer to the concentration of the bed itself. An HLLC flux scheme with original wave speed estimates was applied to that model, which was then applied to several test cases. The importance of bedload for these test cases was assessed with spatio-temporal maps and highlighted the relevancy of the model for applications with a large transport relative to the flow. However, when the transport was more limited, the application of the model did not seem worth the computational effort. Further research about its ability to properly simulate suspension and very-high transport situations is still needed, but the current model’s ability to discern the limits of the lower highly concentrated layer and to cope with the sediment inertia makes it a nice candidate for fast and very concentrated flows such as flushing events.

DOI: https://doi.org/10.3850/IAHR-39WC252171192022799

Year: 2022